Decision curve analysis to evaluate the clinical benefit of prediction models

Establishment and validation of a clinical prediction model for colorectal adenoma risk factors

Colorectal adenomas are benign tumors of the colorectal mucosal epithelium that have malignant potential and are regarded as precancerous lesions of colorectal cancer, for which the specific risk factors are unclear. The present study aimed to identify independent risk factors for colorectal adenoma to develop a prediction model and test its predictive value. A retrospective analysis was performed using data from patients who underwent electronic colonoscopy at the Department of Proctology (Affiliated Hospital of Shandong University of Traditional Chinese Medicine; Jinan, China) from January 2013 to December 2023 and had polyps removed during colonoscopy. Patients with colorectal adenoma were included in the case group, whilst those with no visible abnormalities on endoscopy or with non-adenomatous polyps were included as a control group. The patients were randomly divided into a training and validation group in a 7:3 ratio. Variables were screened using single-component analysis and the filtered variables were employed in multivariate logistic regression to create a clinical prediction model. Finally, the model was internally and externally validated. A total of 730 patients were included in the present study, with 286 assigned to the case group and 444 to the control group. After the initial screening of 39 variables, 12 continued to the next round, resulting in four potential predictors including age, daily number of bowel movements, thrombin time and the number of polyps. A prediction model was created based on these variables. Regarding internal validation, the C-index was 0.7054 [95% confidence interval (CI), 0.6596-0.7512] and the prediction probability in the calibration curve was close to the diagonal line of the calibration graph, indicating that the prediction probability of the model was reasonable. Regarding external validation, the C-index in the validation cohort was 0.6306 (95% CI, 0.5560-0.7053) and the calibration curve also demonstrated good identification capabilities. The Hosmer-Lemeshow test revealed that the model had a reasonable calibration degree, with χ2=9.7893, degree of freedom=8 and P=0.28. The receiver operating characteristic curve and decision curve analysis for the training and validation cohorts demonstrated good efficacy and an ideal application value. In conclusion, the model constructed in the present study demonstrated moderate predictive accuracy for colorectal adenoma risk, laying the groundwork for early detection of colorectal adenoma and secondary prevention of colorectal cancer.

Early differential diagnosis models of Talaromycosis and Tuberculosis in HIV-negative hosts using clinical data and machine learning

BACKGROUND

Talaromyces marneffei is an emerging pathogen, and the number of infections in HIV-negative individuals is increasing. In HIV-negative individuals, talaromycosis is usually misdiagnosed as another disease, especially tuberculosis (TB).

METHODS

We retrospectively extracted the clinical data of HIV-negative patients with Talaromyces marneffei infection from 2018 to 2023, analyzed the differences between TB patients and talaromycosis patients and attempted to establish differential diagnosis models utilizing clinical prediction models for these two diseases.

RESULTS

Overall, 718 patients, including 137 patients with talaromycosis and 581 patients with pulmonary tuberculosis (PTB), were enrolled in this study. According to the multivariate analysis, age > 65 years, expectoration, and PLT count were independent predictors for TB. Fever, chest pain, gasping, rash, lymphadenectasis, osteolysis, Neu count, EOS count, and ALB were independent predictors for talaromycosis. Receiver operating characteristic (ROC) curve analysis of the training set showed that the area under the curve (AUC) (95 % CI) of the clinical differential model based on logistic regression analysis was 0.918 (0.884-0.953). The model was verified in the validation set. ROC curve analysis of the validation set showed that the AUC (95 % CI) was 0.900 (0.841-0.959).

CONCLUSION

These new differential diagnosis models can calculate the probability of either talaromycosis or tuberculosis.

Development and validation of a nomogram for prognosis of bone metastatic disease in patients with esophageal squamous cell carcinoma: A retrospective study in the SEER database and China cohort

Purpose

Esophageal squamous cell carcinoma (ESCC) is a prevalent malignant tumor worldwide, and individuals with ESCC and bone metastasis (BM) often face a challenging prognosis. Our objective was to identify the risk and prognostic factors associated with BM in patients with ESCC and develop a nomogram for predicting Cancer-Specific Survival (CSS) which following the occurrence of BM.

Methods

We conducted a retrospective analysis of data pertaining to ESCC patients with BM registered in the Surveillance, Epidemiology, and End Results (SEER) database from 2010 to 2015, as well as those treated at a Chinese institution from 2006 to 2020. Significant prognostic factors for CSS were assessed through univariate and multivariate Cox regression analyses. Subsequently, a nomogram was developed utilizing the SEER database and externally validated using real-world evidence from a Chinese cohort.

Results

A total of 266 patients from the SEER database and 168 patients from the Chinese cohort were included in the analysis. In the SEER cohort, multivariate analysis indicated that chemotherapy, radiotherapy, liver metastasis, brain metastasis, and sex were independent prognostic factors for ESCC with BM. The prognostic nomogram demonstrated areas under the ROC curve (AUCs) of 0.823, 0.796, and 0.800, respectively, for predicting 3-, 6-, and 12-month CSS. In the Chinese validation cohort, the nomogram exhibited acceptable discrimination (AUCs: 0.822, 0.763, and 0.727) and calibration ability.

Conclusion

The study developed a prognostic nomogram to predict CSS in ESCC patients with BM, which can help clinicians assess survival and make individualized treatment decisions.

Prediction of coronary heart disease based on klotho levels using machine learning

The diagnostic accuracy for coronary heart disease (CHD) needs to be improved. Some studies have indicated that klotho protein levels upon admission comprise an independent risk factor for CHD and have clinical value for predicting CHD. This study aimed to construct a tool to predict CHD risk by analyzing klotho levels and clinically relevant indicators by using a machine learning (ML) method. We randomly assigned the dataset of the National Health and Nutrition Examination Survey (NHANES) 2007-2016 to training and test sets at a ratio of 70:30. We evaluated the ability of five models constructed using logistic regression, neural networks, random forest, support vector machine, and eXtreme Gradient Boosting to predict CHD. We determined their predictive performance using the following parameters: area under the receiver operating characteristic curve, accuracy, precision, recall, F1, and Brier scores. We analyzed data from 11,583 persons in US NHANES and entered 13 potential predictive variables, including klotho and other clinically relevant indicators, into the feature screening process. We established that the five ML models could predict the onset of CHD. The RF model showed the best predictive performance among the five ML models.

Comprehensive analysis of risk factors and nomogram development for predicting hepatic metastasis following radical resection of adenocarcinoma of the esophagogastric junction

BACKGROUND

Adenocarcinoma of the esophagogastric junction (AEG) often presents with subtle early symptoms and delayed diagnosis, frequently resulting in liver metastasis and a poor prognosis. This study aimed to investigate the primary risk factors influencing postoperative liver metastasis in AEG and to develop a simple predictive model to facilitate clinical risk stratification and individualized follow-up strategies.

METHODS

This retrospective study analyzed data from 524 patients with AEG who underwent radical resection, with patients randomly divided into a training group (368 cases) and a validation group (156 cases). Clinical and pathological information was collected, and independent factors significantly associated with postoperative liver metastasis were identified using univariate and multivariate Cox regression analyses. Based on these findings, a nomogram model was constructed to predict the 1-year and 3-year liver metastasis-free survival rates, and the model's predictive performance and clinical utility were evaluated using the C-index, ROC curves, and calibration curves.

RESULTS

Multivariate analysis revealed that thoracoabdominal surgery, higher N stage (N1 and N2/N3), moderate-to-poor differentiation, the presence of vascular tumor thrombus, intestinal type according to Lauren classification, and P53 status were independent risk factors for postoperative liver metastasis. The nomogram model based on these six indicators demonstrated high predictive accuracy in both the training group (C-index = 0.966) and the validation group (C-index = 0.976), with ROC AUCs for both the 1-year and 3-year predictions exceeding 0.96 and favorable calibration curves, confirming the model's strong predictive efficacy.

CONCLUSIONS

The predictive model developed in this study can effectively assess the risk of postoperative liver metastasis in patients with AEG, thereby providing a scientific basis for postoperative monitoring and individualized treatment, with the potential to improve patient outcomes in clinical practice.

Integrated bioinformatics, machine learning, and molecular docking reveal crosstalk genes and potential drugs between periodontitis and systemic lupus erythematosus

Evidence indicates a connection between periodontitis (PD) and systemic lupus erythematosus (SLE), though the underlying co-morbid mechanisms remain unclear. This study sought to identify the genetic factors and potential therapeutic agents involved in the interaction between PD and SLE. We employed multi-omics methodologies, encompassing differential expression analysis, weighted gene co-expression network analysis (WGCNA), functional enrichment (GO/KEGG), LASSO regression, diagnostic model construction, protein-protein interaction (PPI) networks, immune infiltration profiling, computational drug prediction, molecular docking, and disease subtyping, to analyze PD and SLE expression datasets from the Gene Expression Omnibus (GEO) database (GSE10334, GSE16134, GSE50772, and GSE81622). Cross-analysis identified 32 crosstalk genes (CGs) common to both PD and SLE. LASSO analysis pinpointed three key diagnostic genes (TAGLN, MMP9, TNFAIP6) for both conditions. The resulting diagnostic models demonstrated robust efficacy in both training and validation datasets. Four topological algorithms in Cytoscape highlighted four central crosstalk genes (TAGLN, MMP9, TNFAIP6, IL1B). Additionally, hesperidin, doxycycline, and cytochalasin D emerged as potential therapeutic agents. Two subtypes (C1 and C2) of PD and SLE were delineated based on CG expression profiles. The development of diagnostic models, potential drug identification, and disease subtype classification are poised to enhance diagnosis and treatment. These findings aim to deepen the understanding of PD and SLE complexities.

Prediction Models of Medication Adherence in Chronic Disease Patients: Systematic Review and Critical Appraisal

AIMS AND OBJECTIVES

To summarise the currently developed risk prediction models for medication adherence in patients with chronic diseases and evaluate their performance and applicability.

BACKGROUND

Ensuring medication adherence is crucial in effectively managing chronic diseases. Although numerous studies have endeavoured to construct risk prediction models for predicting medication adherence in patients with chronic illnesses, the reliability and practicality of these models remain uncertain.

DESIGN

Systematic review.

METHODS

We conducted searches on PubMed, Web of Science, Cochrane, CINAHL, Embase and Medline from inception until 16 July 2023. Two authors independently screened risk prediction models for medication adherence that met the predefined inclusion criteria. The Prediction Model Risk of Bias Assessment Tool (PROBAST) was employed to evaluate both the risk of bias and clinical applicability of the included studies. This systematic review adhered to the 2020 PRISMA checklist.

RESULTS

The study included a total of 11 risk prediction models from 11 studies. Medication regimen and age were the most common predictors. The use of PROBAST revealed that some essential methodological details were not thoroughly reported in these models. Due to limitations in methodology, all models were rated as having a high-risk for bias.

CONCLUSIONS

According to PROBAST, the current models for predicting medication adherence in patients with chronic diseases exhibit a high risk of bias. Future research should prioritise enhancing the methodological quality of model development and conducting external validations on existing models.

RELEVANCE TO CLINICAL PRACTICE

Based on the review findings, recommendations have been provided to refine the construction methodology of prediction models with an aim of identifying high-risk individuals and key factors associated with low medication adherence in chronic diseases.

PATIENT OR PUBLIC CONTRIBUTION

This systematic review was conducted without patient or public participation.

Concomitant Procedures, Black Race, Male Sex, and General Anesthesia Show Fair Predictive Value for Prolonged Rotator Cuff Repair Operative Time: Analysis of the National Quality Improvement Program Database Using Machine Learning

PURPOSE

To develop machine learning models using the American College of Surgeons National Quality Improvement Program (ACS-NSQIP) database to predict prolonged operative time (POT) for rotator cuff repair (RCR), as well as use the trained machine learning models, cross-referenced with traditional multivariate logistic regression (MLR), to determine the key perioperative variables that may predict POT for RCR.

METHODS

Data were obtained from a large national database (ACS-NSQIP) from 2021. Patients with unilateral RCR procedures were included. Demographic, preoperative, and operative variables were analyzed. An MLR model and various other machine learning techniques, including random forest (RF) and artificial neural network, were compared using area under the curve, calibration, Brier score, and decision curve analysis. Feature importance was identified from the overall best-performing model.

RESULTS

A total of 6,690 patients met inclusion criteria. The RF machine learning model had the highest area under the curve upon internal validation (0.706) and the lowest Brier score (0.15), outperforming the other models. The RF model also demonstrated strong performance upon assessment of the calibration curves (slope = 0.86, intercept = 0.08) and decision curve analysis. The model identified concomitant procedure, specifically labral repair and biceps tenodesis, as the most important variable for determining POT, followed by age <30 years, Black or African American race, male sex, and general anesthesia.

CONCLUSIONS

Despite the advanced machine learning models used in this study, the ACS-NSQIP data set was only able to fairly predict POT following RCR. The RF model identified concomitant procedures, specifically labral repair and biceps tenodesis, as the most important variables for determining POT. Additionally, demographic factors such as age <30 years, Black race, and general anesthesia were significant predictors. While male sex was identified as important in the RF model, the MLR model indicated that its predictive value is primarily in conjunction with specific procedures like biceps tenodesis and subacromial decompression.

LEVEL OF EVIDENCE

Level IV, retrospective comparative prognostic trial.

Development and validation of a prediction model for early recurrence in upper tract urothelial carcinoma treated with radical nephroureterectomy

BACKGROUND

Most cases of upper tract urothelial carcinoma (UTUC) exhibit recurrence within the first year following surgery. The time from surgery to recurrence significantly impacts cancer-specific survival. In this study, we analyzed patients with localized UTUC (pTis-3N0/xcM0) who experienced postoperative recurrence to identify an appropriate early recurrence time point and the associated risk factors.

METHODS

From July 1988 to October 2022, we retrospectively analyzed 3435 localized UTUC patients after undergoing radical nephroureterectomy using Taiwan's UTUC Collaboration Group Database. Early recurrence time point was defined according to oncologic outcome. Variables including clinical and pathological characteristics were assessed in relation to early recurrence. A prediction model was constructed by factors associated with early recurrence and externally validated.

RESULTS

Early recurrence time point in localized UTUC was determined at 9 months post-surgery, with patients experiencing early recurrence exhibiting worse overall and cancer specific survival. Diabetes mellitus, multifocality, lympho-vascular invasion, tumor necrosis and pathologic T stage were independent factors associated with early recurrence. The predictive model for early recurrence achieved an area under the curve (AUC) of 0.84 (95%CI: 0.82-0.86). External validation demonstrated that the model exhibited good discrimination (AUC: 0.76, 95%CI: 0.73-0.79), calibration (Brier score: 0.08) and clinical utility in a distinct cohort.

CONCLUSIONS

This study identified the optimal time point for early recurrence and its associated risk factors. A prediction model for early recurrence was developed based on these factors and validated externally, demonstrating good generalizability. This clinical tool can facilitate early identification of high-risk patients, enabling targeted surveillance and timely intervention. Future studies should explore effective treatment strategies for preventing early recurrence.

Factors Influencing Early Diagnosis of Ruptured Abdominal Aortic Aneurysms: The Role of Neutrophils

Background

Currently, there is no effective and convenient indicator for the early differential diagnosis of ruptured abdominal aortic aneurysms (rAAAs) from unruptured abdominal aortic aneurysms (AAAs).

Objective

The aim of this study was to explore indicators for the early differential diagnosis of rAAAs in a clinical setting.

Methods

This case‒control study included 276 subjects within the last 5 years (220 patients with unruptured AAAs; 56 patients with rAAAs) in the initial analysis and 229 subjects (186 patients with unruptured AAA's; 43 patients with rAAA's) after subgroup analysis. The meaningful indicators were screened via univariate analysis and logistic regression analysis. The diagnostic performance and clinical usefulness of the indicators were assessed and compared using receiver operating characteristic (ROC) curve analysis, decision curve analysis (DCA) and clinical impact curve (CIC).

Results

A high venous blood neutrophil counts (OR = 1.316, P = 0.007) was found to be a risk factor for rAAAs in the initial model. After subgroup analysis, the levels of neutrophils (OR = 1.394, P = 0.017) and D-dimer (OR = 1.023, P = 0.043) were both significantly greater in patients with a rAAA. Abdominal pain (OR = 32.613, P = 0.044) and back pain (OR=91.946, P = 0.036) were strongly associated with the rupture of AAA. The results of the receiver operating characteristic (ROC) analysis revealed that neutrophils (AUC: 0.847, 95% CI: 0.774-0.921) and NLR (AUC: 0.795, 95% CI: 0.717-0.873) had good diagnostic performance for rAAA. DCA demonstrated that the net benefit of neutrophils was greater than that of other indicators. The CIC confirmed that the model has good clinical usefulness.

Conclusion

The use of neutrophils may enhance the early diagnostic accuracy for identifying rAAAs and holds potential for clinical and scientific applications.

Breast Cancer: Habitat imaging based on intravoxel incoherent motion for predicting pathologic complete response to neoadjuvant chemotherapy

BACKGROUND

Radiomics research based on whole tumors is limited by the unclear biological significance of radiomics features, which therefore lack clinical interpretability.

PURPOSE

We aimed to determine whether features extracted from subregions defined by habitat imaging, reflecting tumor heterogeneity, could identify breast cancer patients who will benefit from neoadjuvant chemotherapy (NAC), to optimize treatment.

METHODS

143 women with stage II-III breast cancer were divided into a training set (100 patients, 36 with pathologic complete response [pCR]) and a test set (43 patients, 16 with pCR). Patients underwent 3-T magnetic resonance imaging (MRI) before NAC. With the pathological results as the gold standard, we used the training set to build models for predicting pCR based on whole-tumor radiomics (ModelWH), intravoxel incoherent motion (IVIM)-based habitat imaging (ModelHabitats), conventional MRI features (ModelCF), and immunohistochemical findings (ModelIHC). We also built the combined models ModelHabitats+CF and ModelHabitats+CF+IHC. In the test set, we compared the performance of the combined models with that of the invasive ModelIHC by using the area under the receiver operating characteristic curve (AUC) and decision curve analysis (DCA). Receiver operating characteristic (ROC) curve analysis was performed to evaluate the predictive value of the model. The DeLong test was used to compare diagnostic efficiency across different parameters.

RESULTS

In the prediction of pCR, ModelWH, ModelHabitats, ModelCF, ModelIHC, ModelHabitats+CF, ModelCF+IHC and ModelHabitats+CF+IHC achieved AUCs of 0.895, 0.757, 0.705, 0.807, 0.800, 0.856, and 0.891 respectively, in the training set and 0.549, 0.708, 0.700, 0.788, 0.745, 0.909, and 0.891 respectively, in the test set. The DeLong test revealed no significant difference between ModelIHC versus ModelHabitats+CF (p = 0.695) and ModelHabitats+CF+IHC versus ModelCF+IHC (p = 0.382) but showed a significant difference between ModelIHC and ModelHabitats+CF+IHC (p = 0.043).

CONCLUSION

The habitat model we established from first-order features combined with conventional MRI features and IHC findings accurately predicted pCR before NAC. This model can facilitate decision-making during individualized treatment for breast cancer.

Deep learning-based MVIT-MLKA model for accurate classification of pancreatic lesions: a multicenter retrospective cohort study

BACKGROUND

Accurate differentiation between benign and malignant pancreatic lesions is critical for effective patient management. This study aimed to develop and validate a novel deep learning network using baseline computed tomography (CT) images to predict the classification of pancreatic lesions.

METHODS

This retrospective study included 864 patients (422 men, 442 women) with confirmed histopathological results across three medical centers, forming a training cohort, internal testing cohort, and external validation cohort. A novel hybrid model, Multi-Scale Large Kernel Attention with Mobile Vision Transformer (MVIT-MLKA), was developed, integrating CNN and Transformer architectures to classify pancreatic lesions. The model's performance was compared with traditional machine learning methods and advanced deep learning models. We also evaluated the diagnostic accuracy of radiologists with and without the assistance of the optimal model. Model performance was assessed through discrimination, calibration, and clinical applicability.

RESULTS

The MVIT-MLKA model demonstrated superior performance in classifying pancreatic lesions, achieving an AUC of 0.974 (95% CI 0.967-0.980) in the training set, 0.935 (95% CI 0.915-0.954) in the internal testing set, and 0.924 (95% CI 0.902-0.945) in the external validation set, outperforming traditional models and other deep learning models (P < 0.05). Radiologists aided by the MVIT-MLKA model showed significant improvements in diagnostic accuracy and sensitivity compared to those without model assistance (P < 0.05). Grad-CAM visualization enhanced model interpretability by effectively highlighting key lesion areas.

CONCLUSION

The MVIT-MLKA model efficiently differentiates between benign and malignant pancreatic lesions, surpassing traditional methods and significantly improving radiologists' diagnostic performance. The integration of this advanced deep learning model into clinical practice has the potential to reduce diagnostic errors and optimize treatment strategies.

A nomogram based on systemic inflammation response index and clinical risk factors for prediction of short-term prognosis of very elderly patients with hypertensive intracerebral hemorrhage

Objective

To develop and validate a nomogram based on systemic inflammation response index (SIRI) and clinical risk factors to predict short-term prognosis in very elderly patients with hypertensive intracerebral hemorrhage (HICH).

Methods

A total of 324 very elderly HICH patients from January 2017 to June 2024 were retrospectively enrolled and randomly divided into two cohorts for training (n = 227) and validation (n = 97) according to the ratio of 7:3. Independent predictors of poor prognosis were analyzed using univariate and multivariate logistic regression analyses. Furthermore, a nomogram prediction model was built. The area under the receiver operating characteristic curves (AUC), calibration plots and decision curve analysis (DCA) were used to evaluate the performance of the nomogram in predicting the prognosis of very elderly HICH.

Results

By univariate and stepwise multivariate logistic regression analyses, GCS score (p < 0.001), hematoma expansion (p = 0.049), chronic obstructive pulmonary disease (p = 0.010), and SIRI (p = 0.005) were independent predictors for the prognosis in very elderly patients with HICH. The nomogram showed the highest predictive efficiency in the training cohort (AUC = 0.940, 95% CI: 0.909 to 0.971) and the validation cohort (AUC = 0.884, 95% CI: 0.813 to 0.954). The calibration curve indicated that the nomogram had good calibration. DCA showed that the nomogram had high applicability in clinical practice.

Conclusion

The nomogram incorporated with the SIRI and clinical risk factors has good potential in predicting the short-term prognosis of very elderly HICH.

Construction and validation of machine learning-based predictive model for colorectal polyp recurrence one year after endoscopic mucosal resection

BACKGROUND

Colorectal polyps are precancerous diseases of colorectal cancer. Early detection and resection of colorectal polyps can effectively reduce the mortality of colorectal cancer. Endoscopic mucosal resection (EMR) is a common polypectomy procedure in clinical practice, but it has a high postoperative recurrence rate. Currently, there is no predictive model for the recurrence of colorectal polyps after EMR.

AIM

To construct and validate a machine learning (ML) model for predicting the risk of colorectal polyp recurrence one year after EMR.

METHODS

This study retrospectively collected data from 1694 patients at three medical centers in Xuzhou. Additionally, a total of 166 patients were collected to form a prospective validation set. Feature variable screening was conducted using univariate and multivariate logistic regression analyses, and five ML algorithms were used to construct the predictive models. The optimal models were evaluated based on different performance metrics. Decision curve analysis (DCA) and SHapley Additive exPlanation (SHAP) analysis were performed to assess clinical applicability and predictor importance.

RESULTS

Multivariate logistic regression analysis identified 8 independent risk factors for colorectal polyp recurrence one year after EMR (P < 0.05). Among the models, eXtreme Gradient Boosting (XGBoost) demonstrated the highest area under the curve (AUC) in the training set, internal validation set, and prospective validation set, with AUCs of 0.909 (95%CI: 0.89-0.92), 0.921 (95%CI: 0.90-0.94), and 0.963 (95%CI: 0.94-0.99), respectively. DCA indicated favorable clinical utility for the XGBoost model. SHAP analysis identified smoking history, family history, and age as the top three most important predictors in the model.

CONCLUSION

The XGBoost model has the best predictive performance and can assist clinicians in providing individualized colonoscopy follow-up recommendations.

Exploring Potential Associations between Benzo[a]pyrene, Nicotine Exposure, and Lung Cancer: Molecular Insights, Prognostic Biomarkers, and Immune Cell Infiltration

Benzo[a]pyrene (BaP) and nicotine exposure have been implicated in lung cancer development. This study aims to elucidate the molecular mechanisms and potential biomarkers associated with this exposure in lung cancer patients. We integrated gene expression data from The Cancer Genome Atlas lung cancer cohort and the Comparative Toxicogenomics Database to identify differentially expressed genes (DEGs) associated with BaP and nicotine exposure. Enrichment analyses, survival analyses, and immune cell infiltration analyses were conducted to interpret the biological significance of these DEGs. A risk score model and a nomogram were constructed for a prognostic evaluation. We identified 163 DEGs related to BaP and nicotine exposure in lung cancer. Enrichment analysis revealed significant biological processes and pathways, including "IL-17 signaling", "cellular senescence", and "p53 signaling". From the DEGs, 34 prognostic genes were identified, with CLDN5, DNASE1L3, and GPR37 being independent prognostic factors. A risk score model based on these genes showed significant prognostic value, with high-risk patients exhibiting poorer survival outcomes. Additionally, a nomogram based on these risk scores demonstrated good predictive accuracy and clinical utility. Kaplan-Meier analyses confirmed that high expression of CLDN5 and GPR37 correlated with poor survival, while high DNASE1L3 expression indicated better survival. Single-gene enrichment analyses linked these genes to immune responses, cell adhesion, and DNA methylation. Immune cell infiltration analysis revealed significant correlations between the expression of these genes and the infiltration of various immune cell types. Our findings highlight the significant role of CLDN5, DNASE1L3, and GPR37 in lung cancer associated with BaP and nicotine exposure. The constructed risk score model and nomogram provide valuable tools for prognostication, and the identified genes offer potential targets for therapeutic intervention. Understanding the influence of toxic exposure on the tumor-immune microenvironment can guide future research and treatment strategies.

The challenges for developing prognostic prediction models for acute kidney injury in hospitalized children: A systematic review

Importance

Acute kidney injury (AKI) is common in hospitalized children which could rapidly progress into chronic kidney disease if not timely diagnosed. Prognostic prediction models for AKI were established to identify AKI early and improve children's prognosis.

Objective

To appraise prognostic prediction models for pediatric AKI.

Methods

Four English and four Chinese databases were systematically searched from January 1, 2010, to June 6, 2022. Articles describing prognostic prediction models for pediatric AKI were included. The data extraction was based on the CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies checklist. The risk of bias (ROB) was assessed according to the Prediction model Risk of Bias Assessment Tool guideline. The quantitative synthesis of the models was not performed due to the lack of methods regarding the meta-analysis of prediction models.

Results

Eight studies with 16 models were included. There were significant deficiencies in reporting and all models were considered at high ROB. The area under the receiver operating characteristic curve to predict AKI ranged from 0.69 to 0.95. However, only about one-third of models have completed internal or external validation. The calibration was provided only in four models. Three models allowed easy bedside calculation or electronic automation, and two models were evaluated for their impacts on clinical practice.

Interpretation

Besides the modeling algorithm, the challenges for developing prediction models for pediatric AKI reflected by the reporting deficiencies included ways of handling baseline serum creatinine and age-dependent blood biochemical indexes. Moreover, few prediction models for pediatric AKI were performed for external validation, let alone the transformation in clinical practice. Further investigation should focus on the combination of prediction models and electronic automatic alerts.

Predicting abnormal C-reactive protein level for improving utilization by deep neural network model

BACKGROUND

C-reactive protein (CRP) is an inflammatory biomarker frequently used in clinical practice. However, insufficient evidence-based ordering inevitably results in its overuse or underuse. This study aims to predict its normal and abnormal levels using the deep neural network (DNN) models, helping clinicians order this item more appropriately and intelligently.

METHODS

We considered complete blood count (CBC) parameters as feature vectors and 10 mg/L as a cutoff value for CRP. Several models, including linear support vector classification, logistic regression, decision trees, random forests, and DNN, were developed based on a dataset of 53834 medical records to predict binary output. We externally validated DNN models on independent 20723 samples through discrimination, calibration curve, and decision curve analysis.

RESULTS

DNN models has the best area under the receiver operating characteristic curves (AUC). Learning curves revealed that models' AUC, balanced accuracy, and F1 score do not significantly and continuously improve following increasing data volume. In internal validation, the AUC, balanced accuracy, and the F1 score of 10 models were 0.818 (0.95 CI: 0.812-0.824), 0.741 (0.95 CI: 0.736-0.747), and 0.649 (0.95 CI: 0.643-0.656), respectively. These metrics were 0.817 (0.95 CI: 0.816-0.817), 0.741 (0.95 CI: 0.740-0.742), and 0.641 (0.95 CI: 0.640-0.642), respectively, in external validation. AUC and balanced accuracy shown no significant difference (P-values were 0.106 and 0.339). CRP10-C2 model has the lowest Brier score of 0.154, AUC of 0.818, and calibration curve formula of y=1.001x-0.010, which was identified as a target model to deploy in the app.

CONCLUSIONS

DNN models obtained moderate performance, surpassing baseline indices in distinguishing binary CRP levels. They are good generalizations and well-calibrated. The CRP-C2 model can enhance CRP utilization by informing the orders appropriately and can contribute to inflammatory diagnostics in primary health care where CBC is available, but the CRP test is inaccessible.

A validation study of three early warning scores in early identification of gastric cancer patients with deteriorating condition after gastrectomy

OBJECTIVES

Early warning scores (EWS) aim to rapidly identify patients at risk of critical illness or life-threatening events before deterioration occurs in clinical settings. This study aims to validate the ability of three commonly used early warning scores, namely the National Early Warning Score (NEWS), the Early Warning Score (SEWS), and the Modified Early Warning Score (MEWS), to identify patients with deterioration after gastric cancer resection in general wards.

METHODS

This retrospective case-control study included 110 patients who experienced clinical deterioration after gastrectomy for gastric cancer as case group, and 745 patients without deterioration as control group from a tertiary hospital in Guangdong Province, China. The discriminating ability (receiver operating characteristic curves), calibration (goodness-of-fit test) and net benefit (clinical decision curves) of the three EWS (NEWS, SEWS, MEWS) were explored to compare their early warning performance for patients at risk of post-operative deterioration.

RESULTS

MEWS (goodness-of-fit p = 0.123 > 0.05) and SEWS (goodness-of-fit p = 0.235 > 0.05) both demonstrate good calibration and good discrimination ability (AUC 0.710, 95% CI 0.654-0.766;AUC 0.756, 95% CI 0.701-0.811). In contrast, NEWS not only has good calibration (goodness-of-fit p = 0.283 > 0.05) but also exhibits the best discrimination ability among the three scoring systems (AUC 0.835, 95% CI 0.785-0.884) and the highest net benefit.

CONCLUSION

Overall, NEWS may be more suitable for identifying gastric cancer patients at risk of post-operative clinical deterioration, as the early warning scoring model with best performance currently for post-gastrectomy observation.

A nomogram for postoperative pulmonary infections in esophageal cancer patients: a two-center retrospective clinical study

BACKGROUND

Postoperative pulmonary infections (POPIs) occur in approximately 13-38% of patients who undergo surgery for esophageal cancer, negatively impacting patient outcomes and prolonging hospital stays. This study aims to develop a novel clinical prediction model to identify patients at risk for POPIs early, thereby enabling timely intervention by clinicians.

METHODS

This study included 910 patients from two hospitals. Of these, 795 patients from one hospital were randomly assigned to the training cohort (n = 556) and the validation cohort (n = 239) at a 7:3 ratio. The external test cohort consisted of 115 patients from the second hospital. A nomogram was developed via logistic regression to predict the incidence of POPIs. The model's discrimination, precision and clinical benefit were evaluated by constructing a receiver operating characteristic (ROC) curve, calculating the area under the ROC curve (AUC), performing a calibration plot, conducting decision curve analysis (DCA) and clinical impact curves (CIC).

RESULTS

Multivariate logistic regression revealed that age, anemia, neoadjuvant therapy, T stage, thoracic adhesions and duration of surgery were independent risk factors for POPIs. The AUC for the training cohort was 0.8095 (95% CI: 0.7664-0.8527), that for the validation cohort was 0.8039 (95% CI: 0.7436-0.8643), and that for the external test cohort was 0.7174 (95% CI: 0.6145-0.8204). Calibration plots demonstrated good agreement between the predicted and observed probabilities, while DCA and CIC demonstrated good clinical applicability of the model in three cohorts.

CONCLUSION

The nomogram, which incorporates six key factors, effectively predicts the risk of POPIs and can serve as a valuable tool for clinicians in identifying high-risk patients.

A Risk Warning Model for Anemia Based on Facial Visible Light Reflectance Spectroscopy: Cross-Sectional Study

Background

Anemia is a global public health issue causing symptoms such as fatigue, weakness, and cognitive decline. Furthermore, anemia is associated with various diseases and increases the risk of postoperative complications and mortality. Frequent invasive blood tests for diagnosis also pose additional discomfort and risks to patients.

Objective

This study aims to assess the facial spectral characteristics of patients with anemia and to develop a predictive model for anemia risk using machine learning approaches.

Methods

Between August 2022 and September 2023, we collected facial image data from 78 anemic patients who met the inclusion criteria from the Hematology Department of Shanghai Hospital of Traditional Chinese Medicine. Between March 2023 and September 2023, we collected data from 78 healthy adult participants from Shanghai Jiading Community Health Center and Shanghai Gaohang Community Health Center. A comprehensive statistical analysis was performed to evaluate differences in spectral characteristics between the anemic patients and healthy controls. Then, we used 10 different machine learning algorithms to create a predictive model for anemia. The least absolute shrinkage and selection operator was used to analyze the predictors. We integrated multiple machine learning classification models to identify the optimal model and developed Shapley additive explanations (SHAP) for personalized risk assessment.

Results

The study identified significant differences in facial spectral features between anemic patients and healthy controls. The support vector machine classifier outperformed other classification models, achieving an accuracy of 0.875 (95% CI 0.825-0.925) for distinguishing between the anemia and healthy control groups. In the SHAP interpretation of the model, forehead-570 nm, right cheek-520 nm, right zygomatic-570 nm, jaw-570 nm, and left cheek-610 nm were the features with the highest contributions.

Conclusions

Facial spectral data demonstrated clinical significance in anemia diagnosis, and the early warning model for anemia risk constructed based on spectral information demonstrated a high accuracy rate.

A risk prediction model for venous thromboembolism in hospitalized patients with thoracic trauma: a machine learning, national multicenter retrospective study

BACKGROUND

Early treatment and prevention are the keys to reducing the mortality of VTE in patients with thoracic trauma. This study aimed to develop and validate an automatic prediction model based on machine learning for VTE risk screening in patients with thoracic trauma.

METHODS

In this national multicenter retrospective study, the clinical data of chest trauma patients hospitalized in 33 hospitals in China from October 2020 to September 2021 were collected for model training and testing. The data of patients with thoracic trauma at Shanghai Sixth People's Hospital from October 2021 to September 2022 were included for further verification. The performance of the model was measured mainly by the area under the receiver operating characteristic curve (AUROC) and the mean accuracy (mAP), and the sensitivity, specificity, positive predictive value, and negative predictive value were also measured.

RESULTS

A total of 3116 patients were included in the training and validation of the model. External validation was performed in 408 patients. The random forest (RF) model was selected as the final model, with an AUROC of 0·879 (95% CI 0·856-0·902) in the test dataset. In the external validation, the AUROC was 0.83 (95% CI 0.794-0.866), the specificity was 0.756 (95% CI 0.713-0.799), the sensitivity was 0.821 (95% CI 0.692-0.923), the negative predictive value was 0.976 (95% CI 0.958-0.993), and the positive likelihood ratio was 3.364.

CONCLUSIONS

This model can be used to quickly screen for the risk of VTE in patients with thoracic trauma. More than 90% of unnecessary VTE tests can be avoided, which can help clinicians target interventions to high-risk groups and ensure resource optimization. Although further validation and improvement are needed, this study has considerable clinical value.

Development of a predictive model for gastrointestinal side effects of metformin treatment in Chinese individuals with type 2 diabetes based on four randomised clinical trials

AIMS

This study aimed to build a model-based predictive approach to evaluate the gastrointestinal side effects following an initial metformin medication.

MATERIALS AND METHODS

The model was developed from data from four randomised clinical cohorts. A prediction model was established using integrated or simplified indicators. Ten machine learning models were used for the construction of predictive models. The Shapley values were used to report the features' contribution.

RESULTS

Four randomised clinical trial cohorts, including 1736 patients with type 2 diabetes, were first included in the analysis. Seventy percent of participants (1216) were allocated to the training set, 15% (260) were assigned to the internal validation set and 15% (260) were assigned to the test set. The Extra Tree model had the highest area under curve (AUC) (0.87) in the validation and test set. The top five crucial indicators were blood urea nitrogen (BUN), sex, triglyceride (TG), high-density lipoprotein-cholesterol (HDL-C) and total cholesterol (TC), and these five indicators were selected for constructing a simplified predictive model (AUC = 0.76). An online web-based tool was established based on the predictive model with integrated 17 features and top five indicators.

CONCLUSIONS

To predict gastrointestinal side effects in diabetic patients for initial use of metformin, a few easily obtained features are needed to establish the model. The model can be applied to the Chinese population in clinical practice.

Development and internal validation of a risk prediction model for dementia in a rural older population in China

INTRODUCTION

We sought to develop a practical tool for predicting dementia risk among rural-dwelling Chinese older adults.

METHODS

This cohort study included 2220 rural older adults (age ≥ 65 years) who were dementia-free at baseline (2014) and were followed in 2018. Dementia was diagnosed following the DSM-IV criteria. The prediction model was constructed using Cox models. We used C-index and calibration plots to assess model performance, and the decision curve analysis (DCA) to assess clinical usefulness.

RESULTS

During the 4-year follow-up, 134 individuals were diagnosed with dementia. We identified age, education, self-rated AD8 score, marital status, and stroke for the prediction model, with the C-index being 0.79 (95% confidence interval = 0.75-0.83) and the corrected C-index for internal validation being 0.79. Calibration plots showed good performance in predicting up to 4-year dementia risk and DCA indicated good clinical usefulness.

DISCUSSION

The 4-year dementia risk can be accurately predicted using five easily available predictors in a rural Chinese older population.

HIGHLIGHTS

We developed and internally validated a practical tool for dementia risk prediction among a rural older population in China. The prediction tool showed good discrimination and excellent calibration for predicting up to 4-year risk of dementia. The prediction tool can be used to identify individuals at a high risk for dementia for early preventive interventions.

Construction and validation of a nomogram prediction model for the catheter-related thrombosis risk of central venous access devices in patients with cancer: a prospective machine learning study

Central venous access devices (CVADs) are integral to cancer treatment. However, catheter-related thrombosis (CRT) poses a considerable risk to patient safety. It interrupts treatment; delays therapy; prolongs hospitalisation; and increases the physical, psychological and financial burden of patients. Our study aims to construct and validate a predictive model for CRT risk in patients with cancer. It offers the possibility to identify independent risk factors for CRT and prevent CRT in patients with cancer. We prospectively followed patients with cancer and CVAD at Xiangya Hospital of Central South University from January 2021 to December 2022 until catheter removal. Patients with CRT who met the criteria were taken as the case group. Two patients with cancer but without CRT diagnosed in the same month that a patient with cancer and CRT was diagnosed were selected by using a random number table to form a control group. Data from patients with CVAD placement in Qinghai University Affiliated Hospital and Hainan Provincial People's Hospital (January 2023 to June 2023) were used for the external validation of the optimal model. The incidence rate of CRT in patients with cancer was 5.02% (539/10 736). Amongst different malignant tumour types, head and neck (9.66%), haematological (6.97%) and respiratory (6.58%) tumours had the highest risks. Amongst catheter types, haemodialysis (13.91%), central venous (8.39%) and peripherally inserted central (4.68%) catheters were associated with the highest risks. A total of 500 patients with CRT and 1000 without CRT participated in model construction and were randomly assigned to the training (n = 1050) or testing (n = 450) groups. We identified 11 independent risk factors, including age, catheterisation method, catheter valve, catheter material, infection, insertion history, D-dimer concentration, operation history, anaemia, diabetes and targeted drugs. The logistic regression model had the best discriminative ability amongst the three models. It had an area under the curve (AUC) of 0.868 (0.846-0.890) for the training group. The external validation AUC was 0.708 (0.618-0.797). The calibration curve of the nomogram model was consistent with the ideal curve. Moreover, the Hosmer-Lemeshow test showed a good fit (P > 0.05) and high net benefit value for the clinical decision curve. The nomogram model constructed in this study can predict the risk of CRT in patients with cancer. It can help in the early identification and screening of patients at high risk of cancer CRT.

Factors associated with pulmonary complications after hepatectomy and establishment of nomogram: A real-world retrospective study

Background and Aims

Hepatectomy is currently the most effective way to treat liver diseases, and its safety has observably improved. However, the incidence of postoperative complications (POCs) remains high. Therefore, exploring the related influencing factors helps identify high-risk groups early and improve patient prognosis.

Methods

Clinical data were retrospectively collected from a real-world setting. Patients were divided into two groups based on the incidence of postoperative pulmonary complications (PPCs). Univariate analysis, LASSO regression, and logistic regression were applied to analyse the correlation between PPCs and perioperative indicators. A nomogram prediction model was constructed, whose discrimination, accuracy, and clinical effectiveness were evaluated.

Results

The incidence of PPCs was 36.33% among the 1244 patients in this study. The total length of hospital stay and perioperative mortality in the PPCs group were markedly higher (P < 0.001) than in the non-PPCs group. Logistic regression showed that surgical method [odds ratio (OR) =2.469 (95% CI: 1.665, 3.748); P < 0.001], duration of surgery [OR = 1.003 (95% CI: 1.002, 1.005); P < 0.001], postoperative patient destination [OR = 1.453 (95% CI: 1.115, 1.893); P = 0.006], and postoperative international normalised ratio (INR) [OR = 2.245 (95% CI: 1.287, 4.120); P = 0.007] were independent risk factors of PPCs; the number of clamping [OR = 0.988 (95% CI: 0.980, 0.995); P = 0.001] was an independent protective factor of PPCs. The area under the receiver operating characteristic (ROC) curve was 0.675 (95% CI: 0.638, 0.703), the consistency index of the calibration curve was 0.675 (95% CI: 0.641, 0.703), and the Hosmer-Lemeshow goodness-of-fit test yielded P = 0.327.

Conclusions

In this study, the incidence of PPCs after hepatectomy was the highest. Our nomogram model can predict the probability of PPCs after hepatectomy.

Prediction Tools in Spine Surgery: A Narrative Review

There have been increasing reports on prediction models in spine surgery. Interest in prognostic tools or risk calculators can facilitate shared decision-making about treatment between patients and clinicians. In recent years, there has been a steady increase in the number of models developed using varying methods. External validation is an essential component of prediction model testing to ensure the appropriate use of these models in populations outside of the developing center. This narrative review aimed to provide an overview of the literature describing the development and validation of prediction models in the field of spine surgery.

Association between triglyceride glucose index and adverse cardiovascular prognosis in patients with atrial fibrillation without diabetes: a retrospective cohort study

BACKGROUND

Atrial fibrillation (AF) is the most prevalent arrhythmia encountered in clinical practice. Triglyceride glucose index (Tyg), a convenient evaluation variable for insulin resistance, has shown associations with adverse cardiovascular outcomes. However, studies on the Tyg index's predictive value for adverse prognosis in patients with AF without diabetes are lacking.

METHODS

This retrospective study utilized electronic medical records to collect data on patients with AF hospitalized at West China Hospital from January to June 2020. Participants were categorized into three groups based on their Tyg index levels. The primary outcome, major adverse cardiovascular events, included cardiac death, stroke, and myocardial infarction. Kaplan-Meier curve, Cox proportional hazards regression model, and restricted cubic spline were employed to explore the relationship between the Tyg index and outcomes. The predictive performance of the CHA2DS2-VASc model was evaluated after incorporating the Tyg index.

RESULTS

The study comprised 864 participants (mean age 67.69 years, 55.32% male, 57.52% paroxysmal AF). Patients with high Tyg index had a significantly higher risk of developing major adverse cardiovascular events (MACE) (P < 0.001, hazard ratio: 2.05, 95% confidence interval:1.65-2.56). The MACE risk in the middle Tyg group was similar to that in the low Tyg group (P = 0.1) during the 48-month follow-up period. However, focusing on the last 24 months revealed a higher MACE risk (P = 0.015) in the middle Tyg group. The restricted cubic spline analysis revealed an S-shaped correlation between Tyg and MACE. The CHA2DS2-VASc model combined with the Tyg index showed improved predictive performance and net benefit.

CONCLUSIONS

A high Tyg index is associated with poorer prognosis in patients with AF without diabetes. Integrating the Tyg index into the CHA2DS2-VASc model may enhance its predictive performance, offering clinical utility.

Developing a rapid screening tool for high-risk ICU patients of sepsis: integrating electronic medical records with machine learning methods for mortality prediction in hospitalized patients-model establishment, internal and external validation, and visualization

OBJECTIVES

To develop a machine learning-based prediction model using clinical data from the first 24 h of ICU admission to enable rapid screening and early intervention for sepsis patients.

METHODS

This multicenter retrospective cohort study analyzed electronic medical records of sepsis patients using machine learning methods. We evaluated model performance in predicting sepsis outcomes within the first 24 h of ICU admission across US and Chinese healthcare settings.

RESULTS

From 31 clinical features, machine learning models demonstrated significantly better predictive performance than traditional approaches for sepsis outcomes. While linear regression achieved low test scores (0.25), machine learning methods reached scores of 0.78 and AUCs above 0.8 in testing. Importantly, these models maintained robust performance (scores 0.63-0.77) in external validation.

CONCLUSIONS

The application of machine learning-based prediction models for sepsis could significantly improve patient outcomes through early detection and timely intervention in the critical first 24 h of ICU admission, supporting clinical decision-making.

Prediction of outpatient rehabilitation patient preferences and optimization of graded diagnosis and treatment based on XGBoost machine learning algorithm

Background

The Department of Rehabilitation Medicine is key to improving patients' quality of life. Driven by chronic diseases and an aging population, there is a need to enhance the efficiency and resource allocation of outpatient facilities. This study aims to analyze the treatment preferences of outpatient rehabilitation patients by using data and a grading tool to establish predictive models. The goal is to improve patient visit efficiency and optimize resource allocation through these predictive models.

Methods

Data were collected from 38 Chinese institutions, including 4,244 patients visiting outpatient rehabilitation clinics. Data processing was conducted using Python software. The pandas library was used for data cleaning and preprocessing, involving 68 categorical and 12 continuous variables. The steps included handling missing values, data normalization, and encoding conversion. The data were divided into 80% training and 20% test sets using the Scikit-learn library to ensure model independence and prevent overfitting. Performance comparisons among XGBoost, random forest, and logistic regression were conducted using metrics, including accuracy and receiver operating characteristic (ROC) curves. The imbalanced learning library's SMOTE technique was used to address the sample imbalance during model training. The model was optimized using a confusion matrix and feature importance analysis, and partial dependence plots (PDP) were used to analyze the key influencing factors.

Results

XGBoost achieved the highest overall accuracy of 80.21% with high precision and recall in Category 1. random forest showed a similar overall accuracy. Logistic Regression had a significantly lower accuracy, indicating difficulties with nonlinear data. The key influencing factors identified include distance to medical institutions, arrival time, length of hospital stay, and specific diseases, such as cardiovascular, pulmonary, oncological, and orthopedic conditions. The tiered diagnosis and treatment tool effectively helped doctors assess patients' conditions and recommend suitable medical institutions based on rehabilitation grading.

Conclusion

This study confirmed that ensemble learning methods, particularly XGBoost, outperform single models in classification tasks involving complex datasets. Addressing class imbalance and enhancing feature engineering can further improve model performance. Understanding patient preferences and the factors influencing medical institution selection can guide healthcare policies to optimize resource allocation, improve service quality, and enhance patient satisfaction. Tiered diagnosis and treatment tools play a crucial role in helping doctors evaluate patient conditions and make informed recommendations for appropriate medical care.

Precision Prediction of Alzheimer's Disease: Integrating Mitochondrial Energy Metabolism and Immunological Insights

Alzheimer's disease (AD), a prevalent neurodegenerative disorder, is characterized by mitochondrial dysfunction and immune dysregulation. This study is aimed at developing a risk prediction model for AD by integrating multi-omics data and exploring the interplay between mitochondrial energy metabolism-related genes (MEMRGs) and immune cell dynamics. We integrated four GEO datasets (GSE132903, GSE29378, GSE33000, GSE5281) for differential gene expression analysis, functional enrichment, and weighted gene co-expression network analysis (WGCNA). We identified two key gene modules (turquoise and magenta) significantly correlated with AD. Subsequently, we constructed a risk prediction model incorporating five MEMRGs (MRPL15, RBP4, ABCA1, MPV17, and MRPL37) and clinical factors using LASSO regression. The model demonstrated robust predictive performance (AUC > 0.815) in both internal and external validation (GSE44770) cohorts. Downregulation of MRPL15, RBP4, MPV17, and MRPL37 in AD brain regions (validated using AlzData and qRT-PCR) suggests impaired mitochondrial function. Conversely, ABCA1 upregulation may represent a compensatory response. Furthermore, significant differences in immune cell proportions, particularly gamma delta T cells (p = 0.002) and activated CD4 memory T cells (p = 0.027), were found between AD and non-demented samples. We observed significant correlations between MEMRG expression and specific immune cell fractions, indicating a potential link between mitochondrial dysfunction and immune dysregulation in AD. Our study provides a reliable risk prediction model for AD and highlights the crucial roles of MEMRGs and immune responses in disease pathogenesis, offering potential targets for therapeutic interventions.

Baseline characteristics and 2-year functional outcome data of patients undergoing an arthroscopic rotator cuff repair in Switzerland, results of the ARCR_Pred study

The ARCR_Pred study was initiated to document and predict the safety and effectiveness of arthroscopic rotator cuff repair (ARCR) in a representative Swiss patient cohort. In the present manuscript, we aimed to describe the overall and baseline characteristics of the study, report on functional outcome data and explore case-mix adjustment and differences between public and private hospitals. Between June 2020 and November 2021, primary ARCR patients were prospectively enrolled in a multicenter cohort across 18 Swiss and one German orthopedic center. Baseline characteristics, including sociodemographic and diagnostic variables, were reported. Clinical scores and patient-reported outcome measures were assessed up to 24-month follow-up. After screening 2350 individuals, 973 patients with ARCR were included. Follow-up rates reached 99%, 95%, 89% and 88% at 6 weeks, 6, 12, and 24 months, respectively. While the proportion of massive tears was higher in the study population (44% vs. 20%, Std. Diff. = 0.56), there were no other major differences in key characteristics between enrolled and non-enrolled patients or in patients lost to follow-up. Functional scores improved over time, with positive changes rates ranging from 83% to 92% at 6-month, reaching 91% to 97% at 12- and 24-month follow-up. In linear mixed models, used to estimate the associations between baseline factors, hospital type and standardized 0-100 scores, marginal effects for time ranged from 20 to 30, 28 to 39 and 34 to 41 points at the 6-, 12- and 24-month follow-up, respectively. Except at the 12-month follow-up, where marginal effects for the interaction terms ranged from -5 to -4 points in the standardized scores, there were no consistent outcome differences between public and private hospitals. Increasing number of years of education was consistently associated with better scores, greater feelings of depression and anxiety, smoking and ASA group III-IV were consistently associated with worse scores. Tear severity showed a consistent negative association solely for the Constant-Score. The ARCR_Pred study shows high potential for generalizability to the population of patients undergoing an ARCR in Switzerland. Further analyses are needed to establish relevant clinimetrics for the Swiss population and to compare outcomes for surgical techniques, surgeon experiences profiles and post-operative management.

Derivation and validation of a prediction model for inadequate bowel preparation in Chinese outpatients

The quality of bowel preparation is an important factor in the success of colonoscopy. However, multiple influencing factors that function together can lead to inadequate bowel preparation. The main objective of this study was to explore the specific factors that affect the quality of bowel preparation, with the goal of deriving and validating a predictive model for inadequate bowel preparation in Chinese outpatients. A prospective observational study. We conducted a prospective study in a tertiary hospital in Zhejiang Province that included elective colonoscopy outpatients treated between December 15, 2022 and August 12, 2023. Clinical data related to the patient characteristics and patient bowel preparation were collected to analyze the risk factors of inadequate bowel preparation in outpatients. The quality of bowel preparation was assessed by using the Boston bowel preparation scale (BBPS). Inadequate bowel preparation was defined as a total BBPS score of < 6 or any segment score of < 2. The predictive model was constructed based on multivariate logistic regression, and the discrimination and calibration of the prediction model were evaluated via internal and external validation; additionally, a clinical decision curve was drawn to evaluate the clinical utility of the model. A total of 1314 patients were included from December 15, 2022 through May 31, 2023 (derivation cohort, n = 1035) and from June 1 through August 12, 2023 (external validation cohort, n = 279). Inadequate bowel preparation occurred in 260 of 1035 patients in the derivation cohort (25.1%). Multivariate analysis identified that male sex (OR = 1.690, 95% CI: 1.242-2.300), diabetes (OR = 1.769, 95% CI: 1.059-2.954), constipation (OR = 2.375, 95% CI: 1.560-3.617), history of colorectal surgery (OR = 2.915, 95% CI: 1.455-5.840), a high fiber diet used at 24 h before examination (OR = 2.662, 95% CI: 1.636-4.334) and the time interval from the end of the bowel preparation to the start of the colonoscopy (PC) >5 h (OR = 2.471, 95% CI: 1.814-3.366) were independent risk factors. We derived a model to identify patients with inadequate cleansing by using data from patients in the derivation cohort and tested it in the validation cohort. The area under the curve (AUC) was 0.704 (95% CI: 0.667-0.741), with a calibration value of p = 0.632 in the derivation cohort. Bootstrap cross-validation showed a good model calibration condition. For the validation cohort, the AUC was 0.704 (95% CI: 0.628-0.779), and the calibration value was p = 0.376. We identified the influencing factors of outpatient bowel cleansing that are associated with patient clinical characteristics and bowel preparation-related behaviors. A model was constructed and validated to identify patients who were at high risk of inadequate bowel preparation by using six simple variables, which included male sex, diabetes, constipation, history of colorectal surgery, a high fiber diet used at 24 h prior to examination, and PC > 5 h. The clinical decision curve showed that the constructed prediction model has some clinical utility based on results from the derivation cohort and validation cohort.

Development and Validation of a Machine Learning-Based Nomogram for Prediction of Unplanned Reoperation Postspinal Surgery Within 30 Days

BACKGROUND

Unplanned reoperation postspinal surgery (URPS) leads to prolonged hospital stays, higher costs, decreased patient satisfaction, and adversely affects postoperative rehabilitation. This study aimed to develop and validate prediction models (nomograms) for early URPS risk factors using machine learning methods, aiding spine surgeons in designing prevention strategies, promoting early recovery, reducing complications, and improving patient satisfaction.

METHODS

Medical records of 639 patients who underwent reoperation postspinal surgery from the First Affiliated Hospital of Air Force Medical University (2018-2022) were collected, including baseline indicators, perioperative indicators, and laboratory indicators. After applying inclusion and exclusion criteria, 122 URPS and 155 non-URPS patients were identified and randomly divided into training (82 URPS and 111 non-URPS) and validation (40 URPS and 44 non-URPS) cohorts. Three machine learning methods (least absolute shrinkage and selection operator regression, Random Forest, and Support Vector Machine Recursive Feature Elimination) were used to select feature variables, and their intersection was used to develop the prediction model, tested on the validation cohort.

RESULTS

Six factors-implant, postoperative suction drainage, gelatin sponge, anticoagulants, antibiotics, and disease type-were identified to construct a nomogram diagnostic model. The area under the curve of this nomogram was 0.829 (95% confidence interval 0.771-0.886) in the training cohort and 0.854 (95% confidence interval 0.775-0.933) in the validation cohort. Calibration curves demonstrated satisfactory agreement between predictions and actual probabilities. The decision curve indicated clinical usefulness with a threshold between 1% and 90%.

CONCLUSIONS

The established model can effectively predict URPS in patients and can assist spine surgeons in devising personalized and rational clinical prevention strategies.

Development of a Predictive Hospitalization Model for Skilled Nursing Facility Patients

OBJECTIVES

Identifying skilled nursing facility (SNF) patients at risk for hospitalization or death is of interest to SNFs, patients, and patients' families because of quality measures, financial penalties, and limited clinical staffing. We aimed to develop a predictive model that identifies SNF patients likely to be hospitalized or die within the next 7 days and validate the model's performance against clinician judgment.

DESIGN

Retrospective multivariate prognostic model development study.

SETTING AND PARTICIPANTS

Patients in US SNFs that use the PointClickCare electronic health record (EHR) system. We used data from the first 100 days of skilled stays for 5,642,474 patients in 8440 SNFs, from January 1, 2019, through March 31, 2023.

METHODS

We used data collected in the course of clinical care to develop a machine learning model to predict the likelihood of patient hospitalization or death within the next 7 days. The data included vital signs, diagnoses, laboratory results, food intake, and clinical notes. We also asked SNF nurses and hospital case managers to make their own predictions as a comparison. The EHR was used as the source of information on whether the patient died or was hospitalized.

RESULTS

The model had sensitivity of 35%, specificity of 92%, positive predictive value (PPV) of 18%, and area under the receiver operator curve (AUC) of 0.75. A variation of the model in which we did not include progress notes and food intake achieved an AUC of 0.70. Nurse raters achieved a sensitivity of 61%, specificity of 73%, and PPV of 10%.

CONCLUSIONS AND IMPLICATIONS

Machine learning models can accurately predict the likelihood of hospitalization or death within the next 7 days among SNF patients. These models do not require additional SNF staff time and may be useful in readmission reduction programs by targeting more frequent monitoring proactively to those at highest risk.

Radiomics and deep learning features of pericoronary adipose tissue on non-contrast computerized tomography for predicting non-calcified plaques

BACKGROUND

Inflammation of coronary arterial plaque is considered a key factor in the development of coronary heart disease. Early the plaque detection and timely treatment of the atherosclerosis could effectively reduce the risk of cardiovascular events. However, there is no study combining radiomics with deep learning techniques to predict non-calcified plaques (NCP) in coronary artery at present.

OBJECTIVE

To investigate the value of combination of radiomics and deep learning features based on non-contrast computerized tomography (CT) scans of pericoronary adipose tissue (PCAT), integrating with clinical risk factors of patients, in identifying coronary inflammation and predicting the presence of NCP.

METHODS

The clinical and imaging data of 353 patients were analyzed. The region of interest (ROI) of PCAT was manually outlined on non-contrast CT scan images, like coronary CT calcium score sequential images, then the radiomics and deep learning features in ROIs were extracted respectively. In training set (Center 1), after performing feature selection, radiomics and deep learning feature models were established, meanwhile, clinical models were built. Finally, combined models were developed out via integrating clinical, radiomics, and deep learning features. The predictive performance of the four feature model groups (clinical, radiomics, deep learning, and three combination) was assessed by seven different machine learning models through generation of receiver operating characteristic curves (ROC) and the calculation of area under the curve (AUC), sensitivity, specificity, and accuracy. Furthermore, the predictive performance of each model was validated in an external validation set (Center 2).

RESULTS

For the single model comparation, eXtreme Gradient Boosting (XGBoost) showed the best performance among the clinical model group in the validation set. And Random Forest (RF) exhibited the best indicative performance not only among the radiomics feature group but also in the deep learning feature model group. What's more, among the combined model group, RF still displayed the best predictive performance, with the value of AUC, sensitivity, specificity, and accuracy in the validation set are 0.963, 0.857, 0.929, and 0.905.

CONCLUSION

The RF model in the combined model group based on non-contrast CT scan PCAT can predict the presence of NCP more accurately and has the potential for preliminary screening of the NCP.

Creation of nomograms that combine clinical, CT, and radiographic features to separate benign from malignant diseases using spiculation or (and) lobulation signs

BACKGROUND

Distinguishing between benign and malignant pulmonary nodules based on CT imaging features such as the spiculation sign and/or lobulation sign remains challenging and these nodules are often misinterpreted as malignant tumors. this retrospective study aimed to develop a prediction model to estimate the likelihood of benign and malignant lung nodules exhibiting spiculation and/or lobulation signs.

METHODS

A total of 500 patients with pulmonary nodules from June 2022 to August 2024 were retrospectively analyzed. Among them, 190 patients with spiculation sign and lobar sign or both on CT scan were included in this study. This investigation collected the clinical information, preoperative chest CT imaging characteristics, and postoperative histopathologic results from patients.Univariate and multivariate logistic regression analyses were employed to identify independent risk factors, from which a prediction model and nomogram were developed. In addition, The model performance was assessed through receiver operating characteristic(ROC) curve analysis, calibration curve analysis, and decision curve analysis (DCA).

RESULTS

In our study, 190 patients with pulmonary nodules underwent lung biopsy in 10 patients and surgical resection in 180 patients, of whom 53 were benign nodules and 137 were malignant nodules. When combined with the spiculation sign or (and) the lobulation sign, the vascular cluster sign, bronchial architectural distortion, bubble-like translucent area, nodule density, and CEA were found to be significant independent predictors for determining the benignity and malignancy of pulmonary nodules. The nomogram prediction model demonstrated high predictive accuracy with an area under the ROC curve (AUC) of 0.904. Furthermore, the model's calibration curve demonstrated adequate calibration. DCA confirmed the prediction model's validity.

CONCLUSION

The model can assist clinicians in making more accurate preoperative diagnoses and in guiding clinical decision-making regarding treatment, potentially reducing unnecessary surgical interventions.

Prognostic value of lymph node metrics in lung squamous cell carcinoma: an analysis of the SEER database

INTRODUCTION

Although the Tumor-Node-Metastasis (TNM) staging system is widely used for staging lung squamous cell carcinoma (LSCC), the TNM system primarily emphasizes tumor size and metastasis, without adequately considering lymph node involvement. Consequently, incorporating lymph node metastasis as an additional prognostic factor is essential for predicting outcomes in LSCC patients.

METHODS

This retrospective study included patients diagnosed with LSCC between 2004 and 2018 and was based on data from the Surveillance, Epidemiology, and End Results (SEER) database of the National Cancer Institute. The primary endpoint of the study was cancer-specific survival (CSS), and demographic characteristics, tumor characteristics, and treatment regimens were incorporated into the predictive model. The study focused on the value of indicators related to pathological lymph node testing, including the lymph node ratio (LNR), regional node positivity (RNP), and lymph node examination count (RNE), in the prediction of cancer-specific survival in LSCC. A prognostic model was established using a multivariate Cox regression model, and the model was evaluated using the C index, Kaplan-Meier, the Akaike information criterion (AIC), decision curve analysis (DCA), continuous net reclassification improvement (NRI), and integrated discrimination improvement (IDI), and the predictive efficacy of different models was compared.

RESULTS

A total of 14,200 LSCC patients (2004-2018) were divided into training and validation cohorts. The 10-year CSS rate was approximately 50%, with no significant survival differences between cohorts (p = 0.8). The prognostic analysis revealed that models incorporating LNR, RNP, and RNE demonstrated superior performance over the TNM model. The LNR and RNP models demonstrated better model fit, discrimination, and reclassification, with AUC values of 0.695 (training) and 0.665 (validation). The RNP and LNR models showed similar predictive performance, significantly outperforming the TNM and RNE models. Calibration curves and decision curve analysis confirmed the clinical utility and net benefit of the LNR and RNP models in predicting long-term CSS for LSCC patients, highlighting their value in clinical decision-making.

CONCLUSION

This study confirms that RNP status is an independent prognostic factor for CSS in LSCC, with predictive efficacy comparable to LNR, with both models enhancing survival prediction beyond TNM staging.

Validation and modification of existing bleeding complications prediction models for percutaneous renal biopsy: a prospective study

Background

Bleeding complications following percutaneous renal biopsy (PRB) are a significant clinical concern. This study aimed to validate and refine existing prediction models for post-biopsy bleeding to support more accurate clinical decision-making.

Methods

Clinical data from 471 PRB patients were examined in this prospective analysis. Ultrasounds were performed immediately and 6 h post-biopsy to identify perinephric hematomas. Patients exhibiting severe pain, a hemoglobin drop of >10 g/L, symptomatic hypotension, hematuria within 7 days post-procedure underwent repeat ultrasound to assess for bleeding complications. Univariate and multivariable logistic regression analyses were conducted to identify factors associated with bleeding risk. The predictive performance of three kidney biopsy risk calculators (KBRC) was evaluated using the area under the receiver operating characteristic (AUROC) curve, net reclassification improvement (NRI), integrated discrimination improvement (IDI), and decision curve analysis (DCA) to determine clinical utility. Nomograms were developed for each model to facilitate clinical application.

Results

Univariate analysis identified body mass index (BMI), hemoglobin, and ultrasound findings as significant predictors of bleeding complications. In multivariable analysis, BMI, immediate ultrasound, and 6-h ultrasound data remained significant (p < 0.05). The three models compared included: KBRC-5 (age, body mass index (BMI), platelet count, hemoglobin, kidney size), KBRC-5 with immediate ultrasound data (IKBRC), and KBRC-5 with 6-h hematoma size (SKBRC). The AUROC values for these models were 0.683, 0.786, and 0.867, respectively (p < 0.001). NRI and IDI analyses demonstrated that adding immediate or 6-h ultrasound data significantly improved the risk reclassification ability of the KBRC-5 model (p < 0.05). DCA indicated that IKBRC provided the highest net benefit for risk thresholds between 25% and 77%, while SKBRC was superior for thresholds between 10% and 95%. Nomograms were constructed for each model, allowing clinicians to estimate the probability of bleeding complications by summing scores for each predictor. Calibration curves showed good agreement between predicted and observed probabilities.

Conclusion

Incorporating real-time ultrasound data post-PRB significantly enhances the predictive accuracy and risk reclassification capability of bleeding risk models. These findings provide critical insights for guiding clinical management decisions in patients undergoing renal biopsy.

Predicting Early Hospital Discharge Following Revision Total Hip Arthroplasty: An Analysis of a Large National Database Using Machine Learning

BACKGROUND

Revision total hip arthroplasty (rTHA) was recently removed from the Medicare inpatient-only list. However, appropriate candidate selection for outpatient rTHA remains paramount. The purpose of this study was to evaluate the utility of a large national database using machine learning (ML) and traditional multivariable logistic regression (MLR) models in predicting early hospital discharge (EHD) (< 24 hours) following rTHA. Furthermore, this study aimed to use the trained ML models, cross-referenced with traditional MLR, to determine key perioperative variables predictive of EHD following rTHA.

METHODS

Data were obtained from a large national database from 2021. Patients who had unilateral rTHA procedures were included. Demographic, preoperative, and operative variables were analyzed as inputs for the models. An ML regression model and various ML techniques were used to predict EHD and were compared using the area under the curve, calibration, Brier score, and decision curve analysis. Feature importance was identified from the overall best-performing model. Of the 3,097 patients in this study, 866 (27.96%) underwent EHD.

RESULTS

The random forest model performed the best overall and identified aseptic surgical indication, operative time < three hours, absence of anemia (hematocrit < 40% in men and < 35% in women), neuraxial anesthesia type, White race, men, independent functional status, body mass index > 20, age < 75 years, and the presence of home support as factors predictive of EHD. Each of these variables was also significant in the MLR model.

CONCLUSIONS

Each ML model and MLR displayed good performance and identified clinically important variables for determining candidates for EHD following rTHA. Machine learning (ML) techniques such as random forest may allow clinicians to accurately risk stratify their patients preoperatively to optimize resources and improve patient outcomes.

Machine learning-based diagnostic model of lymphatics-associated genes for new therapeutic target analysis in intervertebral disc degeneration

Background

Low back pain resulting from intervertebral disc degeneration (IVDD) represents a significant global social problem. There are notable differences in the distribution of lymphatic vessels (LV) in normal and pathological intervertebral discs. Nevertheless, the molecular mechanisms of lymphatics-associated genes (LAGs) in the development of IVDD remain unclear. An in-depth exploration of this area will help to reveal the biological and clinical significance of LAGs in IVDD and may lead to the search for new therapeutic targets for IVDD.

Methods

Data sets were obtained from the Gene Expression Omnibus (GEO) database. Following quality control and normalization, the datasets (GSE153761, GSE147383, and GSE124272) were merged to form the training set, with GSE150408 serving as the validation set. LAGs from GeneCards, MSigDB, Gene Ontology, and KEGG database. The Venn diagram was employed to identify differentially expressed lymphatic-associated genes (DELAGs) that were differentially expressed in the normal and IVDD groups. Subsequently, four machine learning algorithms (SVM-RFE, Random Forest, XGB, and GLM) were used to select the method to construct the diagnostic model. The receiver operating characteristic (ROC) curve, nomogram, and Decision Curve Analysis (DCA) were used to evaluate the model effect. In addition, we constructed a potential drug regulatory network and competitive endogenous RNA (ceRNA) network for key LAGs.

Results

A total of 15 differentially expressed LAGs were identified. By comparing four machine learning methods, the top five genes of importance in the XGB model (MET, HHIP, SPRY1, CSF1, TOX) were identified as lymphatics-associated gene diagnostic signatures. This signature was used to predict the diagnosis of IVDD with strong accuracy and an area under curve (AUC) value of 0.938. Furthermore, the diagnostic model was validated in an external dataset (GSE150408), with an AUC value of 0.772. The nomogram and DCA further prove that the diagnosis model has good performance and predictive value. Additionally, drug regulatory networks and ceRNA networks were constructed, revealing potential therapeutic drugs and post-transcriptional regulatory mechanisms.

Conclusion

We developed and validated a lymphatics-associated genes diagnostic model by machine learning algorithms that effectively identify IVDD patients. These five key LAGs may be potential therapeutic targets for IVDD patients.

Development and validation of a nomogram for predicting survival in gastric signet ring cell carcinoma patients treated with radiotherapy

There is no effective clinical prediction model to predict the prognosis of gastric signet ring cell carcinoma (GSRC) patients treated with radiotherapy. This study retrospectively analyzed the clinical data of 20-80-year-old patients diagnosed with GSRC between 2004 and 2019 from the Surveillance, Epidemiology, and End Results (SEER) database. Using Cox regression analyses revealed independent prognostic factors, and a nomogram was constructed. The C-index, net reclassification index (NRI) and integrated discrimination improvement (IDI) of the nomogram were greater than those of the TNM staging system for predicting OS, indicating that the nomogram predicted prognosis with greater accuracy. The area under the curve (AUC) values were 0.725, 0.753 and 0.745 for the training group; 0.725, 0.763 and 0.752 for the internal validation group; and 0.795, 0.764 and 0.765 for the external validation group, respectively. Calibration plots demonstrated high agreement between the nomogram's prediction and the actual observations. The risk stratification system was able to accurately stratify patients who underwent radiotherapy for GSRC into high- and low-risk subgroups, with significant differences in prognosis. The Kaplan‒Meier survival analysis according to different treatments indicated that surgery combined with chemoradiotherapy is a more effective treatment strategy for improving OS in for GSRC patients. The nomogram is sufficiently accurate to predict the prognostic factors of GSRC receiving radiotherapy, allowing for clinicians to predict the 1-, 3-, and 5-year OS.

Validating Machine Learning Models Against the Saline Test Gold Standard for Primary Aldosteronism Diagnosis

Background

In this study, we developed and validated machine learning models to predict primary aldosteronism (PA) in hypertensive East-Asian patients, comparing their performance against the traditional saline infusion test. The motivation for this development arises from the need to provide a more efficient and standardized diagnostic approach, because the saline infusion test, although considered a gold standard, is often cumbersome, is time-consuming, and lacks uniform protocols. By offering an alternative diagnostic method, this study seeks to enhance patient care through quicker and potentially more reliable PA detection.

Objectives

This study sought to both develop and evaluate the performance of machine learning models in detecting PA among hypertensive participants, in comparison to the standard saline loading test.

Methods

We used patient data from 3 distinct cohorts: TAIPAI (Taiwan Primary Aldosteronism Investigation), CONPASS (Chongqing Primary Aldosteronism Study), and a South Korean cohort. Random Forest's importance scores, XGBoost, and deep learning techniques are adopted to identify the most predictive features of primary aldosteronism.

Results

We present detailed results of the model's performance, including accuracy, sensitivity, and specificity. The Random Forest model achieved an accuracy of 0.673 (95% CI: 0.640-0.707), significantly outperforming the baseline models.

Conclusions

In our discussion, we address both the strengths and limitations of our study. Although the machine learning models demonstrated superior performance in predicting primary aldosteronism, the generalizability of these findings may be limited to East-Asian hypertensive populations. Future studies are needed to validate these models in diverse demographic settings to enhance their applicability.

Predictive models of sepsis-associated acute kidney injury based on machine learning: a scoping review

BACKGROUND

With the development of artificial intelligence, the application of machine learning to develop predictive models for sepsis-associated acute kidney injury has made potential breakthroughs in early identification, grading, diagnosis, and prognosis determination.

METHODS

Here, we conducted a systematic search of the PubMed, Cochrane Library, Embase (Ovid), Web of Science, and Scopus databases on April 28, 2023, and screened relevant literature. Then, we comprehensively extracted relevant data related to machine learning algorithms, predictors, and predicted objectives. We subsequently performed a critical evaluation of research quality, data aggregation, and analyses.

RESULTS

We screened 25 studies on predictive models for sepsis-associated acute kidney injury from a total of originally identified 2898 studies. The most commonly used machine learning algorithm is traditional logistic regression, followed by eXtreme gradient boosting. We categorized these predictive models into early identification models (60%), prognostic prediction models (32%), and subtype identification models (8%) according to their predictive purpose. The five most commonly used predictors were serum creatinine levels, lactate levels, age, blood urea nitrogen concentration, and diabetes mellitus. In addition, a single data source, insufficient assessment of clinical utility, lack of model bias assessment, and hyperparameter adjustment may be the main reasons for the low quality of the current research.

CONCLUSIONS

However, studies on the nondeath prognostic outcomes, the long-term clinical outcomes, and the subtype identification models are insufficient. Additionally, the poor quality of the research and the insufficient practicality of the model are problems that need to be addressed urgently.

Validation of a Urine-Based Proteomics Test to Predict Clinically Significant Prostate Cancer: Complementing mpMRI Pathway

INTRODUCTION

Prostate cancer (PCa) is the most frequently diagnosed cancer among men. A major clinical need is to accurately predict clinically significant PCa (csPCa). A proteomics-based 19-biomarker model (19-BM) was previously developed using capillary electrophoresis-mass spectrometry (CE-MS) and validated in close to 1,000 patients at risk for PCa. This study aimed to validate 19-BM in a multicenter prospective cohort of 101 biopsy-naive patients using current diagnostic pathways.

METHODS

Urine samples from 101 patients with suspicious of PCa were analyzed using CE-MS. All patients underwent multiparametric or magnetic resonance imaging (mpMRI) using a 3-T system. The 19-BM score was estimated using support vector machine-based software (MosaCluster v1.7.0), employing the previously published cut-off criterion of -0.07. Diagnostic nomograms were investigated along with mpMRI.

RESULTS

Independent validation of 19-BM yielded a sensitivity of 77% and a specificity of 85% (AUC:0.81). This performance surpassed those of prostate-specific antigen (PSA; AUC:0.56) and PSA density (AUC:0.69). For PI-RADS≤ 3 patients, 19-BM showed a sensitivity of 86% and a specificity of 88%. Integrating 19-BM with mpMRI resulted in significantly better accuracy (AUC:0.90) compared to individual investigations alone (AUC19BM = 0.81; p = 0.004 and AUCmpMRI: 0.79; p = 0.001). Examining the decision curve analysis, 19-BM with mpMRI surpassed other approaches for the prevailing risk interval from a 30% cut-off.

CONCLUSIONS

19-BM exhibited favorable reproducibility for the prediction of csPCa. In patients with PI-RADS ≤3, 19-BM correctly classified 88% of the patients with insignificant PCa at the cost of 1 missed csPCa patient. Utilizing the 19-BM test could prove valuable in complementing mpMRI and reducing the need for unnecessary biopsies.

Predicting the Time to Relapse Following Withdrawal from Different Biologics in Patients with Psoriasis who Responded to Therapy: A 12-Year Multicenter Cohort Study

BACKGROUND

For patients with psoriasis, discontinuation of biologics following remission has become more common in daily practice.

OBJECTIVE

We aimed to identify predictors and construct a predictive model for time to relapse following withdrawal from biologics.

METHODS

This 12-year, multicenter, observational cohort study was performed in six dermatology centers between February 2011 and February 2024. We identified biological treatment episodes in patients with moderate-to-severe psoriasis and included only treatment episodes in which a clinical response (≥ 50% reduction in Psoriasis Area and Severity Index score [PASI 50] from baseline) was achieved and the patient withdrew from biological therapy with a well-controlled status (PASI < 10 and ≥ 50% improvement in PASI from baseline). The primary outcome was time to relapse, which was defined as the period from the last biologic administration to relapse. An extended multivariate Cox proportional hazards analysis (Prentice-Williams-Peterson Gap time model) was used to predict relapse and generate a predictive model.

RESULTS

This study screened 1613 biological treatment episodes, and 991 treatment episodes were enrolled. The time to relapse decreased significantly as the number of previous withdrawals from biological treatment increased (p < 0.001). Similarly, the time to relapse decreased significantly as the number of previous biologics used increased (p < 0.001). The maximum PASI improvement during biological treatment decreased and the PASI score at withdrawal of biological treatment increased in parallel as the number of prior withdrawals from biologics increased. The time to relapse following withdrawal was longest for interleukin (IL)-23 inhibitors (IL-23i), followed by the IL-12/23i, IL-17 inhibitors (IL-17i), and tumor necrosis factor-α inhibitors. After adjustment, multivariate Cox regression identified the following significant predictors of relapse following withdrawal: the mechanisms of action of biologics (hazard ratio [HR] for IL-17i vs IL-12/23i, 1.59; HR for IL-23i vs IL-12/23i, 0.60), number of previous withdrawals from biological treatment (HR 1.23; 95% confidence interval [CI] 1.13‒1.33), time to achieve PASI 50 (HR 1.01; 95% CI 1.00‒1.02), maximum PASI improvement on biologics (HR 0.98; 95% CI 0.98‒0.99), and PASI at the end of therapy (HR 1.03; 95% CI 1.01‒1.05). The model had good predictive and discriminative ability.

CONCLUSIONS

These results have the potential to help physicians and patients make individualized treatment decisions; information on the risk of relapse of psoriasis at specific timepoints following the withdrawal of biologics is particularly valuable for patients considering discontinuation of biologics or as-needed biologic therapy. However, the benefit and risk of repeated withdrawals of biologics should be carefully weighed, as the treatment efficacy and duration of remission decline as the number of withdrawals increases.

Construction of ceRNA Network and Disease Diagnosis Model for Keloid Based on Tumor Suppressor ERRFI1

The aim of this study is to identify the key biomarker of keloid (KD) with significant diagnostic value and to construct the related competing endogenous RNA (ceRNA) network and disease diagnostic model to provide new ideas for the early diagnosis and prevention of KD. Public databases were used to identify the key gene of KD. Enrichment analysis and immune cell infiltration (ICI) analysis revealed its functional and immune characteristics. Then, a ceRNA network was constructed to explore the potential pathways of it. Random forest (RF) analysis was applied to construct a predictive model for the disease diagnosis of KD. Finally, immunohistochemistry (IHC) and RT-qPCR were used to verify the differential expression of key gene. ERRFI1 was identified as a key biomarker in KD and was lowly expressed in KD. The ceRNA network revealed that H0TAIRM1-has-miR-148a-3p-ERRFI1 may be a potential pathway in KD. Finally, a 2-gene diagnostic prediction model (ERRFI1, HSD3B7) was constructed and externally validated and the results suggested that the model had good diagnostic performance. ERRFI1 is a downregulated gene in KD and is expected to be a promising predictive marker and disease diagnostic gene. ICI may play a role in the progression of KD. The ceRNA network may provide new clues to the potential pathogenesis of KD. Finally, the new KD diagnostic model could be an effective tool for assessing the risk of KD development.

Development and validation of a nomogram for assessing comorbidity and frailty in triage: a multicentre observational study

Assessing patient frailty in the Emergency Department (ED) is crucial; however, triage frailty and comorbidity assessment scores developed in recent years are unsatisfactory. The underlying causes of this phenomenon could reside in the nature of the tools used, which were not designed specifically for the emergency context and, thus, are difficult to adapt to the emergency environment. The objective of this study was to create and internally validate a nomogram for identifying different levels of patient frailty during triage. Multicenter, prospective, observational exploratory study conducted in two ED. The study was conducted from April 1 to October 31, 2022. Following the triage assessment, the nurse collected variables related to the patient's comorbidities and chronic conditions using a predefined form. The primary outcome was the 90-day mortality rate. A total of 1345 patients were enrolled in this study; 6% died within 90 days. In the multivariate analysis, the Charlson Comorbidity Index, an altered motor condition, an altered cognitive condition, an autonomous chronic condition, arrival in an ambulance, and a previous hospitalization within 90 days were independently associated with death. The internal validation of the nomogram reported an area under the receiver operating characteristic of 0.91 (95% CI 0.884-0.937). A nomogram was created for assessing comorbidity and frailty during triage and was demonstrated to be capable of determining comorbidity and frailty in the ED setting. Integrating a tool capable of identifying frail patients at the first triage assessment could improve patient stratification.

Early death prediction model for breast cancer with synchronous lung metastases: an analysis of the SEER database

Background

Breast cancer with lung metastases (BCLM) is a serious condition that often leads to early death. This study aims to screen the risk factors of early death in BCLM patients and establish a simple and accurate nomogram prediction model. Identifying prognostic markers and developing accurate prediction models can help guide clinical decision-making.

Methods

The Surveillance, Epidemiology, and End Results (SEER) database was used to analyze a sizable sample of data, encompassing 4,238 BCLM patients diagnosed between 2010 and 2015. Stepwise regression was used to manage the collinearity of variables and to construct a prediction model based on the histogram. The results were subjected to internal validation and contrasted with those of related investigations.

Results

Of the 4,238 BCLM patients in this study, 3,232 did not die early. Of the 1,006 premature deaths, 891 were cancer specific. Lymph node involvement, tumor size, age, and race were all recognized as prognostic markers for premature mortality. A nomogram was constructed based on these factors to reliably predict cancer-specific death and early all-cause death.

Conclusions

This study gives new insights into the prognosis of individuals with BCLM and finds critical prognostic variables for early mortality. The created nomogram might assist physicians in identifying individuals at high risk of early mortality and making treatment options.

Nationwide study on development and validation of a risk prediction model for CIN3+ and cervical cancer in Estonia

Transitioning to an individualized risk-based approach can significantly enhance cervical cancer screening programs. We aimed to derive and internally validate a prediction model for assessing the risk of cervical intraepithelial neoplasia grade 3 or higher (CIN3+) and cancer in women eligible for screening. This retrospective study utilized data from the Estonian electronic health records, including 517,884 women from the health insurance database and linked health registries. We employed Cox proportional hazard regression, incorporating reproductive and medical history variables (14 covariates), and utilized the least absolute shrinkage and selection operator (LASSO) for variable selection. A 10-fold cross-validation for internal validation of the model was used. The main outcomes were the performance of discrimination and calibration. Over the 8-year follow-up, we identified 1326 women with cervical cancer and 5929 with CIN3+, with absolute risks of 0.3% and 1.1%, respectively. The prediction model for CIN3 + and cervical cancer had good discriminative power and was well calibrated Harrell's C of 0.74 (0.73-0.74) (calibration slope 1.00 (0.97-1.02) and 0.67 (0.66-0.69) (calibration slope 0.92 (0.84-1.00) respectively. A developed model based on nationwide electronic health data showed potential utility for risk stratification to supplement screening efforts. This work was supported through grants number PRG2218 from the Estonian Research Council, and EMP416 from the EEA (European Economic Area) and Norway Grants.

Common Critiques and Recommendations for Studies in Neurology Using Machine Learning Methods

Machine learning (ML) methods are becoming more prevalent in the neurology literature as alternatives to traditional statistical methods to address challenges in the analysis of modern data sets. Despite the increase in the popularity of ML methods in neurology studies, some authors do not fully address all items recommended in reporting guidelines. The authors of this Research Methods article are members of the Neurology® editorial board and have reviewed many studies using ML methods. In their review reports, several critiques often appear, which could be avoided if guidance were available. In this article, we detail common critiques found in ML research studies and make recommendations for how to avoid them. The first critique involves misalignment of the study goals and the analysis conducted. The second critique focuses on ML terminology being appropriately used. Critiques 3-6 are related to the study design: justifying sample sizes and the suitability of the data set for the study goals, describing the ML analysis pipeline sufficiently, quantifying the amount of missing data and providing information about missing data handling, and including uncertainty estimates for key metrics. The seventh critique focuses on fairly describing both strengths and limitations of the ML study, including the analysis methodology and results. We provide examples in neurology for each critique and guidance on how to avoid the critique. Overall, we recommend that authors use ML-specific checklists developed by research consortia for designing and reporting studies using ML. We also recommend that authors involve both a statistician and an ML expert in work that uses ML. Although our list of critiques is not exhaustive, our recommendations should help improve the quality and rigor of ML studies. ML has great potential to revolutionize neurology, but investigators need to conduct and report the results in a way that allows readers to fully evaluate the benefits and limitations of ML approaches.