Machine Learning for Hypertension Prediction: a Systematic Review

Machine learning-based analyses of contributing factors for the development of hypertension: a comparative study

OBJECTIVES

Sufficient attention has not been given to machine learning (ML) models using longitudinal data for investigating important predictors of new onset of hypertension. We investigated the predictive ability of several ML models for the development of hypertension.

METHODS

A total of 15 965 Japanese participants (men/women: 9,466/6,499, mean age: 45 years) who received annual health examinations were randomly divided into a training group (70%, n = 11,175) and a test group (30%, n = 4,790). The predictive abilities of 58 candidates including fatty liver index (FLI), which is calculated by using body mass index, waist circumference and levels of γ-glutamyl transferase and triglycerides, were investigated by statistics analogous to the area under the curve (AUC) in receiver operating characteristic curve analyses using ML models including logistic regression, random forest, naïve Bayes, extreme gradient boosting and artificial neural network.

RESULTS

During a 10-year period (mean period: 6.1 years), 2,132 subjects (19.1%) in the training group and 917 subjects (19.1%) in the test group had new onset of hypertension. Among the 58 parameters, systolic blood pressure, age and FLI were identified as important candidates by random forest feature selection with 10-fold cross-validation. The AUCs of ML models were 0.765-0.825, and discriminatory capacity was significantly improved in the artificial neural network model compared to that in the logistic regression model.

CONCLUSIONS

The development of hypertension can be simply and accurately predicted by each ML model using systolic blood pressure, age and FLI as selected features. By building multiple ML models, more practical prediction might be possible.

Machine learning prediction models for multidrug-resistant organism infections in ICU ventilator-associated pneumonia patients: Analysis using the MIMIC-IV database

OBJECTIVE

This study aims to construct and compare four machine learning models using the MIMIC-IV database to identify high-risk factors for multidrug-resistant organism (MDRO) infection in Ventilator-associated pneumonia (VAP) patients.

METHODS

The study included 972 VAP patients from the MIMIC-IV database. Data encompassing demographic information, vital signs, laboratory results, and other relevant variables were collected. The class imbalance issue was addressed using the Synthetic Minority Over-sampling Technique (SMOTE). The dataset was randomly split into training and testing sets (8:2). LASSO regression and feature importance scores were used for feature selection. Clinical prediction models were built using logistic regression, XGBoost, random forest and gradient boosting machine. The performance of the models was evaluated through receiver operating characteristic(ROC) curve analysis.Model calibration was assessed using calibration curves and Brier scores. The effectiveness was evaluated through Decision Curve Analysis (DCA). SHAP was utilized for model interpretation.

RESULTS

Among 972 patients, 824 were non-MDROs-VAP and 128 were MDROs-VAP. Comparative analysis revealed statistically significant differences in various clinical parameters. XGBoost exhibited the best predictive performance, incorporating 20 features with an AUC of 0.831 (95 % CI: 0.785-0.877) on the test set. Calibration curves demonstrated robust consistency, corroborated by Decision Curve Analysis (DCA) affirming the clinical utility. SHAP analysis identified the most important features: red cell distribution width, duration of mechanical ventilation, anion gap, basophil percentage, and neutrophil percentage.

CONCLUSION

This study established and compared four machine learning models for MDROs infections in VAP patients. XGBoost was identified as the optimal predictor, and SHAP values provided insights into 20 independent risk factors, confirming its excellent predictive value.

IMPLICATIONS FOR CLINICAL PRACTICE

VAP is a common infection in ICU patients with a heightened risk of MDRO and increased mortality. The recognition of high bias in existing models calls for future research to employ rigorous methodologies and robust data sources, aiming to develop and validate more accurate and clinically applicable predictive models for MDROs infections in VAP patients.

Proximity extension assay reveals serum inflammatory biomarkers in two amyotrophic lateral sclerosis cohorts

Amyotrophic lateral sclerosis (ALS) is a rare neurodegenerative disease with both clinical and hereditary heterogeneity. Inflammation has been suggested to play an important role in ALS pathophysiology. In this study, we aimed to identify serum inflammatory alterations and develop effective inflammatory biomarkers to assist in the diagnosis of ALS. Through proximity extension assay (PEA), we investigated serum inflammatory alterations in two ALS cohorts compared with healthy controls (HCs), including sporadic ALS patients and genetic ALS patients. We found that CHIT1, OSM, SIRT2, CDCP1 and 5 other factors were significantly increased in sporadic ALS patients in both cohorts and that SIRT2, CDCP1 and 6 other factors were different between genetic ALS patients and HCs. Using XGBoost and binary logistic regression analysis, we developed a two-serum protein diagnostic panel (CHIT1 and CDCP1), and the area under the curve (AUC) was 0.904 in the original cohort and 0.907 in the replication cohort. Based on Mendelian Randomization (MR), OSM and SIRT2 are significantly associated with the risk of ALS. In conclusion, our study revealed a consistent and replicable serum inflammatory profile and developed a biomarker panel that can differentiate ALS patients from HCs in two cohorts, which may play an important role in advancing our current understanding of the inflammatory process and identifying novel therapeutic strategies for ALS patients.

Interpretable prognostic modeling for long-term survival of Type A aortic dissection patients using support vector machine algorithm

OBJECTIVES

This study aims to develop a reliable and interpretable predictive model for long-term survival in Type A aortic dissection (TAAD) patients, utilizing machine learning (ML) algorithms.

METHODS

We retrospectively reviewed the clinical data of patients diagnosed with TAAD who underwent open surgical repair at the First Affiliated Hospital of Chongqing Medical University, from September 2017 to December 2020, and at the Chongqing University Central Hospital between October 2019 and April 2020. Cases with less than 20% missing data were imputed using random forest algorithms. To identify significant prognostic factors, we performed LASSO (Least Absolute Shrinkage and Selection Operator) Cox regression analysis, including preoperative blood markers, previous medical history and intraoperative condition. Based on the advantages of the model and the characteristics of the data set, we subsequently developed a machine learning-based prognostic model using Support Vector Machine (SVM) and evaluated its performance across key metrics. To further explain the decision-making process of the SVM model, we employed SHapley Additive exPlanation (SHAP) values for model interpretation.

RESULTS

A total of 171 patients with TAAD were included in model training and internal test groups; 73 patients with TAAD were included in external test group. Through LASSO Cox regression, univariate analysis, and clinical relevance assessment, seven feature variables were selected for modeling. Performance evaluation revealed that the SVM model showed excellent performance in both the training and test sets, with no significant overfitting, indicating strong clinical applicability. In the training set, the model achieved an AUC of 0.9137 (95% CI 0.9081-0.9203) and in the internal and external testing set, 0.8533 (95% CI 0.8503-0.8624) and 0.8770 (95% CI 0.8698-0.8982), respectively. The accuracy values were 0.8366, 0.8481 and 0.8030; precision values were 0.8696, 0.8374 and 0.8235; recall values were 0.8421, 0.7933 and 0.7651; F1 scores were 0.8290, 0.8148 and 0.7928; Brier scores were 0.1213, 0.1417 and 0.1323; average precision (AP) values were 0.9019, 0.8789 and 0.8548, respectively. SHAP analysis revealed that longer operation time, extended cardiopulmonary bypass (CPB) duration, prolonged aortic cross-clamp (ACC) time, advanced age, higher plasma transfusion volume, elevated serum creatinine and increased white blood cell (WBC) count significantly contributed to higher model predictions.

CONCLUSIONS

This study developed an interpretable predictive model based on the SVM algorithm to assess long-term survival in TAAD patients. The model demonstrated accuracy, precision, and robustness in identifying high-risk patients, providing reliable evidence for clinicians.

Thyro-GenAI: A Chatbot Using Retrieval-Augmented Generative Models for Personalized Thyroid Disease Management

Background: Large language models (LLMs) have the potential to enhance information processing and clinical reasoning in the healthcare industry but are hindered by inaccuracies and hallucinations. The retrieval-augmented generation (RAG) technique may address these problems by integrating external knowledge sources. Methods: We developed a RAG-based chatbot called Thyro-GenAI by integrating a database of textbooks and guidelines with LLM. Thyro-GenAI and three service LLMs: OpenAI's ChatGPT-4o, Perplexity AI's ChatGPT-4o, and Anthropic's Claude 3.5 Sonnet, were asked personalized clinical questions about thyroid disease. Three thyroid specialists assessed the quality of the generated responses and references without being blinded, which allowed them to interact with different chatbot interfaces. Results: Thyro-GenAI achieved the highest inverse-weighted mean rank for overall response quality. The overall inverse-weighted mean rankings for Thyro-GenAI, ChatGPT, Perplexity, and Claude were 3.0, 2.3, 2.8, and 1.9, respectively. Thyro-GenAI also achieved the second-highest inverse-weighted mean rank for overall reference quality. The overall inverse-weighted mean rankings for Thyro-GenAI, ChatGPT, Perplexity, and Claude were 3.1, 2.3, 3.2, and 1.8, respectively. Conclusions: Thyro-GenAI produced patient-specific clinical reasoning output based on a vector database, with fewer hallucinations and more reliability, compared to service LLMs. This emphasis on evidence-based responses ensures its safety and validity, addressing a critical limitation of existing LLMs. By integrating RAG with LLMs, it has the potential to support frontline clinical decision-making, especially helping first-line physicians by offering reliable decision support while managing thyroid disease patients.

Controversy in Hypertension: Pro-Side of the Argument Using Artificial Intelligence for Hypertension Diagnosis and Management

Hypertension presents the largest modifiable public health challenge due to its high prevalence, its intimate relationship to cardiovascular diseases, and its complex pathogenesis and pathophysiology. Low awareness of blood pressure elevation and suboptimal hypertension diagnosis serve as the major hurdles in effective hypertension management. Advances in artificial intelligence in hypertension have permitted the integrative analysis of large data sets including omics, clinical (with novel sensor and wearable technologies), health-related, social, behavioral, and environmental sources, and hold transformative potential in achieving large-scale, data-driven approaches toward personalized diagnosis, treatment, and long-term management. However, although the emerging artificial intelligence science may advance the concept of precision hypertension in discovery, drug targeting and development, patient care, and management, its clinical adoption at scale today is lacking. Recognizing that clinical implementation of artificial intelligence-based solutions need evidence generation, this opinion statement examines a clinician-centric perspective of the state-of-art in using artificial intelligence in the management of hypertension and puts forward recommendations toward equitable precision hypertension care.

Machine Learning-Based Pathomics Model Predicts Angiopoietin-2 Expression and Prognosis in Hepatocellular Carcinoma

Angiopoietin-2 (ANGPT2) shows promise as prognostic marker and therapeutic target in hepatocellular carcinoma (HCC). However, assessing ANGPT2 expression and prognostic potential using histopathology images viewed with naked eye is challenging. Herein, machine learning was employed to develop a pathomics model for analyzing histopathology images to predict ANGPT2 status. HCC cases obtained from The Cancer Genome Atlas (TCGA-HCC; n = 267) were randomly assigned to the training or testing set, and cases from a single center were employed as a validation set (n = 91). In the TCGA-HCC cohort, the group with high ANGPT2 expression had a significantly lower overall survival compared with the group with low ANGPT2. Histopathologic features in the training set were extracted, screened, and incorporated into a gradient-boosting machine model that generated a pathomics score, which successfully predicted ANGPT2 expression in the three data sets and showed remarkable risk stratification for overall survival in both the TCGA-HCC (P < 0.0001) and single-center cohorts (P = 0.001). Multivariate analysis suggested that the pathomics score could serve as a predictor of prognosis (P < 0.001). Bioinformatics analysis illustrated a distinction in tumor growth and development related gene-enriched pathways, vascular endothelial growth factor-related gene expression, and immune cell infiltration between high and low pathomics scores. This study indicates that the use of histopathology image features can enhance the prediction of molecular status and prognosis in HCC. The integration of image features with machine learning may improve prognosis prediction in HCC.

Development and Validation of a Hypertension Risk Prediction Model Based on Particle Swarm Optimization-Support Vector Machine

BACKGROUND

Hypertension is a prevalent health issue, especially among the elderly, and is linked to multiple complications. Early and accurate detection is crucial for effective management. Traditional detection methods may be limited in accuracy and efficiency, prompting the exploration of advanced computational techniques. Machine learning algorithms, combined with optimization methods, show potential in enhancing hypertension detection.

METHODS

In 2022, data from 1460 hypertensive and 1416 non-hypertensive individuals aged 65 and above were collected from the Lujingdong Outpatient Department of the Guangdong Second Traditional Chinese Medicine Hospital. Support Vector Machine (SVM) and Particle Swarm Optimization-Support Vector Machine (PSO-SVM) models were developed, validated using the holdout method, and evaluated based on sensitivity, specificity, positive predictive value (PPV), accuracy, G-mean, F1 score, Matthews correlation coefficient (MCC), and the area under the curve (AUC) of the receiver operating characteristic curve (ROC curve).

RESULTS

The PSO-SVM model outperformed the standard SVM, especially in sensitivity (93.9%), F1 score (0.838), and AUC-ROC (0.871).

CONCLUSION

The PSO-SVM model is effective for complex classifications, particularly in hypertension detection, providing a basis for early diagnosis and treatment.

Explainable machine learning model for predicting decline in platelet count after interventional closure in children with patent ductus arteriosus

Background

This study aimed to apply four machine learning algorithms to develop the optimal model to predict decline in platelet count (DPC) after interventional closure in children with patent ductus arteriosus (PDA).

Methods

Data from children with PDA who underwent successful transcatheter closure at the Second Affiliated Hospital of Wenzhou Medical University and Yuying Children's Hospital from January 2016, to December 2022, were collected. The cohort data were split into training and testing sets. DPC following the intervention is defined as a percentage DPC ≥25% [(baseline platelet count-nadir platelet count)/baseline platelet count]. The extra tree algorithm was used for feature selection and four ML algorithms [random forest (RF), adaptive boosting, extreme gradient boosting, and logistic regression] were established. Moreover, SHapley Additive exPlanation (SHAP) to explain the importance of features and the ML models.

Results

This study included 330 children who underwent successful transcatheter closure of PDA, of which 113 (34.2%) experienced DPC. After 62 clinical features were considered, the extra tree algorithm selected six clinical features to build the ML models. Amongst the four ML algorithms, the RF model achieved the greatest AUC. SHAP analysis revealed that pulmonary artery systolic pressure, size of defect and weight were the top three most important clinical features in the RF model. Furthermore, clinical descriptions of two children with PDA, with accurate predictions, and explanations of the prediction results were provided.

Conclusion

In this study, an ML model (RF) capable of predicting post-intervention DPC in children with PDA undergoing transcatheter closure was established.

Relative Fat Mass and Physical Indices as Predictors of Gallstone Formation: Insights From Machine Learning and Logistic Regression

Purpose

Gallstones (GS), a prevalent disorder of the biliary tract, markedly impair patients' quality of life. This study aims to construct predictive models employing diverse machine learning algorithms to elucidate risk factors linked to gallstone formation.

Patients and Methods

This study integrated data from the National Health and Nutrition Examination Survey (NHANES) with a cohort of 7868 participants from Wuxi People's Hospital and Wuxi Second People's Hospital, including 830 individuals diagnosed with gallstones. To develop our predictive model, we employed four algorithms-Logistic Regression, Gaussian Naive Bayes (GNB), Multi-Layer Perceptron (MLP), and Support Vector Machine (SVM). The models were validated internally through k-fold cross-validation and externally using independent datasets. Furthermore, we substantiated the link between relative fat mass (RFM) and gallstone formation by employing four logistic regression models, conducting subgroup analyses, and applying restricted cubic spline (RCS) curves.

Results

The logistic regression algorithm demonstrated superior predictive capability for all risk factors associated with gallstone occurrence compared to other machine learning models. SHAP analysis identified RFM, weight-to-waist index (WWI), waist circumference (WC), waist-to-height ratio (WHtR), and body mass index (BMI) as prominent predictors of gallstone occurrence, with RFM emerging as the primary determinant. A fully adjusted multivariate logistic regression analysis revealed a robust positive association between RFM and gallstones. Subgroup analysis further indicated that subgroup factors did not alter the positive relationship between RFM and gallstone prevalence.

Conclusion

Among the four algorithmic models, logistic regression proved most effective in predicting gallstone occurrence. The model developed in this study offers clinicians a valuable tool for identifying critical prognostic factors, facilitating personalized patient monitoring and tailored management.

Feature Selection for Hypertension Risk Prediction Using XGBoost on Single Nucleotide Polymorphism Data

OBJECTIVES

Hypertension, commonly known as high blood pressure, is a prevalent and serious condition affecting a significant portion of the adult population globally. It is a chronic medical issue that, if left unaddressed, can lead to severe health complications, including kidney problems, heart disease, and stroke. This study aims to develop a feature selection model using the XGBoost algorithm to identify specific single nucleotide polymorphisms (SNPs) as biomarkers for detecting hypertension risk.

METHODS

We propose using the high dimensionality of genetic variations (i.e., SNPs) to build a classifier model for prediction. In this study, SNPs were used as markers for hypertension in patients. We utilized the OpenSNP dataset, which includes 19,697 SNPs from 2,052 samples. Extreme gradient boosting (XGBoost) is an ensemble machine learning method employed here for feature selection, which incrementally adjusts weights in a series of steps.

RESULTS

The experimental results identified 292 SNPs that exhibited high performance, with an F1-score of 98.55%, precision of 98.73%, recall of 98.38%, and overall accuracy of 98%. This study provides compelling evidence that the XGBoost feature selection method outperforms other representative feature selection methods, such as genetic algorithms, analysis of variance, chi-square, and principal component analysis, in predicting hypertension risk, demonstrating its effectiveness.

CONCLUSIONS

We developed a model for predicting hypertension using the SNPs dataset. The high dimensionality of SNP data was effectively managed to identify significant features as biomarkers using the XGBoost feature selection method. The results indicate high performance in predicting the risk of hypertension.

Machine learning-enabled risk prediction of self-neglect among community-dwelling older adults in China

BACKGROUND

Elder self-neglect (ESN) is usually ignored as a private problem and impairs the health outcomes of older adults. It is essential to construct a robust and efficient tool for risk prediction which can better detect and prevent self-neglect among older adults.

METHODS

This study included 2494 study participants from the Ma'anshan Healthy Ageing Cohort (MHAC). First, the group-based trajectory model (GBTM) was used to estimate ESN development trajectory groups. Then, feature selection methods were used to select variables; after that, we compared six machine learning models (Decision Tree Classifier (DT), K-Nearest Neighbour (KNN), Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM) and XGBoost (XGB)). In addition, Synthetic Minority Oversampling Technique (SMOTE) is used to address the data imbalance problem.

RESULTS

The results show that the ESN can be defined as two trajectory groups (rising and stable). After feature selection, the final model contains eight predictors. The area under the curve (AUC) of the raw dataset was 0.637-0.769. In the dataset with SMOTE, the AUC was 0.635-0.765 and RF was the optimal model. The top five most important characteristics were quality of life, psychological resilience, social support, education, and income.

CONCLUSIONS

The RF developed in this study may be considered a simple and scientific aid in the risk prediction of self-neglect among community-dwelling old adults.

Machine learning algorithms in constructing prediction models for assisted reproductive technology (ART) related live birth outcomes

Currently applicable models for predicting live birth outcomes in patients who received assisted reproductive technology (ART) have methodological or study design limitations that greatly obstruct their dissemination and application. Models suitable for Chinese couples have not yet been identified. We conducted a retrospective study by using a database includes a total of 11,938 couples who underwent in vitro fertilization (IVF) treatment between January 2015 and December 2022 in a medical institution of southwest China Yunnan province. Multiple candidate predictors were screened out by using the importance scores. Four machine learning (ML) algorithms including random forest, extreme gradient boosting, light gradient boosting machine and binary logistic regression were used to construct prediction models. An initial assessment of the predictive performance was conducted and validated by using cross-validation and bootstrap methods. A total of seven predictors were identified, namely maternal age, duration of infertility, basal follicle-stimulating hormone (FSH), progressive sperm motility, progesterone (P) on HCG day, estradiol (E2) on HCG day, and luteinizing hormone (LH) on HCG day. Of the four predictive models, the random forest model and the logistic regression model were considered to have the optimal performance, with the areas under the receiver operating characteristic curve (AUROC) curves of 0.671 (95% CI 0.630-0.713) and 0.674 (95% CI 0.627-0.720). The Brier scores were 0.183 (95% CI 0.170-0.196) and 0.183 (95% CI 0.170-0.196), respectively. Considering the simplicity of model fitting, we recommend the logistic regression model as the best predictive model for live birth. Furthermore, maternal age, P on HCG day and E2 on HCG day were deemed to have the highest contribution to model prediction.

Antihypertensive Drug Recommendations for Reducing Arterial Stiffness in Patients With Hypertension: Machine Learning-Based Multicohort (RIGIPREV) Study

BACKGROUND

High systolic blood pressure is one of the leading global risk factors for mortality, contributing significantly to cardiovascular diseases. Despite advances in treatment, a large proportion of patients with hypertension do not achieve optimal blood pressure control. Arterial stiffness (AS), measured by pulse wave velocity (PWV), is an independent predictor of cardiovascular events and overall mortality. Various antihypertensive drugs exhibit differential effects on PWV, but the extent to which these effects vary depending on individual patient characteristics is not well understood. Given the complexity of selecting the most appropriate antihypertensive medication for reducing PWV, machine learning (ML) techniques offer an opportunity to improve personalized treatment recommendations.

OBJECTIVE

This study aims to develop an ML model that provides personalized recommendations for antihypertensive medications aimed at reducing PWV. The model considers individual patient characteristics, such as demographic factors, clinical data, and cardiovascular measurements, to identify the most suitable antihypertensive agent for improving AS.

METHODS

This study, known as the RIGIPREV study, used data from the EVA, LOD-DIABETES, and EVIDENT studies involving individuals with hypertension with baseline and follow-up measurements. Antihypertensive drugs were grouped into classes such as angiotensin-converting enzyme inhibitors (ACEIs), angiotensin receptor blockers (ARBs), β-blockers, diuretics, and combinations of diuretics with ACEIs or ARBs. The primary outcomes were carotid-femoral and brachial-ankle PWV, while the secondary outcomes included various cardiovascular, anthropometric, and biochemical parameters. A multioutput regressor using 6 random forest models was used to predict the impact of each antihypertensive class on PWV reduction. Model performance was evaluated using the coefficient of determination (R2) and mean squared error.

RESULTS

The random forest models exhibited strong predictive capabilities, with internal validation yielding R2 values between 0.61 and 0.74, while external validation showed a range of 0.26 to 0.46. The mean squared values ranged from 0.08 to 0.22 for internal validation and from 0.29 to 0.45 for external validation. Variable importance analysis revealed that glycated hemoglobin and weight were the most critical predictors for ACEIs, while carotid-femoral PWV and total cholesterol were key variables for ARBs. The decision tree model achieved an accuracy of 84.02% in identifying the most suitable antihypertensive drug based on individual patient characteristics. Furthermore, the system's recommendations for ARBs matched 55.3% of patients' original prescriptions.

CONCLUSIONS

This study demonstrates the utility of ML techniques in providing personalized treatment recommendations for antihypertensive therapy. By accounting for individual patient characteristics, the model improves the selection of drugs that control blood pressure and reduce AS. These findings could significantly aid clinicians in optimizing hypertension management and reducing cardiovascular risk. However, further studies with larger and more diverse populations are necessary to validate these results and extend the model's applicability.

Explore key genes of Crohn's disease based on glycerophospholipid metabolism: A comprehensive analysis Utilizing Mendelian Randomization, Multi-Omics integration, Machine Learning, and SHAP methodology

BACKGROUND AND AIMS

Crohn's disease (CD) is a chronic, complex inflammatory condition with increasing incidence and prevalence worldwide. However, the causes of CD remain incompletely understood. We identified CD-related metabolites, inflammatory factors, and key genes by Mendelian randomization (MR), multi-omics integration, machine learning (ML), and SHAP.

METHODS

We first performed a mediation MR analysis on 1400 serum metabolites, 91 inflammatory factors, and CD. We found that certain phospholipids are causally related to CD. In the scRNA-seq data, monocytes were categorized into high and low metabolism groups based on their glycerophospholipid metabolism scores. The differentially expressed genes of these two groups of cells were extracted, and transcription factor prediction, cell communication analysis, and GSEA analysis were performed. After further screening of differentially expressed genes (FDR<0.05, log2FC>1), least absolute shrinkage and selection operator (LASSO) regression was performed to obtain hub genes. Models for hub genes were built using the Catboost, XGboost, and NGboost methods. Further, we used the SHAP method to interpret the models and obtain the gene with the highest contribution to each model. Finally, qRT-PCR was used to verify the expression of these genes in the peripheral blood mononuclear cells (PBMC) of CD patients and healthy subjects.

RESULT

MR results showed 1-palmitoyl-2-stearoyl-gpc (16:0/18:0) levels, 1-stearoyl-2-arachidonoyl-GPI (18:0/20:4) levels, 1-arachidonoyl-gpc (20:4n6) levels, 1-palmitoyl-2-arachidonoyl-gpc (16:0/20:4n6) levels, and 1-arachidonoyl-GPE (20:4n6) levels were significantly associated with CD risk reduction (FDR<0.05), with CXCL9 acting as a mediation between these phospholipids and CD. The analysis identified 19 hub genes, with Catboost, XGboost, and NGboost achieving AUC of 0.91, 0.88, and 0.85, respectively. The SHAP methodology obtained the three genes with the highest model contribution: G0S2, S100A8, and PLAUR. The qRT-PCR results showed that the expression levels of S100A8 (p = 0.0003), G0S2 (p < 0.0001), and PLAUR (p = 0.0141) in the PBMC of CD patients were higher than healthy subjects.

CONCLUSION

MR findings suggest that certain phospholipids may lower CD risk. G0S2, S100A8, and PLAUR may be potential pathogenic genes in CD. These phospholipids and genes could serve as novel diagnostic and therapeutic targets for CD.

Clinical phenotype of ARDS based on K-means cluster analysis: A study from the eICU database

Purpose

To explore the characteristics of the clinical phenotype of ARDS based on Machine Learning.

Methods

This is a study on Machine Learning. Screened cases of acute respiratory distress syndrome (ARDS) in the eICU database collected basic information in the cases and clinical data on the Day 1, Day 3, and Day 7 after the diagnosis of ARDS, respectively. Using the Calinski-Harabasz criterion, Gap Statistic, and Silhouette Coefficient, we determine the optimal clustering number k value. By the K-means cluster analysis to derive clinical phenotype, we analyzed the data collected within the first 24 h. We compared it with the survival of cases under the Berlin standard classification, and also examined the phenotypic conversion within the first 24 h, on day 3, and on day 7 after the diagnosis of ARDS.

Results

We collected 5054 cases and derived three clinical phenotypes using K-means cluster analysis. Phenotype-I is characterized by fewer abnormal laboratory indicators, higher oxygen partial pressure, oxygenation index, APACHE IV score, systolic and diastolic blood pressure, and lower respiratory rate and heart rate. Phenotype-II is characterized by elevated white blood cell count, blood glucose, creatinine, temperature, heart rate, and respiratory rate. Phenotype-III is characterized by elevated age, partial pressure of carbon dioxide, bicarbonate, GCS score, albumin. The differences in ICU length of stay and in-hospital mortality were significantly different between the three phenotypes (P < 0.05), with phenotype I having the lowest in-hospital mortality (10 %) and phenotype II having the highest (31.8 %). To compare the survival analysis of ARDS patients classified by phenotype and those classified according to Berlin criteria. The results showed that the differences in survival between phenotypes were statistically significant (P < 0.05) under phenotypic classification.

Conclusions

The clinical classification of ARDS based on K-means clustering analysis is beneficial for further identifying ARDS patients with different characteristics. Compared to the Berlin standard, the new clinical classification of ARDS provides a clearer display of the survival status of different types of patients, which helps to predict patient prognosis.

Assessing personalized molecular portraits underlying endothelial-to-mesenchymal transition within pulmonary arterial hypertension

Pulmonary arterial hypertension (PAH) is a progressive and rapidly fatal disease with an intricate etiology. Identifying biomarkers for early PAH lesions based on the exploration of subtle biological processes is significant for timely diagnosis and treatment. In the present study, nine distinct cell populations identified based on gene expression profiles revealed high heterogeneity in cell composition ratio, biological function, distribution preference, and communication patterns in PAH. Notably, compared to other cells, endothelial cells (ECs) showed prominent variation in multiple perspectives. Further analysis demonstrated the endothelial-to-mesenchymal transition (EndMT) in ECs and identified a subgroup exhibiting a contrasting phenotype. Based on these findings, a machine-learning integrated program consisting of nine learners was developed to create a PAH Endothelial-to-mesenchymal transition Signature (PETS). This study identified cell populations underlying EndMT and furnished a potential tool that might be valuable for PAH diagnosis and new precise therapies.

Machine learning-based prognostic prediction for hospitalized HIV/AIDS patients with cryptococcus infection in Guangxi, China

OBJECTIVE

To develop and validate a machine learning model for predicting mortality-associated prognostic factors in order to reduce in-hospital mortality rates among HIV/AIDS patients with Cryptococcus infection in Guangxi, China.

METHODS

This retrospective prognostic study included HIV/AIDS patients with cryptococcosis in the Fourth People's Hospital of Nanning from October 2011 to June 2019. Clinical features were extracted and used to train ten machine learning models, including Logistic Regression, KNN, DT, RF, Adaboost, Xgboost, LightGBM, Catboost, SVM, and NBM, to predict the outcome of HIV patients with cryptococcosis infection. The sensitivity, specificity, AUC, and F1 value were applied to assess model performance in both the testing and training sets. The optimal model was selected and interpreted.

RESULTS

A total of 396 patients were included in the study. The average in-hospital mortality of HIV/AIDS patients with cryptococcosis was 12.9% from 2012 to 2019. After feature screening, 20 clinical features were selected for model construction, accounting for 93.8%, including ART, Electrolyte disorder, Anemia, and 17 laboratory tests. The RF model (AUC 0.9787, Sensitivity 0.9535, Specificity 0.8889, F1 0.7455) and the SVM model (AUC 0.9286, Sensitivity 0.7907, Specificity 0.9786, F1 0.8293) had excellent performance. The SHAP analysis showed that the primary risk factors for prognosis prediction were identified as BUN/CREA, Electrolyte disorder, NEUT%, Urea, and IBIL.

CONCLUSIONS

RF and SVM machine learning models have shown promising predictive abilities for the prognosis of hospitalized HIV/AIDS patients with cryptococcosis, which can aid clinical assessment and treatment decisions for patient prognosis.

Smart solutions in hypertension diagnosis and management: a deep dive into artificial intelligence and modern wearables for blood pressure monitoring

Hypertension, a widespread cardiovascular issue, presents a major global health challenge. Traditional diagnosis and treatment methods involve periodic blood pressure monitoring and prescribing antihypertensive drugs. Smart technology integration in healthcare offers promising results in optimizing the diagnosis and treatment of various conditions. We investigate its role in improving hypertension diagnosis and treatment effectiveness using machine learning algorithms for early and accurate detection. Intelligent models trained on diverse datasets (encompassing physiological parameters, lifestyle factors, and genetic information) to detect subtle hypertension risk patterns. Adaptive algorithms analyze patient-specific data, optimizing treatment plans based on medication responses and lifestyle habits. This personalized approach ensures effective, minimally invasive interventions tailored to each patient. Wearables and smart sensors provide real-time health insights for proactive treatment adjustments and early complication detection.

Machine learning-based prediction of the risk of moderate-to-severe catheter-related bladder discomfort in general anaesthesia patients: a prospective cohort study

BACKGROUND

Catheter-related bladder discomfort (CRBD) commonly occurs in patients who have indwelling urinary catheters while under general anesthesia. And moderate-to-severe CRBD can lead to significant adverse events and negatively impact patient health outcomes. However, current screening studies for patients experiencing moderate-to-severe CRBD after waking from general anesthesia are insufficient. Constructing predictive models with higher accuracy using multiple machine learning techniques for early identification of patients at risk of experiencing moderate-to-severe CRBD during general anesthesia resuscitation.

METHODS

Eight hundred forty-six patients with indwelling urinary catheters who were resuscitated in a post-anesthesia care unit (PACU). Trained researchers used the CRBD 4-level assessment method to evaluate the severity of a patient's CRBD. They then inputted 24 predictors into six different machine learning algorithms. The performance of the models was evaluated using metrics like the area under the curve (AUC).

RESULTS

The AUCs of the six models ranged from 0.82 to 0.89. Among them, the RF model displayed the highest predictive ability, with an AUC of 0.89 (95%CI: 0.87, 0.91). Additionally, it achieved an accuracy of 0.93 (95%CI: 0.91, 0.95), 0.80 sensitivity, 0.98 specificity, 0.94 positive predictive value (PPV), 0.92 negative predictive value (NPV), 0.87 F1 score, and 0.07 Brier score. The logistic regression (LR) model has achieved good results (AUC:0.87) and converted into a nomogram.

CONCLUSIONS

The study has successfully developed a machine learning prediction model that exhibits excellent predictive capabilities in identifying patients who may develop moderate-to-severe CRBD after undergoing general anesthesia. Furthermore, the study also presents a nomogram, which serves as a valuable tool for clinical healthcare professionals, enabling them to intervene at an early stage for better patient outcomes.

The application and clinical translation of the self-evolving machine learning methods in predicting diabetic retinopathy and visualizing clinical transformation

Objective

This study aims to analyze the application and clinical translation value of the self-evolving machine learning methods in predicting diabetic retinopathy and visualizing clinical outcomes.

Methods

A retrospective study was conducted on 300 diabetic patients admitted to our hospital between January 2022 and October 2023. The patients were divided into a diabetic retinopathy group (n=150) and a non-diabetic retinopathy group (n=150). The improved Beetle Antennae Search (IBAS) was used for hyperparameter optimization in machine learning, and a self-evolving machine learning model based on XGBoost was developed. Value analysis was performed on the predictive features for diabetic retinopathy selected through multifactor logistic regression analysis, followed by the construction of a visualization system to calculate the risk of diabetic retinopathy occurrence.

Results

Multifactor logistic regression analysis revealed that being male, having a longer disease duration, higher systolic blood pressure, fasting blood glucose, glycosylated hemoglobin, low-density lipoprotein cholesterol, and urine albumin-to-creatinine ratio were risk factors for the development of diabetic retinopathy, while non-pharmacological treatment was a protective factor. The self-evolving machine learning model demonstrated significant performance advantages in early diagnosis and prediction of diabetic retinopathy occurrence.

Conclusion

The application of the self-evolving machine learning models can assist in identifying features associated with diabetic retinopathy in clinical settings, enabling early prediction of disease occurrence and aiding in the formulation of treatment plans to improve patient prognosis.

Gene therapy for polygenic or complex diseases

Gene therapy utilizes nucleic acid drugs to treat diseases, encompassing gene supplementation, gene replacement, gene silencing, and gene editing. It represents a distinct therapeutic approach from traditional medications and introduces novel strategies for genetic disorders. Over the past two decades, significant advancements have been made in the field of gene therapy, leading to the approval of various gene therapy drugs. Gene therapy was initially employed for treating genetic diseases and cancers, particularly monogenic conditions classified as orphan diseases due to their low prevalence rates; however, polygenic or complex diseases exhibit higher incidence rates within populations. Extensive research on the etiology of polygenic diseases has unveiled new therapeutic targets that offer fresh opportunities for their treatment. Building upon the progress achieved in gene therapy for monogenic diseases and cancers, extending its application to polygenic or complex diseases would enable targeting a broader range of patient populations. This review aims to discuss the strategies of gene therapy, methods of gene editing (mainly CRISPR-CAS9), and carriers utilized in gene therapy, and highlight the applications of gene therapy in polygenic or complex diseases focused on applications that have either entered clinical stages or are currently undergoing clinical trials.

Improving cardiovascular risk prediction with machine learning: a focus on perivascular adipose tissue characteristics

BACKGROUND

Timely prevention of major adverse cardiovascular events (MACEs) is imperative for reducing cardiovascular diseases-related mortality. Perivascular adipose tissue (PVAT), the adipose tissue surrounding coronary arteries, has attracted increased amounts of attention. Developing a model for predicting the incidence of MACE utilizing machine learning (ML) integrating clinical and PVAT features may facilitate targeted preventive interventions and improve patient outcomes.

METHODS

From January 2017 to December 2019, we analyzed a cohort of 1077 individuals who underwent coronary CT scanning at our facility. Clinical features were collected alongside imaging features, such as coronary artery calcium (CAC) scores and perivascular adipose tissue (PVAT) characteristics. Logistic regression (LR), Framingham Risk Score, and ML algorithms were employed for MACE prediction.

RESULTS

We screened seven critical features to improve the practicability of the model. MACE patients tended to be older, smokers, and hypertensive. Imaging biomarkers such as CAC scores and PVAT characteristics differed significantly between patients with and without a 3-year MACE risk in a population that did not exhibit disparities in laboratory results. The ensemble model, which leverages multiple ML algorithms, demonstrated superior predictive performance compared with the other models. Finally, the ensemble model was used for risk stratification prediction to explore its clinical application value.

CONCLUSIONS

The developed ensemble model effectively predicted MACE incidence based on clinical and imaging features, highlighting the potential of ML algorithms in cardiovascular risk prediction and personalized medicine. Early identification of high-risk patients may facilitate targeted preventive interventions and improve patient outcomes.

Predicting new cases of hypertension in Swedish primary care with a machine learning tool

Background

Many individuals with hypertension remain undiagnosed. We aimed to develop a predictive model for hypertension using diagnostic codes from prevailing electronic medical records in Swedish primary care.

Methods

This sex- and age-matched case-control (1:5) study included patients aged 30-65 years living in the Stockholm Region, Sweden, with a newly recorded diagnosis of hypertension during 2010-19 (cases) and individuals without a recorded hypertension diagnosis during 2010-19 (controls), in total 507,618 individuals. Patients with diagnoses of cardiovascular diseases or diabetes were excluded. A stochastic gradient boosting machine learning model was constructed using the 1,309 most registered ICD-10 codes from primary care for three years prior the hypertension diagnosis.

Results

The model showed an area under the curve (95 % confidence interval) of 0.748 (0.742-0.753) for females and 0.745 (0.740-0.751) for males for predicting diagnosis of hypertension within three years. The sensitivity was 63 % and 68 %, and the specificity 76 % and 73 %, for females and males, respectively. The 25 diagnoses that contributed the most to the model for females and males all exhibited a normalized relative influence >1 %. The codes contributing most to the model, all with an odds ratio of marginal effects >1 for both sexes, were dyslipidaemia, obesity, and encountering health services in other circumstances.

Conclusions

This machine learning model, using prevailing recorded diagnoses within primary health care, may contribute to the identification of patients at risk of unrecognized hypertension. The added value of this predictive model beyond information of blood pressure warrants further study.

Transforming Healthcare: The AI Revolution in the Comprehensive Care of Hypertension

This review explores the transformative role of artificial intelligence (AI) in hypertension care, summarizing and analyzing published works from the last three years in this field. Hypertension contributes to a significant healthcare burden both at an individual and global level. We focus on five key areas: risk prediction, diagnosis, education, monitoring, and management of hypertension, supplemented with a brief look into the works on hypertensive disease of pregnancy. For each area, we discuss the advantages and disadvantages of integrating AI. While AI, in its current rudimentary form, cannot replace sound clinical judgment, it can still enhance faster diagnosis, education, prevention, and management. The integration of AI in healthcare is poised to revolutionize hypertension care, although careful implementation and ongoing research are essential to mitigate risks.

Using machine learning or deep learning models in a hospital setting to detect inappropriate prescriptions: a systematic review

OBJECTIVES

The emergence of artificial intelligence (AI) is catching the interest of hospital pharmacists. A massive collection of health data is now available to train AI models and hold the promise of disrupting codes and practices. The objective of this systematic review was to examine the state of the art of machine learning or deep learning models that detect inappropriate hospital medication orders.

METHODS

A systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. MEDLINE and Embase databases were searched from inception to May 2023. Studies were included if they reported and described an AI model intended for use by clinical pharmacists in hospitals. Risk of bias was assessed using the Prediction model Risk Of Bias ASsessment Tool (PROBAST).

RESULTS

13 articles were selected after review: 12 studies were judged to have high risk of bias; 11 studies were published between 2020 and 2023; 8 were conducted in North America and Asia; 6 analysed orders and detected inappropriate prescriptions according to patient profiles and medication orders; and 7 detected specific inappropriate prescriptions, such as detecting antibiotic resistance, dosage abnormality in prescriptions, high alert drugs errors from prescriptions or predicting the risk of adverse drug events. Various AI models were used, mainly supervised learning techniques. The training datasets used were very heterogeneous; the length of study varied from 2 weeks to 7 years and the number of prescription orders analysed went from 31 to 5 804 192.

CONCLUSIONS

This systematic review points out that, to date, few original research studies report AI tools based on machine or deep learning in the field of hospital clinical pharmacy. However, these original articles, while preliminary, highlighted the potential value of integrating AI into clinical hospital pharmacy practice.

Computed tomography-based prediction of pancreatitis following biliary metal stent placement with the convolutional neural network

Background and study aims Pancreatitis is a potentially lethal adverse event of endoscopic transpapillary placement of a self-expandable metal stent (SEMS) for malignant biliary obstruction (MBO). Deep learning-based image recognition has not been investigated in predicting pancreatitis in this setting. Patients and methods We included 70 patients who underwent endoscopic placement of a SEMS for nonresectable distal MBO. We constructed a convolutional neural network (CNN) model for pancreatitis prediction using a series of pre-procedure computed tomography images covering the whole pancreas (≥ 120,960 augmented images in total). We examined the additional effects of the CNN-based probabilities on the following machine learning models based on clinical parameters: logistic regression, support vector machine with a linear or RBF kernel, random forest classifier, and gradient boosting classifier. Model performance was assessed based on the area under the curve (AUC) in the receiver operating characteristic analysis, positive predictive value (PPV), accuracy, and specificity. Results The CNN model was associated with moderate levels of performance metrics: AUC, 0.67; PPV, 0.45; accuracy, 0.66; and specificity, 0.63. When added to the machine learning models, the CNN-based probabilities increased the performance metrics. The logistic regression model with the CNN-based probabilities had an AUC of 0.74, PPV of 0.85, accuracy of 0.83, and specificity of 0.96, compared with 0.72, 0.78, 0.77, and 0.96, respectively, without the probabilities. Conclusions The CNN-based model may increase predictability for pancreatitis following endoscopic placement of a biliary SEMS. Our findings support the potential of deep learning technology to improve prognostic models in pancreatobiliary therapeutic endoscopy.

A Systematic Review of Natural Language Processing Methods and Applications in Thyroidology

This study aimed to review the application of natural language processing (NLP) in thyroid-related conditions and to summarize current challenges and potential future directions. We performed a systematic search of databases for studies describing NLP applications in thyroid conditions published in English between January 1, 2012 and November 4, 2022. In addition, we used a snowballing technique to identify studies missed in the initial search or published after our search timeline until April 1, 2023. For included studies, we extracted the NLP method (eg, rule-based, machine learning, deep learning, or hybrid), NLP application (eg, identification, classification, and automation), thyroid condition (eg, thyroid cancer, thyroid nodule, and functional or autoimmune disease), data source (eg, electronic health records, health forums, medical literature databases, or genomic databases), performance metrics, and stages of development. We identified 24 eligible NLP studies focusing on thyroid-related conditions. Deep learning-based methods were the most common (38%), followed by rule-based (21%), and traditional machine learning (21%) methods. Thyroid nodules (54%) and thyroid cancer (29%) were the primary conditions under investigation. Electronic health records were the dominant data source (17/24, 71%), with imaging reports being the most frequently used (15/17, 88%). There is increasing interest in NLP applications for thyroid-related studies, mostly addressing thyroid nodules and using deep learning-based methodologies with limited external validation. However, none of the reviewed NLP applications have reached clinical practice. Several limitations, including inconsistent clinical documentation and model portability, need to be addressed to promote the evaluation and implementation of NLP applications to support patient care in thyroidology.

Development and trends in research on hypertension and atrial fibrillation: A bibliometric analysis from 2003 to 2022

BACKGROUND

This study aimed to comprehensively analyze research related to hypertension and atrial fibrillation, 2 common cardiovascular diseases with significant global public health implications, using bibliometric methods from 2003 to 2022.

METHODS

From the Web of Science Core Collection database, literature on the theme of hypertension and atrial fibrillation was retrieved. Subsequently, comprehensive bibliometric analyses were conducted across multiple dimensions utilizing software tools such as VOSviewer, Citespace, Pajek, Scimago Graphica, and ClusterProfiler. These analyses encompassed examinations of the literature according to country/region, institution, authors, journals, citation relationships, and keywords.

RESULTS

It revealed an increasing interest and shifting focus in research over the years. The analysis covered 7936 relevant publications, demonstrating a gradual rise in research activity regarding hypertension combined with atrial fibrillation over the past 2 decades, with a stable growth trend in research outcomes. Geographically, Europe and the Americas, particularly the United States, have shown the most active research in this field, while China has also gained importance in recent years. Regarding institutional contributions, internationally renowned institutions such as the University of Birmingham and the Mayo Clinic have emerged as core forces in this research direction. Additionally, Professor Lip Gregory, with his prolific research output, has stood out among numerous scholars. The American Journal of Cardiology has become a primary platform for publishing research related to hypertension and atrial fibrillation, highlighting its central role in advancing knowledge dissemination in this field. The research focus has shifted from exploring the pathophysiological mechanisms to investigating the treatment of complications and risk factors associated with hypertension and atrial fibrillation. Future research will focus on in-depth exploration of genetic and molecular mechanisms, causal relationship exploration through Mendelian randomization studies, and the application of machine learning techniques in prediction and treatment, aiming to promote the development of precision medicine for cardiovascular diseases.

CONCLUSION

In conclusion, this study provides a comprehensive overview of the developmental trajectory of research on hypertension and atrial fibrillation, presenting novel insights into trends and future research directions, thus offering information support and guidance for research in this crucial field of cardiovascular medicine.

Prediction of tuberculosis clusters in the riverine municipalities of the Brazilian Amazon with machine learning

OBJECTIVE

Tuberculosis (TB) is the second most deadly infectious disease globally, posing a significant burden in Brazil and its Amazonian region. This study focused on the "riverine municipalities" and hypothesizes the presence of TB clusters in the area. We also aimed to train a machine learning model to differentiate municipalities classified as hot spots vs. non-hot spots using disease surveillance variables as predictors.

METHODS

Data regarding the incidence of TB from 2019 to 2022 in the riverine town was collected from the Brazilian Health Ministry Informatics Department. Moran's I was used to assess global spatial autocorrelation, while the Getis-Ord GI* method was employed to detect high and low-incidence clusters. A Random Forest machine-learning model was trained using surveillance variables related to TB cases to predict hot spots among non-hot spot municipalities.

RESULTS

Our analysis revealed distinct geographical clusters with high and low TB incidence following a west-to-east distribution pattern. The Random Forest Classification model utilizes six surveillance variables to predict hot vs. non-hot spots. The machine learning model achieved an Area Under the Receiver Operator Curve (AUC-ROC) of 0.81.

CONCLUSION

Municipalities with higher percentages of recurrent cases, deaths due to TB, antibiotic regimen changes, percentage of new cases, and cases with smoking history were the best predictors of hot spots. This prediction method can be leveraged to identify the municipalities at the highest risk of being hot spots for the disease, aiding policymakers with an evidenced-based tool to direct resource allocation for disease control in the riverine municipalities.

Development and Validation of a Machine Learning Model to Predict Post-dispatch Cancellation of Physician-staffed Rapid Car

Objectives

This study aimed to develop and validate a machine learning prediction model for post-dispatch cancellation of physician-staffed rapid car.

Materials

Data were extracted from the physician-staffed rapid response car database at our Hospital between April 2017 and March 2019.

Methods

After obtaining 2019 cases, we divided the dataset into a training set for developing the model and a test set for validation using stratified random sampling with an 8 : 2 allocation ratio. We selected random forest as the machine-learning classifier. The outcome was the post-dispatch cancellation of a rapid car. The model was trained using predictor variables, including 18 different reasons for rapid car request, age and gender of a patient, date (month), and distance from the hospital.

Results

This machine learning model predicted the occurrence of post-dispatch cancellation of rapid cars with an accuracy of 75.5% [95% confidence interval (CI): 71.0-79.6], sensitivity of 81.5% (CI: 75.0-86.9), specificity of 70.8% (CI: 64.4-76.6), and an area under the receiver operating characteristic value of 0.83 (CI: 0.79-0.87). The important features were distance from the hospital to the scene, age, suspicion of non-witnessed cardiac arrest, farthest geographic area, and date (months).

Conclusions

We developed a favorable machine learning model to predict post-dispatch cancellation of rapid cars in a local district. This study suggests the potential of machine-learning models in improving the efficiency of dispatching physicians outside hospitals.

Creating machine learning models that interpretably link systemic inflammatory index, sex steroid hormones, and dietary antioxidants to identify gout using the SHAP (SHapley Additive exPlanations) method

Background

The relationship between systemic inflammatory index (SII), sex steroid hormones, dietary antioxidants (DA), and gout has not been determined. We aim to develop a reliable and interpretable machine learning (ML) model that links SII, sex steroid hormones, and DA to gout identification.

Methods

The dataset we used to study the relationship between SII, sex steroid hormones, DA, and gout was from the National Health and Nutrition Examination Survey (NHANES). Six ML models were developed to identify gout by SII, sex steroid hormones, and DA. The seven performance discriminative features of each model were summarized, and the eXtreme Gradient Boosting (XGBoost) model with the best overall performance was selected to identify gout. We used the SHapley Additive exPlanation (SHAP) method to explain the XGBoost model and its decision-making process.

Results

An initial survey of 20,146 participants resulted in 8,550 being included in the study. Selecting the best performing XGBoost model associated with SII, sex steroid hormones, and DA to identify gout (male: AUC: 0.795, 95% CI: 0.746- 0.843, accuracy: 98.7%; female: AUC: 0.822, 95% CI: 0.754- 0.883, accuracy: 99.2%). In the male group, The SHAP values showed that the lower feature values of lutein + zeaxanthin (LZ), vitamin C (VitC), lycopene, zinc, total testosterone (TT), vitamin E (VitE), and vitamin A (VitA), the greater the positive effect on the model output. In the female group, SHAP values showed that lower feature values of E2, zinc, lycopene, LZ, TT, and selenium had a greater positive effect on model output.

Conclusion

The interpretable XGBoost model demonstrated accuracy, efficiency, and robustness in identifying associations between SII, sex steroid hormones, DA, and gout in participants. Decreased TT in males and decreased E2 in females may be associated with gout, and increased DA intake and decreased SII may reduce the potential risk of gout.

Validation of mitochondrial biomarkers and immune dynamics in polycystic ovary syndrome

PROBLEM

Polycystic ovary syndrome (PCOS), a prevalent endocrine-metabolic disorder, presents considerable therapeutic challenges due to its complex and elusive pathophysiology.

METHOD OF STUDY

We employed three machine learning algorithms to identify potential biomarkers within a training dataset, comprising GSE138518, GSE155489, and GSE193123. The diagnostic accuracy of these biomarkers was rigorously evaluated using a validation dataset using area under the curve (AUC) metrics. Further validation in clinical samples was conducted using PCR and immunofluorescence techniques. Additionally, we investigate the complex interplay among immune cells in PCOS using CIBERSORT to uncover the relationships between the identified biomarkers and various immune cell types.

RESULTS

Our analysis identified ACSS2, LPIN1, and NR4A1 as key mitochondria-related biomarkers associated with PCOS. A notable difference was observed in the immune microenvironment between PCOS patients and healthy controls. In particular, LPIN1 exhibited a positive correlation with resting mast cells, whereas NR4A1 demonstrated a negative correlation with monocytes in PCOS patients.

CONCLUSION

ACSS2, LPIN1, and NR4A1 emerge as PCOS-related diagnostic biomarkers and potential intervention targets, opening new avenues for the diagnosis and management of PCOS.

Automatic identification of hypertension and assessment of its secondary effects using artificial intelligence: A systematic review (2013-2023)

Artificial Intelligence (AI) techniques are increasingly used in computer-aided diagnostic tools in medicine. These techniques can also help to identify Hypertension (HTN) in its early stage, as it is a global health issue. Automated HTN detection uses socio-demographic, clinical data, and physiological signals. Additionally, signs of secondary HTN can also be identified using various imaging modalities. This systematic review examines related work on automated HTN detection. We identify datasets, techniques, and classifiers used to develop AI models from clinical data, physiological signals, and fused data (a combination of both). Image-based models for assessing secondary HTN are also reviewed. The majority of the studies have primarily utilized single-modality approaches, such as biological signals (e.g., electrocardiography, photoplethysmography), and medical imaging (e.g., magnetic resonance angiography, ultrasound). Surprisingly, only a small portion of the studies (22 out of 122) utilized a multi-modal fusion approach combining data from different sources. Even fewer investigated integrating clinical data, physiological signals, and medical imaging to understand the intricate relationships between these factors. Future research directions are discussed that could build better healthcare systems for early HTN detection through more integrated modeling of multi-modal data sources.

Predicting hypertension control using machine learning

Hypertension is a widely prevalent disease and uncontrolled hypertension predisposes affected individuals to severe adverse effects. Though the importance of controlling hypertension is clear, the multitude of therapeutic regimens and patient factors that affect the success of blood pressure control makes it difficult to predict the likelihood to predict whether a patient's blood pressure will be controlled. This project endeavors to investigate whether machine learning can accurately predict the control of a patient's hypertension within 12 months of a clinical encounter. To build the machine learning model, a retrospective review of the electronic medical records of 350,008 patients 18 years of age and older between January 1, 2015 and June 1, 2022 was performed to form model training and testing cohorts. The data included in the model included medication combinations, patient laboratory values, vital sign measurements, comorbidities, healthcare encounters, and demographic information. The mean age of the patient population was 65.6 years with 161,283 (46.1%) men and 275,001 (78.6%) white. A sliding time window of data was used to both prohibit data leakage from training sets to test sets and to maximize model performance. This sliding window resulted in using the study data to create 287 predictive models each using 2 years of training data and one week of testing data for a total study duration of five and a half years. Model performance was combined across all models. The primary outcome, prediction of blood pressure control within 12 months demonstrated an area under the curve of 0.76 (95% confidence interval; 0.75-0.76), sensitivity of 61.52% (61.0-62.03%), specificity of 75.69% (75.25-76.13%), positive predictive value of 67.75% (67.51-67.99%), and negative predictive value of 70.49% (70.32-70.66%). An AUC of 0.756 is considered to be moderately good for machine learning models. While the accuracy of this model is promising, it is impossible to state with certainty the clinical relevancy of any clinical support ML model without deploying it in a clinical setting and studying its impact on health outcomes. By also incorporating uncertainty analysis for every prediction, the authors believe that this approach offers the best-known solution to predicting hypertension control and that machine learning may be able to improve the accuracy of hypertension control predictions using patient information already available in the electronic health record. This method can serve as a foundation with further research to strengthen the model accuracy and to help determine clinical relevance.

Predicting the Rapid Progression of Mild Cognitive Impairment by Intestinal Flora and Blood Indicators through Machine Learning Method

INTRODUCTION

The aim of the work was to establish a prediction model of mild cognitive impairment (MCI) progression based on intestinal flora by machine learning method.

METHOD

A total of 1,013 patients were recruited, in which 87 patients with MCI finished a two-year follow-up. To establish a prediction model, 61 patients were randomly divided into a training set and 26 patients were divided into a testing set. A total of 121 features including demographic characteristics, hematological indicators, and intestinal flora abundance were analyzed.

RESULTS

Of the 87 patients who finished a two-year follow-up, 44 presented rapid progression. Model 1 was established based on 121 features with the accuracy 85%, sensitivity 85%, and specificity 83%. Model 2 was based on the first fifteen features of model 1 (triglyceride, uric acid, alanine transaminase, F-Clostridiaceae, G-Megamonas, S-Megamonas, G-Shigella, G-Shigella, S-Shigella, average hemoglobin concentration, G-Alistipes, S-Collinsella, median cell count, average hemoglobin volume, low-density lipoprotein), with the accuracy 97%, sensitivity 92%, and specificity 100%. Model 3 was based on the first ten features of model 1, with the accuracy 97%, sensitivity 86%, and specificity 100%. Other models based on the demographic characteristics, hematological indicators, or intestinal flora abundance features presented lower sensitivity and specificity.

CONCLUSION

The 15 features (including intestinal flora abundance) could establish an effective model for predicting rapid MCI progression.

Machine learning-based models for prediction of the risk of stroke in coronary artery disease patients receiving coronary revascularization

BACKGROUND

To construct several prediction models for the risk of stroke in coronary artery disease (CAD) patients receiving coronary revascularization based on machine learning methods.

METHODS

In total, 5757 CAD patients receiving coronary revascularization admitted to ICU in Medical Information Mart for Intensive Care IV (MIMIC-IV) were included in this cohort study. All the data were randomly split into the training set (n = 4029) and testing set (n = 1728) at 7:3. Pearson correlation analysis and least absolute shrinkage and selection operator (LASSO) regression model were applied for feature screening. Variables with Pearson correlation coefficient<9 were included, and the regression coefficients were set to 0. Features more closely related to the outcome were selected from the 10-fold cross-validation, and features with non-0 Coefficent were retained and included in the final model. The predictive values of the models were evaluated by sensitivity, specificity, area under the curve (AUC), accuracy, and 95% confidence interval (CI).

RESULTS

The Catboost model presented the best predictive performance with the AUC of 0.831 (95%CI: 0.811-0.851) in the training set, and 0.760 (95%CI: 0.722-0.798) in the testing set. The AUC of the logistic regression model was 0.789 (95%CI: 0.764-0.814) in the training set and 0.731 (95%CI: 0.686-0.776) in the testing set. The results of Delong test revealed that the predictive value of the Catboost model was significantly higher than the logistic regression model (P<0.05). Charlson Comorbidity Index (CCI) was the most important variable associated with the risk of stroke in CAD patients receiving coronary revascularization.

CONCLUSION

The Catboost model was the optimal model for predicting the risk of stroke in CAD patients receiving coronary revascularization, which might provide a tool to quickly identify CAD patients who were at high risk of postoperative stroke.

Machine-learning-based prediction of a diagnostic model using autophagy-related genes based on RNA sequencing for patients with papillary thyroid carcinoma

Papillary thyroid carcinoma (PTC) is the most common type of thyroid cancer and belongs to the category of malignant tumors of the thyroid gland. Autophagy plays an important role in PTC. The purpose of this study is to develop a novel diagnostic model using autophagy-related genes (ARGs) in patients. In this study, RNA sequencing data of PTC samples and normal samples were obtained from GSE33630 and GSE29265. Then, we analyzed GSE33630 datasets and identified 127 DE-ARGs. Functional enrichment analysis suggested that 127 DE-ARGs were mainly enriched in pathways in cancer, protein processing in endoplasmic reticulum, toll-like receptor pathway, MAPK pathway, apoptosis, neurotrophin signaling pathway, and regulation of autophagy. Subsequently, CALCOCO2, DAPK1, and RAC1 among the 127 DE-ARGs were identified as diagnostic genes by support vector machine recursive feature elimination and least absolute shrinkage and selection operator algorithms. Then, we developed a novel diagnostic model using CALCOCO2, DAPK1, and RAC1 and its diagnostic value was confirmed in GSE29265 and our cohorts. Importantly, CALCOCO2 may be a critical regulator involved in immune microenvironment because its expression was related to many types of immune cells. Overall, we developed a novel diagnostic model using CALCOCO2, DAPK1, and RAC1 which can be used as diagnostic markers of PTC.

Identification and verification of diagnostic biomarkers based on mitochondria-related genes related to immune microenvironment for preeclampsia using machine learning algorithms

Preeclampsia is one of the leading causes of maternal and fetal morbidity and mortality worldwide. Preeclampsia is linked to mitochondrial dysfunction as a contributing factor in its progression. This study aimed to develop a novel diagnostic model based on mitochondria-related genes(MRGs) for preeclampsia using machine learning and further investigate the association of the MRGs and immune infiltration landscape in preeclampsia. In this research, we analyzed GSE75010 database and screened 552 DE-MRGs between preeclampsia samples and normal samples. Enrichment assays indicated that 552 DE-MRGs were mainly related to energy metabolism pathway and several different diseases. Then, we performed LASSO and SVM-RFE and identified three critical diagnostic genes for preeclampsia, including CPOX, DEGS1 and SH3BP5. In addition, we developed a novel diagnostic model using the above three genes and its diagnostic value was confirmed in GSE44711, GSE75010 datasets and our cohorts. Importantly, the results of RT-PCR confirmed the expressions of CPOX, DEGS1 and SH3BP5 were distinctly increased in preeclampsia samples compared with normal samples. The results of the CIBERSORT algorithm revealed a striking dissimilarity between the immune cells found in preeclampsia samples and those found in normal samples. In addition, we found that the levels of SH3BP5 were closely associated with several immune cells, highlighting its potential involved in immune microenvironment of preeclampsia. Overall, this study has provided a novel diagnostic model and diagnostic genes for preeclampsia while also revealing the association between MRGs and immune infiltration. These findings offer valuable insights for further research and treatment of preeclampsia.

Construction of prognosis prediction model and visualization system of acute paraquat poisoning based on improved machine learning model

Objective

This study aims to develop a prognosis prediction model and visualization system for acute paraquat poisoning based on an improved machine learning model.

Methods

101 patients with acute paraquat poisoning admitted to 6 hospitals from March 2020 to March 2022 were selected for this study. After expiry of the treatment period (one year of follow-up for survivors and up to the time of death for deceased patients) and they were categorized into the survival group (n = 37) and death group (n = 64). The biochemical indexes of the patients were analyzed, and a prognosis prediction model was constructed using HHO-XGBoost, an improved machine-learning algorithm. Multivariate logistic analysis was used to verify the value of the self-screening features in the model.

Results

Seven features were selected in the HHO-XGBoost model, including oral dose, serum creatinine, alanine aminotransferase (ALT), white blood cell (WBC) count, neutrophil count, urea nitrogen level, and thrombin time. Univariate analysis showed statistically significant differences between these features' survival and death groups (P < 0.05). Multivariate logistic analysis identified four features significantly associated with prognosis- serum creatinine level, oral dose, ALT level, and WBC count - indicating their critical significance in predicting outcomes.

Conclusion

The HHO-XGBoost model based on machine learning is precious in constructing a prognosis prediction model and visualization system for acute paraquat poisoning, which can help clinical prognosis prediction of patients with paraquat poisoning.

Machine learning aided non-invasive diagnosis of coronary heart disease based on tongue features fusion

BACKGROUND

Coronary heart disease (CHD) is the first cause of death globally. Hypertension is considered to be the most important independent risk factor for CHD. Early and accurate diagnosis of CHD in patients with hypertension can plays a significant role in reducing the risk and harm of hypertension combined with CHD.

OBJECTIVE

To propose a non-invasive method for early diagnosis of coronary heart disease according to tongue image features with the help of machine learning techniques.

METHODS

We collected standard tongue images and extract features by Diagnosis Analysis System (TDAS) and ResNet-50. On the basis of these tongue features, a common machine learning method is used to customize the non-invasive CHD diagnosis algorithm based on tongue image.

RESULTS

Based on feature fusion, our algorithm has good performance. The results showed that the XGBoost model with fused features had the best performance with accuracy of 0.869, the AUC of 0.957, the AUPR of 0.961, the precision of 0.926, the recall of 0.806, and the F1-score of 0.862.

CONCLUSION

We provide a feasible, convenient, and non-invasive method for the diagnosis and large-scale screening of CHD. Tongue image information is a possible effective marker for the diagnosis of CHD.

Taking precautions in advance: a lower level of activities of daily living may be associated with a higher likelihood of memory-related diseases

Introduction

Memory-related diseases (MDs) pose a significant healthcare challenge globally, and early detection is essential for effective intervention. This study investigates the potential of Activities of Daily Living (ADL) as a clinical diagnostic indicator for MDs. Utilizing data from the 2018 national baseline survey of the China Health and Retirement Longitudinal Study (CHARLS), encompassing 10,062 Chinese individuals aged 45 or older, we assessed ADL using the Barthel Index (BI) and correlated it with the presence of MDs. Statistical analysis, supplemented by machine learning algorithms (Support Vector Machine, Decision Tree, and Logistic Regression), was employed to elucidate the relationship between ADL and MDs.

Background

MDs represent a significant public health concern, necessitating early detection and intervention to mitigate their impact on individuals and society. Identifying reliable clinical diagnostic signs for MDs is imperative. ADL have garnered attention as a potential marker. This study aims to rigorously analyze clinical data and validate machine learning algorithms to ascertain if ADL can serve as an indicator of MDs.

Methods

Data from the 2018 national baseline survey of the China Health and Retirement Longitudinal Study (CHARLS) were employed, encompassing responses from 10,062 Chinese individuals aged 45 or older. ADL was assessed using the BI, while the presence of MDs was determined through health report questions. Statistical analysis was executed using SPSS 25.0, and machine learning algorithms, including Support Vector Machine (SVM), Decision Tree Learning (DT), and Logistic Regression (LR), were implemented using Python 3.10.2.

Results

Population characteristics analysis revealed that the average BI score for individuals with MDs was 70.88, significantly lower than the average score of 87.77 in the control group. Pearson's correlation analysis demonstrated a robust negative association (r = -0.188, p < 0.001) between ADL and MDs. After adjusting for covariates such as gender, age, smoking status, drinking status, hypertension, diabetes, and dyslipidemia, the negative relationship between ADL and MDs remained statistically significant (B = -0.002, β = -0.142, t = -14.393, 95% CI = -0.002, -0.001, p = 0.000). The application of machine learning models further confirmed the predictive accuracy of ADL for MDs, with area under the curve (AUC) values as follows: SVM-AUC = 0.69, DT-AUC = 0.715, LR-AUC = 0.7. Comparative analysis of machine learning outcomes with and without the BI underscored the BI's role in enhancing predictive abilities, with the DT model demonstrating superior performance.

Conclusion

This study establishes a robust negative correlation between ADL and MDs through comprehensive statistical analysis and machine learning algorithms. The results validate ADL as a promising diagnostic indicator for MDs, with enhanced predictive accuracy when coupled with the Barthel Index. Lower levels of ADL are associated with an increased likelihood of developing memory-related diseases, underscoring the clinical relevance of ADL assessment in early disease detection.

Machine learning and experiments identifies SPINK1 as a candidate diagnostic and prognostic biomarker for hepatocellular carcinoma

Machine learning techniques have been widely used in predicting disease prognosis, including cancer prognosis. One of the major challenges in cancer prognosis is to accurately classify cancer types and stages to optimize early screening and detection, and machine learning techniques have proven to be very useful in this regard. In this study, we aimed at identifying critical genes for diagnosis and outcomes of hepatocellular carcinoma (HCC) patients using machine learning. The HCC expression dataset was downloaded from GSE65372 datasets and TCGA datasets. Differentially expressed genes (DEGs) were identified between 39 HCC and 15 normal samples. For the purpose of locating potential biomarkers, the LASSO and the SVM-RFE assays were performed. The ssGSEA method was used to analyze the TCGA to determine whether there was an association between SPINK1 and tumor immune infiltrates. RT-PCR was applied to examine the expression of SPINK1 in HCC specimens and cells. A series of functional assays were applied to examine the function of SPINK1 knockdown on the proliferation of HCC cells. In this study, 103 DEGs were obtained. Based on LASSO and SVM-RFE analysis, we identified nine critical diagnostic genes, including C10orf113, SPINK1, CNTLN, NRG3, HIST1H2AI, GPRIN3, SCTR, C2orf40 and PITX1. Importantly, we confirmed SPINK1 as a prognostic gene in HCC. Multivariate analysis confirmed that SPINK1 was an independent prognostic factor for overall survivals of HCC patients. We also found that SPINK1 level was positively associated with Macrophages, B cells, TFH, T cells, Th2 cells, iDC, NK CD56bright cells, Th1 cells, aDC, while negatively associated with Tcm and Eosinophils. Finally, we demonstrated that SPINK1 expression was distinctly increased in HCC specimens and cells. Functionally, silence of SPINK1 distinctly suppressed the proliferation of HCC cells via regulating Wnt/β-catenin pathway. The evidence provided suggested that SPINK1 may possess oncogenic properties by inducing dysregulated immune infiltration in HCC. Additionally, SPINK1 was identified as a novel biomarker and therapeutic target for HCC.

Prognostic estimation for acute ischemic stroke patients undergoing mechanical thrombectomy within an extended therapeutic window using an interpretable machine learning model

Background

Mechanical thrombectomy (MT) is effective for acute ischemic stroke with large vessel occlusion (AIS-LVO) within an extended therapeutic window. However, successful reperfusion does not guarantee positive prognosis, with around 40-50% of cases yielding favorable outcomes. Preoperative prediction of patient outcomes is essential to identify those who may benefit from MT. Although machine learning (ML) has shown promise in handling variables with non-linear relationships in prediction models, its "black box" nature and the absence of ML models for extended-window MT prognosis remain limitations.

Objective

This study aimed to establish and select the optimal model for predicting extended-window MT outcomes, with the Shapley additive explanation (SHAP) approach used to enhance the interpretability of the selected model.

Methods

A retrospective analysis was conducted on 260 AIS-LVO patients undergoing extended-window MT. Selected patients were allocated into training and test sets at a 3:1 ratio following inclusion and exclusion criteria. Four ML classifiers and one logistic regression (Logit) model were constructed using pre-treatment variables from the training set. The optimal model was selected through comparative validation, with key features interpreted using the SHAP approach. The effectiveness of the chosen model was further evaluated using the test set.

Results

Of the 212 selected patients, 159 comprised the training and 53 the test sets. Extreme gradient boosting (XGBoost) showed the highest discrimination with an area under the curve (AUC) of 0.93 during validation, and maintained an AUC of 0.77 during testing. SHAP analysis identified ischemic core volume, baseline NHISS score, ischemic penumbra volume, ASPECTS, and patient age as the top five determinants of outcome prediction.

Conclusion

XGBoost emerged as the most effective for predicting the prognosis of AIS-LVO patients undergoing MT within the extended therapeutic window. SHAP interpretation improved its clinical confidence, paving the way for ML in clinical decision-making.

Machine Learning-Enhanced Biomass Pressure Sensor with Embedded Wrinkle Structures Created by Surface Buckling

Flexible piezoresistive sensors are core components of many wearable devices to detect deformation and motion. However, it is still a challenge to conveniently prepare high-precision sensors using natural materials and identify similar short vibration signals. In this study, inspired by microstructures of human skins, biomass flexible piezoresistive sensors were prepared by assembling two wrinkled surfaces of konjac glucomannan and k-carrageenan composite hydrogel. The wrinkle structures were conveniently created by hardness gradient-induced surface buckling and coated with MXene sheets to capture weak pressure signals. The sensor was applied to detect various slight body movements, and a machine learning method was used to enhance the identification of similar and short throat vibration signals. The results showed that the sensor exhibited a high sensitivity of 5.1 kPa-1 under low pressure (50 Pa), a fast response time (104 ms), and high stability over 100 cycles. The XGBoost machine learning model accurately distinguished short voice vibrations similar to those of individual English letters. Moreover, experiments and numerical simulations were carried out to reveal the mechanism of the wrinkle structure preparation and the excellent sensing performance. This biomass sensor preparation and the machine learning method will promote the optimization and application of wearable devices.

Machine learning models for predicting the risk factor of carotid plaque in cardiovascular disease

Introduction

Cardiovascular disease (CVD) is a group of diseases involving the heart or blood vessels and represents a leading cause of death and disability worldwide. Carotid plaque is an important risk factor for CVD that can reflect the severity of atherosclerosis. Accordingly, developing a prediction model for carotid plaque formation is essential to assist in the early prevention and management of CVD.

Methods

In this study, eight machine learning algorithms were established, and their performance in predicting carotid plaque risk was compared. Physical examination data were collected from 4,659 patients and used for model training and validation. The eight predictive models based on machine learning algorithms were optimized using the above dataset and 10-fold cross-validation. The Shapley Additive Explanations (SHAP) tool was used to compute and visualize feature importance. Then, the performance of the models was evaluated according to the area under the receiver operating characteristic curve (AUC), feature importance, accuracy and specificity.

Results

The experimental results indicated that the XGBoost algorithm outperformed the other machine learning algorithms, with an AUC, accuracy and specificity of 0.808, 0.749 and 0.762, respectively. Moreover, age, smoke, alcohol drink and BMI were the top four predictors of carotid plaque formation. It is feasible to predict carotid plaque risk using machine learning algorithms.

Conclusions

This study indicates that our models can be applied to routine chronic disease management procedures to enable more preemptive, broad-based screening for carotid plaque and improve the prognosis of CVD patients.

Artificial intelligence in systematic reviews: promising when appropriately used

BACKGROUND

Systematic reviews provide a structured overview of the available evidence in medical-scientific research. However, due to the increasing medical-scientific research output, it is a time-consuming task to conduct systematic reviews. To accelerate this process, artificial intelligence (AI) can be used in the review process. In this communication paper, we suggest how to conduct a transparent and reliable systematic review using the AI tool 'ASReview' in the title and abstract screening.

METHODS

Use of the AI tool consisted of several steps. First, the tool required training of its algorithm with several prelabelled articles prior to screening. Next, using a researcher-in-the-loop algorithm, the AI tool proposed the article with the highest probability of being relevant. The reviewer then decided on relevancy of each article proposed. This process was continued until the stopping criterion was reached. All articles labelled relevant by the reviewer were screened on full text.

RESULTS

Considerations to ensure methodological quality when using AI in systematic reviews included: the choice of whether to use AI, the need of both deduplication and checking for inter-reviewer agreement, how to choose a stopping criterion and the quality of reporting. Using the tool in our review resulted in much time saved: only 23% of the articles were assessed by the reviewer.

CONCLUSION

The AI tool is a promising innovation for the current systematic reviewing practice, as long as it is appropriately used and methodological quality can be assured.

PROSPERO REGISTRATION NUMBER

CRD42022283952.

Differential diagnosis of secondary hypertension based on deep learning

Secondary hypertension is associated with higher risks of target organ damage and cardiovascular and cerebrovascular disease events. Early aetiology identification can eliminate aetiologies and control blood pressure. However, inexperienced doctors often fail to diagnose secondary hypertension, and comprehensively screening for all causes of high blood pressure increases health care costs. To date, deep learning has rarely been involved in the differential diagnosis of secondary hypertension. Relevant machine learning methods cannot combine textual information such as chief complaints with numerical information such as the laboratory examination results in electronic health records (EHRs), and the use of all features increases health care costs. To reduce redundant examinations and accurately identify secondary hypertension, we propose a two-stage framework that follows clinical procedures. The framework carries out an initial diagnosis process in the first stage, on which basis patients are recommended for disease-related examinations, followed by differential diagnoses of different diseases based on the different characteristics observed in the second stage. We convert the numerical examination results into descriptive sentences, thus blending textual and numerical characteristics. Medical guidelines are introduced through label embedding and attention mechanisms to obtain interactive features. Our model was trained and evaluated using a cross-sectional dataset containing 11,961 patients with hypertension from January 2013 to December 2019. The F1 scores of our model were 0.912, 0.921, 0.869 and 0.894 for primary aldosteronism, thyroid disease, nephritis and nephrotic syndrome and chronic kidney disease, respectively, which are four kinds of secondary hypertension with high incidence rates. The experimental results show that our model can powerfully use the textual and numerical data contained in EHRs to provide effective decision support for the differential diagnosis of secondary hypertension.

A molecular classification of gastric cancer associated with distinct clinical outcomes and validated by an XGBoost-based prediction model

Gastric cancer (GC) is a heterogeneous disease and a leading cause of cancer-related deaths. Discovering robust, clinically relevant molecular classifications is critical for guiding personalized therapies for GC. Here, we propose a refined molecular classification scheme for GC using integrated optimal algorithms and multi-omics data. Based on the important features of mRNA, microRNA, and DNA methylation data selected by the multivariate Cox regression model, three subtypes linked to distinct clinical outcomes were identified by combining similarity network fusion and consensus clustering methods. Three subtypes were validated by an extreme gradient boosting machine learning prediction model with 125 differentially expressed genes in multiple independent cohorts. The molecular characteristics of mutation signatures, characteristic gene sets, driver genes, and chemotherapy sensitivity for each subtype were also identified: subtype 1 was associated with favorable prognosis and characterized by high ARID1A and PIK3CA mutations, subtype 2 was associated with a poor prognosis and harbored high recurrent TP53 mutations, and subtype 3 was associated with high CHD1, APOA1 mutations, and a poor prognosis. The proposed three-subtype scheme achieved a better clinical prediction performance (area under the curve value = 0.71) than The Cancer Genome Atlas classification, which may provide a practical subtyping framework to improve the treatment of GC.