Stable Iterative Variable Selection

Prediction of Early Adverse Events After THA: A Comparison of Different Machine-Learning Strategies Based on 262,356 Observations From the Nordic Arthroplasty Register Association (NARA) Dataset

OBJECTIVE

Preoperative risk prediction models can support shared decision-making before total hip arthroplasties (THAs). Here, we compare different machine-learning (ML) approaches to predict the six-month risk of adverse events following primary THA to obtain accurate yet simple-to-use risk prediction models.

METHODS

We extracted data on primary THAs (N = 262,356) between 2010 and 2018 from the Nordic Arthroplasty Register Association dataset. We benchmarked a variety of ML algorithms in terms of the area under the receiver operating characteristic curve (AUROC) for predicting the risk of revision caused by periprosthetic joint infection (PJI), dislocation or periprosthetic fracture (PPF), and death. All models were internally validated against a randomly selected test cohort (one-third of the data) that was not used for training the models.

RESULTS

The incidences of revisions because of PJI, dislocation, and PPF were 0.8%, 0.4%, and 0.3%, respectively, and the incidence of death was 1.2%. Overall, Lasso regression with stable iterative variable selection (SIVS) produced models using only four to five input variables but with AUROC comparable to more complex models using all 32 variables available. The SIVS-based Lasso models based on age, sex, preoperative diagnosis, bearing couple, fixation, and surgical approach predicted the risk of revisions caused by PJI, dislocations, and PPF, as well as death, with AUROCs of 0.61, 0.67, 0.76, and 0.86, respectively.

CONCLUSION

Our study demonstrates that satisfactory predictive potential for adverse events following THA can be reached with parsimonious modeling strategies. The SIVS-based Lasso models may serve as simple-to-use tools for clinical risk assessment in the future.

Serum hydroxysteroid (17beta) dehydrogenase 1 concentration in pregnant women correlates with pregnancy-associated plasma protein A but does not serve as an independent marker for preeclampsia†

Hydroxysteroid (17beta) dehydrogenase 1 (HSD17B1) is a steroid synthetic enzyme expressed in ovarian granulosa cells and placental syncytiotrophoblasts. Here, HSD17B1 serum concentration was measured with a validated immunoassay during pregnancy at three time points (12-14, 18-20 and 26-28 weeks of gestation). The concentration increased 2.5-fold (P < 0.0001) and 1.7-fold (P = 0.0019) during the follow-up period for control women and women who later developed preeclampsia (PE), respectively, and a significant difference was observed at weeks 26-28 (P = 0.0266). HSD17B1 concentration at all the three time points positively correlated with serum PAPPA measured at the first time point (first time point r = 0.38, P = 1.1 × 10-10; second time point r = 0.27, P = 5.9 × 10-6 and third timepoint r = 0.26, P = 2.3 × 10-5). No correlation was observed between HSD17B1 and placental growth factor (PLGF). Serum HSD17B1 negatively correlated with the mother's weight and body mass index (BMI), mirroring the pattern observed for PAPPA. The univariable logistic regression identified a weak association between HSD17B1 at 26-28 weeks and later development of PE (P = 0.04). The best multivariable model obtained using penalized logistic regression with stable iterative variable selection at 26-28 weeks included HSD17B1, together with PLGF, PAPPA and mother's BMI. While the area under the receiver operating characteristic curve of the model was higher than that of the adjusted PLGF, the difference was not statistically significant. In summary, the serum concentration of HSD17B1 correlated with PAPPA, another protein expressed in syncytiotrophoblasts, and with mother's weight and BMI but could not be considered as an independent marker for PE.

Resting-state EEG dynamic functional connectivity distinguishes non-psychotic major depression, psychotic major depression and schizophrenia

This study aims to identify dynamic patterns within the spatiotemporal feature space that are specific to nonpsychotic major depression (NPMD), psychotic major depression (PMD), and schizophrenia (SCZ). The study also evaluates the effectiveness of machine learning algorithms based on these network manifestations in differentiating individuals with NPMD, PMD, and SCZ. A total of 579 participants were recruited, including 152 patients with NPMD, 45 patients with PMD, 185 patients with SCZ, and 197 healthy controls (HCs). A dynamic functional connectivity (DFC) approach was employed to estimate the principal FC states within each diagnostic group. Incremental proportions of data (ranging from 10% to 100%) within each diagnostic group were used for variability testing. DFC metrics, such as proportion, mean duration, and transition number, were examined among the four diagnostic groups to identify disease-related neural activity patterns. These patterns were then used to train a two-layer classifier for the four groups (HC, NPMD, PMD, and SCZ). The four principal brain states (i.e., states 1,2,3, and 4) identified by the DFC approach were highly representative within and across diagnostic groups. Between-group comparisons revealed significant differences in network metrics of state 2 and state 3, within delta, theta, and gamma frequency bands, between healthy individuals and patients in each diagnostic group (p < 0.01, FDR corrected). Moreover, the identified key dynamic network metrics achieved an accuracy of 73.1 ± 2.8% in the four-way classification of HC, NPMD, PMD, and SCZ, outperforming the static functional connectivity (SFC) approach (p < 0.001). These findings suggest that the proposed DFC approach can identify dynamic network biomarkers at the single-subject level. These biomarkers have the potential to accurately differentiate individual subjects among various diagnostic groups of psychiatric disorders or healthy controls. This work may contribute to the development of a valuable EEG-based diagnostic tool with enhanced accuracy and assistive capabilities.

Single-cell sequencing and transcriptome analyses in the construction of a liquid-liquid phase separation-associated gene model for rheumatoid arthritis

Background: Rheumatoid arthritis (RA) is a disabling autoimmune disease that affects multiple joints. Accumulating evidence suggests that imbalances in liquid-liquid phase separation (LLPS) can lead to altered spatiotemporal coordination of biomolecular condensates, which play important roles in carcinogenesis and inflammatory diseases. However, the role of LLPS in the development and progression of RA remains unclear. Methods: We screened RA and normal samples from GSE12021, GSE55235, and GSE55457 transcriptome datasets and GSE129087 and GSE109449 single-cell sequencing datasets from Gene Expression Omnibus database to investigate the pathogenesis of LLPS-related hub genes at the transcriptome and single cell sequencing levels. Machine learning algorithms and weighted gene co-expression network analysis were applied to screen hub genes, and hub genes were validated using correlation studies. Results: Differential analysis showed that 36 LLPS-related genes were significantly differentially expressed in RA, further random forest and support vector machine identified four and six LLPS-related genes, respectively, and weighted gene co-expression network analysis identified 396 modular genes. Hybridization of the three sets revealed two hub genes, MYC and MAP1LC3B, with AUCs of 0.907 and 0.911, respectively. Further ROC analysis of the hub genes in the GSE55457 dataset showed that the AUCs of MYC and MAP1LC3B were 0.815 and 0.785, respectively. qRT-PCR showed that the expression of MYC and MAP1LC3B in RA synovial tissues was significantly lower than that in the normal control synovial tissues. Correlation analysis between hub genes and the immune microenvironment and single-cell sequencing analysis revealed that both MYC and MAP1LC3B were significantly correlated with the degree of infiltration of various innate and acquired immune cells. Conclusion: Our study reveals a possible mechanism for LLPS in RA pathogenesis and suggests that MYC and MAP1LC3B may be potential novel molecular markers for RA with immunological significance.

Development and validation of COEWS (COVID-19 Early Warning Score) for hospitalized COVID-19 with laboratory features: A multicontinental retrospective study

Background

The emergence of new SARS-CoV-2 variants with significant immune-evasiveness, the relaxation of measures for reducing the number of infections, the waning of immune protection (particularly in high-risk population groups), and the low uptake of new vaccine boosters, forecast new waves of hospitalizations and admission to intensive care units. There is an urgent need for easily implementable and clinically effective Early Warning Scores (EWSs) that can predict the risk of complications within the next 24-48 hr. Although EWSs have been used in the evaluation of COVID-19 patients, there are several clinical limitations to their use. Moreover, no models have been tested on geographically distinct populations or population groups with varying levels of immune protection.

Methods

We developed and validated COVID-19 Early Warning Score (COEWS), an EWS that is automatically calculated solely from laboratory parameters that are widely available and affordable. We benchmarked COEWS against the widely used NEWS2. We also evaluated the predictive performance of vaccinated and unvaccinated patients.

Results

The variables of the COEWS predictive model were selected based on their predictive coefficients and on the wide availability of these laboratory variables. The final model included complete blood count, blood glucose, and oxygen saturation features. To make COEWS more actionable in real clinical situations, we transformed the predictive coefficients of the COEWS model into individual scores for each selected feature. The global score serves as an easy-to-calculate measure indicating the risk of a patient developing the combined outcome of mechanical ventilation or death within the next 48 hr.The discrimination in the external validation cohort was 0.743 (95% confidence interval [CI]: 0.703-0.784) for the COEWS score performed with coefficients and 0.700 (95% CI: 0.654-0.745) for the COEWS performed with scores. The area under the receiver operating characteristic curve (AUROC) was similar in vaccinated and unvaccinated patients. Additionally, we observed that the AUROC of the NEWS2 was 0.677 (95% CI: 0.601-0.752) in vaccinated patients and 0.648 (95% CI: 0.608-0.689) in unvaccinated patients.

Conclusions

The COEWS score predicts death or MV within the next 48 hr based on routine and widely available laboratory measurements. The extensive external validation, its high performance, its ease of use, and its positive benchmark in comparison with the widely used NEWS2 position COEWS as a new reference tool for assisting clinical decisions and improving patient care in the upcoming pandemic waves.

Funding

University of Vienna.

Development of prediction model for alanine transaminase elevations during the first 6 months of conventional synthetic DMARD treatment

Frequent laboratory monitoring is recommended for early identification of toxicity when initiating conventional synthetic disease-modifying antirheumatic drugs (csDMARDs). We aimed at developing a risk prediction model to individualize laboratory testing at csDMARD initiation. We identified inflammatory joint disease patients (N = 1196) initiating a csDMARD in Turku University Hospital 2013-2019. Baseline and follow-up safety monitoring results were drawn from electronic health records. For rheumatoid arthritis patients, diagnoses and csDMARD initiation/cessation dates were manually confirmed. Primary endpoint was alanine transaminase (ALT) elevation of more than twice the upper limit of normal (ULN) within 6 months after treatment initiation. Computational models for predicting incident ALT elevations were developed using Lasso Cox proportional hazards regression with stable iterative variable selection (SIVS) and were internally validated against a randomly selected test cohort (1/3 of the data) that was not used for training the models. Primary endpoint was reached in 82 patients (6.9%). Among baseline variables, Lasso model with SIVS predicted subsequent ALT elevations of > 2 × ULN using higher ALT, csDMARD other than methotrexate or sulfasalazine and psoriatic arthritis diagnosis as important predictors, with a concordance index of 0.71 in the test cohort. Respectively, at first follow-up, in addition to baseline ALT and psoriatic arthritis diagnosis, also ALT change from baseline was identified as an important predictor resulting in a test concordance index of 0.72. Our computational model predicts ALT elevations after the first follow-up test with good accuracy and can help in optimizing individual testing frequency.

Development of a novel pyroptosis-related LncRNA signature with multiple significance in acute myeloid leukemia

Background: Pyroptosis, a programmed cell death (PCD) with highly inflammatory form, has been recently found to be associated with the origin of hematopoietic malignancies. Long noncoding RNA (lncRNA) had emerged as an essential mediator to regulate gene expression and been involved in oncogenesis. However, the roles of pyroptosis-related lncRNA (PRlncRNA) in acute myeloid leukemia (AML) have not yet been completely clarified. Methods: We collected AML datasets from public databases to obtain PRlncRNA associated with survival and constructed a PRlncRNA signature using Lasso-Cox regression analysis. Subsequently, we employed RT-PCR to confirm its expression difference and internal training to further verify its reliability. Next, AML patients were classified into two subgroups by the median risk score. Finally, the differences between two groups in immune infiltration, enrichment analysis and drug sensitivity were further explored. Results: A PRlncRNA signature and an effective nomogram combined with clinicopathological variables to predict the prognosis of AML were constructed. The internal validations showed that the PRlncRNA risk score model was an accurate and productive indicator to predict the outcome of AML. Furthermore, this study indicated that higher inflammatory cell and immunosuppressive cells, and less sensitive to conventional chemotherapy drugs were highlighted in the high-risk group. Conclusion: Through comprehensive analysis of PRlncRNA model, our study may offer a valuable basis for future researches in targeting pyroptosis and tumor microenvironment (TME) and provide new measures for prevention and treatment in AML.

Development and validation of prediction models for hypertension risks: A cross-sectional study based on 4,287,407 participants

Objective

To develop an optimal screening model to identify the individuals with a high risk of hypertension in China by comparing tree-based machine learning models, such as classification and regression tree, random forest, adaboost with a decision tree, extreme gradient boosting decision tree, and other machine learning models like an artificial neural network, naive Bayes, and traditional logistic regression models.

Methods

A total of 4,287,407 adults participating in the national physical examination were included in the study. Features were selected using the least absolute shrinkage and selection operator regression. The Borderline synthetic minority over-sampling technique was used for data balance. Non-laboratory and semi-laboratory analyses were carried out in combination with the selected features. The tree-based machine learning models, other machine learning models, and traditional logistic regression models were constructed to identify individuals with hypertension, respectively. Top features selected using the best algorithm and the corresponding variable importance score were visualized.

Results

A total of 24 variables were finally included for analyses after the least absolute shrinkage and selection operator regression model. The sample size of hypertensive patients in the training set was expanded from 689,025 to 2,312,160 using the borderline synthetic minority over-sampling technique algorithm. The extreme gradient boosting decision tree algorithm showed the best results (area under the receiver operating characteristic curve of non-laboratory: 0.893 and area under the receiver operating characteristic curve of semi-laboratory: 0.894). This study found that age, systolic blood pressure, waist circumference, diastolic blood pressure, albumin, drinking frequency, electrocardiogram, ethnicity (uyghur, hui, and other), body mass index, sex (female), exercise frequency, diabetes mellitus, and total bilirubin are important factors reflecting hypertension. Besides, some algorithms included in the semi-laboratory analyses showed less improvement in the predictive performance compared to the non-laboratory analyses.

Conclusion

Using multiple methods, a more significant prediction model can be built, which discovers risk factors and provides new insights into the prediction and prevention of hypertension.

Vascular Dysfunction Is Central to Alzheimer's Disease Pathogenesis in APOE e4 Carriers

Alzheimer's disease (AD) is the most common form of dementia and the leading risk factor, after age, is possession of the apolipoprotein E epsilon 4 allele (APOE4). Approximately 50% of AD patients carry one or two copies of APOE4 but the mechanisms by which it confers risk are still unknown. APOE4 carriers are reported to demonstrate changes in brain structure, cognition, and neuropathology, but findings have been inconsistent across studies. In the present study, we used multi-modal data to characterise the effects of APOE4 on the brain, to investigate whether AD pathology manifests differently in APOE4 carriers, and to determine if AD pathomechanisms are different between carriers and non-carriers. Brain structural differences in APOE4 carriers were characterised by applying machine learning to over 2000 brain MRI measurements from 33,384 non-demented UK biobank study participants. APOE4 carriers showed brain changes consistent with vascular dysfunction, such as reduced white matter integrity in posterior brain regions. The relationship between APOE4 and AD pathology was explored among the 1260 individuals from the Religious Orders Study and Memory and Aging Project (ROSMAP). APOE4 status had a greater effect on amyloid than tau load, particularly amyloid in the posterior cortical regions. APOE status was also highly correlated with cerebral amyloid angiopathy (CAA). Bulk tissue brain transcriptomic data from ROSMAP and a similar dataset from the Mount Sinai Brain Bank showed that differentially expressed genes between the dementia and non-dementia groups were enriched for vascular-related processes (e.g., "angiogenesis") in APOE4 carriers only. Immune-related transcripts were more strongly correlated with AD pathology in APOE4 carriers with some transcripts such as TREM2 and positively correlated with pathology severity in APOE4 carriers, but negatively in non-carriers. Overall, cumulative evidence from the largest neuroimaging, pathology, and transcriptomic studies available suggests that vascular dysfunction is key to the development of AD in APOE4 carriers. However, further studies are required to tease out non-APOE4-specific mechanisms.

Development and evaluation of a machine learning-based in-hospital COVID-19 disease outcome predictor (CODOP): A multicontinental retrospective study

New SARS-CoV-2 variants, breakthrough infections, waning immunity, and sub-optimal vaccination rates account for surges of hospitalizations and deaths. There is an urgent need for clinically valuable and generalizable triage tools assisting the allocation of hospital resources, particularly in resource-limited countries. We developed and validate CODOP, a machine learning-based tool for predicting the clinical outcome of hospitalized COVID-19 patients. CODOP was trained, tested and validated with six cohorts encompassing 29223 COVID-19 patients from more than 150 hospitals in Spain, the USA and Latin America during 2020-22. CODOP uses 12 clinical parameters commonly measured at hospital admission for reaching high discriminative ability up to 9 days before clinical resolution (AUROC: 0·90-0·96), it is well calibrated, and it enables an effective dynamic risk stratification during hospitalization. Furthermore, CODOP maintains its predictive ability independently of the virus variant and the vaccination status. To reckon with the fluctuating pressure levels in hospitals during the pandemic, we offer two online CODOP calculators, suited for undertriage or overtriage scenarios, validated with a cohort of patients from 42 hospitals in three Latin American countries (78-100% sensitivity and 89-97% specificity). The performance of CODOP in heterogeneous and geographically disperse patient cohorts and the easiness of use strongly suggest its clinical utility, particularly in resource-limited countries.

Derivation, Comprehensive Analysis, and Assay Validation of a Pyroptosis-Related lncRNA Prognostic Signature in Patients With Ovarian Cancer

Background

Pyroptosis is regulated by long non-coding RNAs (lncRNAs) in ovarian cancer (OC). Therefore, a comprehensive analysis of pyroptosis-related lncRNAs (PRLs) in OC is crucial for developing therapeutic strategies and survival prediction.

Methods

Based on public database raw data, mutations in the landscape of pyroptosis-related genes (PRGs) in patients with OC were investigated thoroughly. PRLs were identified by calculating Pearson correlation coefficients. Cox and LASSO regression analyses were performed on PRLs to screen for lncRNAs participating in the risk signature. Furthermore, receiver operating characteristic (ROC) curves, Kaplan-Meier survival analyses, decision curve analysis (DCA) curves, and calibration curves were used to confirm the clinical benefits. To assess the ability of the risk signature to independently predict prognosis, it was included in a Cox regression analysis with clinicopathological parameters. Two nomograms were constructed to facilitate clinical application. In addition, potential biological functions of the risk signature were investigated using gene function annotation. Subsequently, immune-related landscapes and BRCA1/2 mutations were compared in different risk groups using diverse bioinformatics algorithms. Finally, we conducted a meta-analysis and in-vitro assays on alternative lncRNAs.

Results

A total of 374 patients with OC were randomized into training and validation cohorts (7:3). A total of 250 PRLs were selected from all the lncRNAs. Subsequently, a risk signature (DICER1-AS1, MIR600HG, AC083880.1, AC109322.1, AC007991.4, IL6R-AS1, AL365361.1, and AC022098.2) was constructed to distinguish the risk of patient survival. The ROC curve, K-M analysis, DCA curve, and calibration curve indicated excellent predictive performance for determining overall survival (OS) based on the risk signature in each cohort (p < 0.05). The Cox regression analysis indicated that the risk signature was an independent prognostic factor for OS (p < 0.05). Moreover, significant differences in the immune response and BRCA1 mutations were identified in different groups distinguished by the risk signature (p < 0.05). Interestingly, in-vitro assays showed that an alternative lncRNA (DICER1-AS1) could promote OC cell proliferation.

Conclusion

The PRL risk signature could independently predict overall survival and guide treatment in patients with OC.

Machine learning-based dynamic prediction of lateral lymph node metastasis in patients with papillary thyroid cancer

OBJECTIVE

To develop a web-based machine learning server to predict lateral lymph node metastasis (LLNM) in papillary thyroid cancer (PTC) patients.

METHODS

Clinical data for PTC patients who underwent primary thyroidectomy at our hospital between January 2015 and December 2020, with pathologically confirmed presence or absence of any LLNM finding, were retrospectively reviewed. We built all models from a training set (80%) and assessed them in a test set (20%), using algorithms including decision tree, XGBoost, random forest, support vector machine, neural network, and K-nearest neighbor algorithm. Their performance was measured against a previously established nomogram using area under the receiver operating characteristic curve (AUC), decision curve analysis (DCA), precision, recall, accuracy, F1 score, specificity, and sensitivity. Interpretable machine learning was used for identifying potential relationships between variables and LLNM, and a web-based tool was created for use by clinicians.

RESULTS

A total of 1135 (62.53%) out of 1815 PTC patients enrolled in this study experienced LLNM episodes. In predicting LLNM, the best algorithm was random forest. In determining feature importance, the AUC reached 0.80, with an accuracy of 0.74, sensitivity of 0.89, and F1 score of 0.81. In addition, DCA showed that random forest held a higher clinical net benefit. Random forest identified tumor size, lymph node microcalcification, age, lymph node size, and tumor location as the most influentials in predicting LLNM. And the website tool is freely accessible at http://43.138.62.202/.

CONCLUSION

The results showed that machine learning can be used to enable accurate prediction for LLNM in PTC patients, and that the web tool allowed for LLNM risk assessment at the individual level.

Improved risk prediction of chemotherapy-induced neutropenia-model development and validation with real-world data

BACKGROUND

The existing risk prediction models for chemotherapy-induced febrile neutropenia (FN) do not necessarily apply to real-life patients in different healthcare systems and the external validation of these models are often lacking. Our study evaluates whether a machine learning-based risk prediction model could outperform the previously introduced models, especially when validated against real-world patient data from another institution not used for model training.

METHODS

Using Turku University Hospital electronic medical records, we identified all patients who received chemotherapy for non-hematological cancer between the years 2010 and 2017 (N = 5879). An experimental surrogate endpoint was first-cycle neutropenic infection (NI), defined as grade IV neutropenia with serum C-reactive protein >10 mg/l. For predicting the risk of NI, a penalized regression model (Lasso) was developed. The model was externally validated in an independent dataset (N = 4594) from Tampere University Hospital.

RESULTS

Lasso model accurately predicted NI risk with good accuracy (AUROC 0.84). In the validation cohort, the Lasso model outperformed two previously introduced, widely approved models, with AUROC 0.75. The variables selected by Lasso included granulocyte colony-stimulating factor (G-CSF) use, cancer type, pre-treatment neutrophil and thrombocyte count, intravenous treatment regimen, and the planned dose intensity. The same model predicted also FN, with AUROC 0.77, supporting the validity of NI as an endpoint.

CONCLUSIONS

Our study demonstrates that real-world NI risk prediction can be improved with machine learning and that every difference in patient or treatment characteristics can have a significant impact on model performance. Here we outline a novel, externally validated approach which may hold potential to facilitate more targeted use of G-CSFs in the future.