Individual Risk Prediction Research Articles

Machine Learning-Based CT Radiomics Model to Predict the Risk of Hip Fragility Fracture.

This research aimed to develop a combined model based on proximal femur attenuation values and radiomics features at routine CT to predict hip fragility fracture using machine learning methods. A total of 254 patients (training cohort, n=132; test cohort 1, n=56；test cohort 2, n=66) who underwent hip or pelvic CT scans were included. Three different machine learning methods were used to build the Support Vector Machine (SVM) model, Logistic Regression (LR) model and Random Forest (RF) model respectively. The method that exhibited the best performance in the training cohort and test cohort 1 was selected to represent the radiomics model for subsequent studies. The mean CT Hounsfield unit of three-dimensional CT images at the proximal femur was extracted to construct the mean CTHU model. Multivariate logistic regression was performed using mean CT Hounsfield unit together with radiomics features, and the combined model was subsequently developed with a visualized nomogram. Among the radiomics models based on three machine learning methods, the LR model showed the best performance in the training cohort (AUC=0.875, 95% CI=0.806-0.926) and in the test cohort 1 (AUC=0.851, 95% CI=0.730-0.932). Compared to the mean CT model and the LR model, the combined model showed superior discriminatory power in the training cohort (AUC=0.934, 95% CI=0.895-0.972), the test cohort 1 (AUC=0.893, 95% CI=0.812-0.974) and the test cohort 2 (AUC=0.851, 95% CI=0.742-0.927). The combined model, based on the mean CT Hounsfield unit of the proximal femur and radiomics features, can provide an accurate quantitative imaging basis for individualized risk prediction of hip fragility fracture.

Exploring Task-Related EEG for Cross-Subject Early Alzheimer's Disease Susceptibility Prediction in Middle-Aged Adults Using Multitaper Spectral Analysis.

The early prediction of Alzheimer's disease (AD) risk in healthy individuals remains a significant challenge. This study investigates the feasibility of task-state EEG signals for improving detection accuracy. Electroencephalogram (EEG) data were collected from the Multi-Source Interference Task (MSIT) and Sternberg Memory Task (STMT). Time-frequency features were extracted using the Multitaper method, followed by multidimensional reduction techniques. Subspace features (F24 and F216) were selected via t-tests and False Discovery Rate (FDR) multiple comparisons correction, and subsequently analyzed in the Time-Frequency Area Average Test (TFAAT) and Prefrontal Beta Time Series Test (PBTST). The experimental results reveal that the MSIT task achieves optimal cross-subject classification performance using the Support Vector Machine (SVM) approach with the TFAAT feature set, yielding a Receiver Operating Characteristic Area Under the Curve (ROC AUC) of 58%. Similarly, the Sternberg Memory Task demonstrates classification ability with the logistic regression model applied to the PBTST feature set, emphasizing the beta band power spectrum in the prefrontal cortex as a potential marker of AD risk. These findings confirm that task-state EEG provides stronger classification potential compared to resting-state EEG, offering valuable insights for advancing early AD prediction research.

Moving beyond population variable importance: concept, theory and applications of individual variable importance

Abstract In a non-parametric regression setting, we introduce a novel concept of ‘individual variable importance’, which assesses the relevance of certain covariates to an outcome variable among individuals with specific characteristics. This concept holds practical importance for both risk assessment and association identification. For example, it can represent (i) the usefulness of expensive biomarkers in risk prediction for individuals at a specified baseline risk, or (ii) age-specific associations between physiological indicators. We quantify individual variable importance using a ratio parameter between two conditional mean squared errors. To estimate and infer this parameter, we develop fully non-parametric estimators and establish their asymptotic properties. Our method performs well in simulation studies. Applying our approach to analyse a real dataset reveals a scientifically interesting result: the association between body shape and systolic blood pressure diminishes with increasing age. Our finding aligns with the medical literature based on standard parametric regression techniques, but our approach is more reliable due to its robustness to model misspecification. More importantly, the fully non-parametric nature of our method allows it to be applied in settings with complex relationships between variables, which cannot be correctly characterized by traditional parametric interaction analyses.

Systematic review of prognostic models for predicting recurrence and survival in patients with treated oropharyngeal cancer

ObjectivesThis systematic review aims to evaluate externally validated models for individualised prediction of recurrence or survival in adults treated with curative intent for oropharyngeal cancer.DesignSystematic review.SettingHospital care.MethodsSystematic searches were conducted...

Novel Variants in Medium and Low Penetrance Predisposing Genes in a Hungarian Malignant Melanoma Cohort With Increased Risk.

Both germline and somatic variants contribute to the genetic background and pathogenesis of melanoma. Germline variants include the presence of rare pathogenic or likely pathogenic variants of high, medium, and low penetrance melanoma-predisposing genes. Rare variants of high penetrance melanoma-predisposing genes are associated with melanoma development, whereas the medium and low penetrance predisposing genes can significantly increase melanoma risk. In this study, we clarified the germline genetic background of a Hungarian melanoma cohort (n = 17). Using a gene panel of 30 melanoma-predisposing genes, germline genetic variants were identified in 10 of the 17 patients (58.82%). A novel, likely pathogenic, missense variant (p.Y143C) in a medium penetrance melanoma-predisposing gene, melanocortin 1 receptor gene (MC1R), and two novel, likely pathogenic nonsense variants in low penetrance genes, p.Q218Ter in caspase 8 (CASP8) and p.Q40Ter in the fat mass- and obesity-associated (FTO) gene were detected. This study highlights the importance of elucidating the germline genetic background of melanoma, which may improve prediction of individual risk and the risk of family members and to optimize preventive, screening, and therapeutic measures for each patient and melanoma-prone families.

Establishment of prognostic risk model related to disulfidptosis and immune infiltration in hepatocellular carcinoma

BackgroundDisulfidptosis is a newly discovered type of cell death. We aim to identify hub genes associated with both disulfidptosis and immune infiltration in hepatocellular carcinoma (HCC) patients, and to develop an individualized risk prediction model. MethodsThe TCGA-LIHC cohort was utilized as the training set to identify molecular subtypes associated with disulfidptosis and to perform immune infiltration analysis. WGCNA, univariate Cox, and LASSO algorithm were employed to select hub genes for constructing the prognostic model. ICGC-LIRI cohort was utilized as an independent testing set. Validation of the expression of hub genes was performed in vitro using qRT-PCR and Western blot. ResultsCluster 1 was identified as the disulfidptosis associated molecular subtype, characterized by higher expression of disulfidptosis related genes (DRGs) and immune infiltration levels. ANXA2, MSC, and ST6GALNAC4 were identified as hub genes for calculating the risk score. The high-risk group were more likely to benefit from immunotherapy, targeted therapy and chemotherapy. A prognostic model was developed combining clinicopathological factors with satisfactory predictive accuracy. The hub genes were found upregulated in HCC cell line. ConclusionsOur findings provide valuable theoretical support for prognostic prediction and the evaluation of therapeutic outcomes in relation to disulfidptosis and immune infiltration in HCC, highlighting the importance of conducting in-depth research on disulfidptosis-related mechanisms.

Abstract 4134977: Utility of Coronary Artery Calcium Scoring in Low-Risk Patients: the Multi-Ethnic Study of Atherosclerosis (MESA)

Introduction: Coronary artery calcium (CAC) scoring is a powerful tool for atherosclerotic cardiovascular disease (ASCVD) risk assessment. Guidelines recommend consideration of CAC scoring in intermediate risk patients, but it’s utility in low and borderline risk patients is less clear. Methods: We used data from 2,894 participants from MESA with low (<5%) or borderline (5-<7.5%) 10-year ASCVD risk defined by the pooled cohort equations (PCE). We evaluated the association between CAC score and coronary heart disease (CHD) and ASCVD risk using multivariable adjusted Cox proportional hazard models. We also evaluated if the addition of CAC to the pooled cohort equations (PCE) improved risk prediction using Harrell’s C-index and net reclassification improvement (NRI). Results/Data: Mean age was 54.1 ± 5.8 years with 66.9% women. Among low-risk individuals, the ASCVD event rate was 3.3% with CAC=0, and 7.8% with CAC>0, and 4.7% vs. 11.4% among borderline-risk individuals. Natural-log transformed CAC score was associated with increased ASCVD risk in low-risk (HR 1.35, 95% CI 1.22-1.50) and borderline-risk (HR 1.29, 95% CI 1.15-1.45) individuals. Similar results were seen for CHD with higher HRs ( Table 1 ). The addition of CAC to the PCE improved the C-index (SE) among low/borderline risk individuals from 0.620 (0.023) to 0.671 (0.024). Continuous NRI was also significant with the addition of CAC to the PCE in low/borderline risk (0.422, 95% CI 0.248-0.611). Greater improvement was seen for CHD ( Table 2 ). Conclusions: The presence of CAC in low and borderline risk individuals is associated with increased ASCVD and CHD risk. The addition of CAC scoring to the PCE improved risk prediction in low and borderline risk individuals, suggesting potential clinical utility in this population.

Adding ethnicity to cardiovascular risk prediction: external validation and model updating of SCORE2 using data from the multiethnic HELIUS population cohort

Abstract Background Current prediction models for mainland Europe do not include ethnicity, despite ethnic disparities in cardiovascular disease (CVD) risk. Purpose We evaluated the performance of SCORE2 across the largest ethnic groups in the Netherlands with and without the factor of ethnicity. Methods We linked data from 11,614 participants enrolled between 2011 and 2015 in the prospective, multi-ethnic HELIUS cohort study, the Netherlands, aged between 40 and 70 years without CVD with hospital data. Variables (age, sex, smoking status, blood pressure, cholesterol) and outcome first CVD event (myocardial infarction, stroke, CVD death) were consistent with SCORE2. We used Fine and Gray models to calculate sub-distribution hazard ratios (SHR) for South-Asian Surinamese, African Surinamese, Ghanaian, Turkish and Moroccan origin groups, relative to the Dutch origin group, on top of individual SCORE2 risk predictions. Model performance was evaluated by discrimination, calibration and net reclassification index (NRI). Results Overall, 274 fatal and non-fatal CVD events, and 146 non-cardiovascular deaths were observed during a median of 7.8 years follow-up (IQR 6.8–8.8). SHRs for CVD events were 1.86 (95% CI 1.31-2.65, p=0.0005) for the South-Asian Surinamese, 1.09 (95% CI 0.76-1.56) for the African-Surinamese, 1.48 (95% CI 0.94-2.31) for the Ghanaian, 1.63 (95% CI 1.09-2.44, p=0.018) for the Turkish, and 0.67 (95% CI 0.39-1.18) for the Moroccan origin groups. Adding ethnicity to SCORE2 led to comparable calibration and discrimination [0.76 (0.74-0.79) vs. 0.77 (95% CI 0.74-0.78)]. The NRI for adding ethnicity to SCORE2 was 0.24 (95% CI 0.18-0.31) for events and -0.12 (95% CI -0.13--0.12) for non-events. Conclusion Adding ethnic backgrounds improves risk classification of the SCORE2 risk prediction model in a middle-aged, multi-ethnic Dutch population, which may help to address disparities in CVDs through timely treatment.

Intraindividual variability in lipoprotein(a): the value of a repeat measure for reclassifying individuals at intermediate-risk

Abstract Background Lipoprotein(a) [Lp(a)] is an independent risk factor for cardiovascular disease and outcomes. Levels are predominantly genetically determined and repeat measurements are traditionally considered unlikely to be useful. However, the temporal variation of Lp(a) is not well characterized and it is unclear whether a single measure of Lp(a) is sufficient to assess cardiovascular risk in all individuals. Purpose To determine the intraindividual variability of Lp(a) and whether a repeated measure reclassified Lp(a)-specific cardiovascular risk using the European Atherosclerosis Society (EAS) consensus statement risk categories. Methods This retrospective cohort study analyzed initial and repeated Lp(a) measures obtained using the same laboratory assay from 609 individuals in the Nashville Biosciences database accessed through the American Heart Association Precision Medicine Platform. Categorical variables are presented as counts (%) and compared using Fisher’s exact test. Continuous variables are presented as mean ± standard deviation or median (Q1-Q3), tested for normality using the Shapiro-Wilk test, and compared using parametric and nonparametric tests where appropriate. The Spearman’s Rank correlation coefficient was used to assess the relationship between two variables. The significance level was set at p<0.05. Results Median age was 52.5 (37.1-62.8) years, 45% were women, 80.3% were White patients, 13.3% were Black patients, 1.6% were Asian patients, and 4.8% were other/unknown patients. In terms of relevant comorbidities, 41% had ischemic heart disease, 24.1% had chronic kidney disease, and 13.5% had liver disease. The median time between the two Lp(a) measures was 1.07 (0.5-2.1) years. Our analysis shows a significant difference between the paired Lp(a) values (p<0.05), with 38.1% [95% CI 34.2%-42%] changing by ≥10mg/dL, and 40.5% [95%CI 36.6%-44.3%] changing by >25%. There was a positive correlation between the baseline Lp(a) levels and the absolute changes in Lp(a), (r=0.59, p<0.01). The proportions of individuals with variability are similar in those who were never on the potentially Lp(a)-altering agents (statin, niacin, or PCSK9 inhibitor drugs) (n=91), with 32% [95%CI 22.4%-41.6%] changing by ≥10mg/dL, and 34% [95%CI 24.3%-43.7%] changing by >25%. While over 90% of those whose values were initially in the EAS low-risk and high-risk categories remained in those categories, 53% of those in the intermediate "grey-zone" category transitioned to either the low (20%)- or the high (33%)- risk category (Figure). Conclusion(s) Temporal-related changes in Lp(a) variability were present in many individuals and should be included in calculating the expected effect sizes in future studies targeting Lp(a). A repeat Lp(a) measure may allow more precise Lp(a)-specific cardiovascular risk prediction for individuals whose initial value is in the EAS-defined intermediate "grey-zone" category.

Risk stratification model for foreseeing overall survival in Chinese patients with initially metastatic small-cell lung cancer.

The study was outlined to develop and approve a nomogram and chance stratification demonstrate for foreseeing overall survival of Chinese patients with at first metastatic small-cell lung cancer (SCLC). We collected information from the Surveillance, Epidemiology, and End Results (SEER) database approximately Chinese SCLC patients with at first distant metastases between 2010 and 2015. Patients with incomplete data about the follow-up time or clinicopathological information were excluded. The included patients were randomized into the training and validation set. Univariate and multivariate Cox proportional hazard regression models were performed. By integrating the significant variables screened, a prescient nomogram and risk stratification model were developed. In addition, we collected 198 small-cell lung cancer patients with metastasis at diagnosis from the case database of the Affiliated Hospital of Shandong University of Traditional Chinese Medicine as an external validation cohort. In all, 421 patients were screened from the SEER database. Multivariate examination showed that age (P = .049), sex (P = .001), grade (P = .008), chemotherapy (P = .001), liver metastasis (P = .001), and pleural invasion (P = .012) were independent prognostic factors. The C-indicator of the nomogram to anticipate overall survival was higher than that of the eighth edition of the American Joint Committee on Cancer Tumor Node Metastasis classification system (0.75 vs 0.543; P < .001). A risk stratification model was encouraged to be created to precisely classify patients into 2 prognostic bunches. The survival rates anticipated by the nomogram appeared to have critical qualifications from the Kaplan-Meier curves in the entire SEER cohort. Calibration curves and survival predictions also showed strong accuracy and consistency in the external validation cohort. The nomogram provided a clear prognostic superiority over the traditional Tumor Node Metastasis system. It could help clinicians make individual risk predictions for initially metastatic Chinese SCLC cancer patients and give necessary treatment recommendations.

Population Performance and Individual Agreement of Coronary Artery Disease Polygenic Risk Scores.

Polygenic risk scores (PRSs) for coronary artery disease (CAD) are a growing clinical and commercial reality. Whether existing scores provide similar individual-level assessments of disease liability is a critical consideration for clinical implementation that remains uncharacterized. Characterize the reliability of CAD PRSs that perform equivalently at the population level at predicting individual-level risk. Cross-sectional Study. All of Us Research Program (AOU), Penn Medicine Biobank (PMBB), and UCLA ATLAS Precision Health Biobank. Volunteers of diverse genetic backgrounds enrolled in AOU, PMBB, and UCLA with available electronic health record and genotyping data. Polygenic risk for CAD from previously published PRSs and new PRSs developed separately from the testing cohorts. Sets of CAD PRSs that perform population prediction equivalently were identified by comparing calibration and discrimination (Brier score and AUROC) of generalized linear models of prevalent CAD using Bayesian analysis of variance. Among equivalently performing scores, individual-level agreement between risk estimates was tested with intraclass correlation (ICC) and Light's Kappa, measures of inter-rater reliability. 50 PRSs were calculated for 171,095 AOU participants. When included in a model of prevalent CAD, 48 scores had practically equivalent Brier scores and AUROCs (region of practical equivalence = 0.02). Across these scores, 84% of participants had at least one score in both the top and bottom risk quintile. Continuous agreement of individual risk predictions from the 48 scores was poor, with an ICC of 0.351 (95% CI; 0.349, 0.352). Agreement between two statistically equivalent scores was moderate, with an ICC of 0.649 (95% CI; 0.646, 0.652). Light's Kappa, used to evaluate consistency of assignment to high-risk thresholds, did not exceed 0.56 (interpreted as 'fair') across statistically and practically equivalent scores. Repeating the analysis among 41,193 PMBB and 50,748 UCLA participants yielded different sets of statistically and practically equivalent scores which also lacked strong individual agreement. Across three diverse biobanks, CAD PRSs that performed equivalently at the population level produced unreliable individual risk estimates. Approaches to clinical implementation of CAD PRSs must consider the potential for discordant individual risk estimates from otherwise indistinguishable scores.

TdCoxSNN: Time-dependent Cox survival neural network for continuous-time dynamic prediction

Abstract The aim of dynamic prediction is to provide individualized risk predictions over time, which are updated as new data become available. In pursuit of constructing a dynamic prediction model for a progressive eye disorder, age-related macular degeneration (AMD), we propose a time-dependent Cox survival neural network (tdCoxSNN) to predict its progression using longitudinal fundus images. tdCoxSNN builds upon the time-dependent Cox model by utilizing a neural network to capture the nonlinear effect of time-dependent covariates on the survival outcome. Moreover, by concurrently integrating a convolutional neural network with the survival network, tdCoxSNN can directly take longitudinal images as input. We evaluate and compare our proposed method with joint modelling and landmarking approaches through extensive simulations. We applied the proposed approach to two real datasets. One is a large AMD study, the Age-Related Eye Disease Study, in which more than 50,000 fundus images were captured over a period of 12 years for more than 4,000 participants. Another is a public dataset of the primary biliary cirrhosis disease, where multiple laboratory tests were longitudinally collected to predict the time-to-liver transplant. Our approach demonstrates commendable predictive performance in both simulation studies and the analysis of the two real datasets.

Remnant cholesterol is superior to other lipid-related parameters for the prediction of cardiometabolic disease risk in individuals with hypertension: The Kailuan study

BackgroundIt is uncertain which lipid-related parameter is most suitable for predicting the risk of cardiometabolic disease (CMD) in individuals with hypertension. AimsTo explore which lipid-related parameter is most suitable for predicting the risk of CMD. Methods and resultsWe studied 30,378 patients with hypertension who completed the 2006–2007 Kailuan health examination and followed up until December 31, 2021. In the constructed model, the utilities of lipid-related parameters for the prediction of CMD were compared using the C-index, NRI, and IDI. The best predictor (remnant cholesterol, RC) was identified and the participants were grouped according to RC quartile. Cox proportional hazard analysis was then used to evaluate the relationship between RC and the risk of CMD. During a median follow-up period of 14.7 years (IQR 5.3–15.1), 9502 (31.27 %) participants with hypertension developed CMD. The C-index, NRI, and IDI values for RC were higher than those for the other lipid parameters. After adjustment for multiple potential confounding factors, compared with the quartile (Q)1 RC group, the adjusted hazard ratios for CMD of the Q2–Q4 groups were 1.09 (1.03–1.16), 1.17 (1.11–1.24), and 1.25 (1.18–1.33) (P < 0.0001). Restrictive cubic spline analysis revealed dose-dependent relationships of lipid parameters with the risk of CMD. ConclusionsRC is superior to other lipid parameters for the prediction of the risk of CMD in individuals with hypertension. As the concentration of RC increases, the risk of CMD in such individuals also increases.

Fracture risk prediction in diabetes patients based on Lasso feature selection and Machine Learning

Fracture risk among individuals with diabetes poses significant clinical challenges due to the multifaceted relationship between diabetes and bone health. Diabetes not only affects bone density but also alters bone quality and structure, thereby increases the susceptibility to fractures. Given the rising prevalence of diabetes worldwide and its associated complications, accurate prediction of fracture risk in diabetic individuals has emerged as a pressing clinical need. This study aims to investigate the factors influencing fracture risk among diabetic patients. We propose a framework that combines Lasso feature selection with eight classification algorithms. Initially, Lasso regression is employed to select 24 significant features. Subsequently, we utilize grid search and 5-fold cross-validation to train and tune the selected classification algorithms, including KNN, Naive Bayes, Decision Tree, Random Forest, AdaBoost, XGBoost, Multi-layer Perceptron (MLP), and Support Vector Machine (SVM). Among models trained using these important features, Random Forest exhibits the highest performance with a predictive accuracy of 93.87%. Comparative analysis across all features, important features, and remaining features demonstrate the crucial role of features selected by Lasso regression in predicting fracture risk among diabetic patients. Besides, by using a feature importance ranking algorithm, we find several features that hold significant reference values for predicting early bone fracture risk in diabetic individuals.

Polygenic Risk Scores: The Next Step for Improved Risk Stratification in Coronary Artery Disease?

Despite significant advances in the management of coronary artery disease (CAD) and reductions in annual mortality rates in recent decades, this disease remains the leading cause of death worldwide. Consequently, there is an ongoing need for efforts to address this situation. Current clinical algorithms to identify at-risk patients are particularly inaccurate in moderate-risk individuals. For this reason, the need for ancillary tests has been suggested, including predictive genetic screening. As genetic studies rapidly expand and genomic data becomes more accessible, numerous genetic risk scores have been proposed to identify and evaluate an individual's susceptibility to developing diseases, including CAD. The field of genetics has indeed made substantial contributions to risk prediction, particularly in cases where children have parents with premature CAD, resulting in an increased risk of up to 75%. The polygenic risk scores (PRSs) have emerged as a potentially valuable tool for understanding and stratifying an individual's genetic risk. The PRS is calculated as a weighted sum of single-nucleotide variants present throughout the human genome, identifiable through genome-wide association studies, and associated with various cardiometabolic diseases. The use of PRSs holds promise, as it enables the development of personalized strategies for preventing or diagnosing specific pathologies early. Furthermore, it can complement existing clinical scores, increasing the accuracy of individual risk prediction. Consequently, the application of PRSs has the potential to impact the costs and adverse outcomes associated with CAD positively. This narrative review provides an overview of the role of PRSs in the context of CAD.

Intra-individual variability in lipoprotein(a): the value of a repeat measure for reclassifying individuals at intermediate risk.

Lipoprotein(a) [Lp(a)] levels are predominantly genetically determined and repeat measurements are generally considered unlikely to be clinically useful. However, the temporal variation of Lp(a) is not well characterized. Our aim was to determine the intra-individual variability of Lp(a) and whether a repeated measure reclassified Lp(a)-specific cardiovascular risk using the European Atherosclerosis Society (EAS) consensus statement risk categories. This retrospective cohort study analysed initial and repeated serum Lp(a) levels measured using the same methodology from 609 individuals in the Nashville Biosciences database, a de-identified electronic medical records database. Baseline and follow-up paired values were significantly different (P < 0.05), with an absolute change of ≥10 mg/dL in 38.1% [95% CI 34.2-42%] and a >25% change in 40.5% [95% CI 36.6-44.3%] of individuals. Although the categories of those whose values were in the EAS low-risk and high-risk categories did not change, 53% of those in the intermediate 'grey-zone' category transitioned to either the low-risk (20%) or high-risk (33%) category. Black individuals exhibited greater variability than White individuals and women exhibited greater variability than men. There was a positive correlation between the baseline Lp(a) levels and the absolute changes in Lp(a), (r = 0.59, P < 0.01). Temporal-related changes in Lp(a) variability were present in many individuals. A repeat Lp(a) measure may allow more precise Lp(a)-specific cardiovascular risk prediction for individuals whose initial value is in the EAS-defined intermediate 'grey-zone' category. Lp(a) variability should be included in calculating the expected effect sizes in future clinical research studies targeting Lp(a).

Integrative genetic analysis: cornerstone of precision psychiatry.

The role of genetic testing in the domain of neurodevelopmental and psychiatric disorders (NPDs) is gradually changing from providing etiological explanation for the presence of NPD phenotypes to also identifying young individuals at high risk of developing NPDs before their clinical manifestation. In clinical practice, the latter implies a shift towards the availability of individual genetic information predicting a certain liability to develop an NPD (e.g., autism, intellectual disability, psychosis etc.). The shift from mostly a posteriori explanation to increasingly a priori risk prediction is the by-product of the systematic implementation of whole exome or genome sequencing as part of routine diagnostic work-ups during the neonatal and prenatal periods. This rapid uptake of genetic testing early in development has far-reaching consequences for psychiatry: Whereas until recently individuals would come to medical attention because of signs of abnormal developmental and/or behavioral symptoms, increasingly, individuals are presented based on genetic liability for NPD outcomes before NPD symptoms emerge. This novel clinical scenario, while challenging, also creates opportunities for research on prevention interventions and precision medicine approaches. Here, we review why optimization of individual risk prediction is a key prerequisite for precision medicine in the sphere of NPDs, as well as the technological and statistical methods required to achieve this ambition.

A Vision on User-Centered Implementation and Evaluation of Explainable AI for Predicting Hospital-Onset Bacteremia.

In recent years, artificial intelligence (AI) has gained momentum in many fields of daily live. In healthcare, AI can be used for diagnosing or predicting illnesses. However, explainable AI (XAI) is needed to ensure that users understand how the algorithm arrives at a decision. In our research project, machine learning methods are used for individual risk prediction of hospital-onset bacteremia (HOB). This paper presents a vision on a step-wise process for implementation and evaluation of user-centered XAI for risk prediction of HOB. An initial requirement analysis revealed first insights on the users' needs of explainability to use and trust such risk prediction applications. The findings were then used to propose step-wise process towards a user-centered evaluation.

Utilizing machine learning to analyze trunk movement patterns in women with postpartum low back pain

This paper presents an analysis of trunk movement in women with postnatal low back pain using machine learning techniques. The study aims to identify the most important features related to low back pain and to develop accurate models for predicting low back pain. Machine learning approaches showed promise for analyzing biomechanical factors related to postnatal low back pain (LBP). This study applied regression and classification algorithms to the trunk movement proposed dataset from 100 postpartum women, 50 with LBP and 50 without. The Optimized optuna Regressor achieved the best regression performance with a mean squared error (MSE) of 0.000273, mean absolute error (MAE) of 0.0039, and R2 score of 0.9968. In classification, the Basic CNN and Random Forest Classifier both attained near-perfect accuracy of 1.0, the area under the receiver operating characteristic curve (AUC) of 1.0, precision of 1.0, recall of 1.0, and F1-score of 1.0, outperforming other models. Key predictive features included pain (correlation of -0.732 with flexion range of motion), range of motion measures (flexion and extension correlation of 0.662), and average movements (correlation of 0.957 with flexion). Feature selection consistently identified pain, flexion, extension, lateral flexion, and average movement as influential across methods. While limited to this initial dataset and constrained by generalizability, machine learning offered quantitative insight. Models accurately regressed (MSE < 0.01, R2 > 0.95) and classified (accuracy > 0.94) trunk biomechanics distinguishing LBP. Incorporating additional demographic, clinical, and patient-reported factors may enhance individualized risk prediction and treatment personalization. This preliminary application of advanced analytics supported machine learning's potential utility for both LBP risk determination and outcome improvement. This study provides valuable insights into the use of machine learning techniques for analyzing trunk movement in women with postnatal low back pain and can potentially inform the development of more effective treatments.Trial registration: The trial was designed as an observational and cross-section study. The study was approved by the Ethical Committee in Deraya University, Faculty of Pharmacy, (No: 10/2023). According to the ethical standards of the Declaration of Helsinki. This study complies with the principles of human research. Each patient signed a written consent form after being given a thorough description of the trial. The study was conducted at the outpatient clinic from February 2023 till June 30, 2023.

Individual suicide risk factors with resting-state brain functional connectivity patterns in bipolar disorder patients based on latent Dirichlet allocation model

BackgroundThe widespread problem of suicide and its severe burden in bipolar disorder (BD) necessitate the development of objective risk markers, aiming to enhance individual suicide risk prediction in BD. MethodsThis study recruited 123 BD patients (61 patients with prior suicide attempted history (PSAs), 62 without (NSAs)) and 68 healthy controls (HEs). The Latent Dirichlet Allocation (LDA) model was used to decompose the resting state functional connectivity (RSFC) into multiple hyper/hypo-RSFC patterns. Thereafter, according to the quantitative results of individual heterogeneity over latent factor dimensions, the correlations were analyzed to test prediction ability. ResultsModel constructed without introducing suicide-related labels yielded three latent factors with dissociable hyper/hypo-RSFC patterns. In the subsequent analysis, significant differences in the factor distributions of PSAs and NSAs showed biases on the default-mode network (DMN) hyper-RSFC factor (factor 3) and the salience network (SN) and central executive network (CEN) hyper-RSFC factor (factor 1), indicating predictive value. Correlation analysis of the individuals' expressions with their Nurses' Global Assessment of Suicide Risk (NGASR) revealed factor 3 positively correlated (r = 0.4180, p < 0.0001) and factor 1 negatively correlated (r = − 0.2492, p = 0.0055) with suicide risk. Therefore, it could be speculated that patterns more associated with suicide reflected hyper-connectivity in DMN and hypo-connectivity in SN, CEN. ConclusionsThis study provided individual suicide-associated risk factors that could reflect the abnormal RSFC patterns, and explored the suicide related brain mechanisms, which is expected to provide supports for clinical decision-making and timely screening and intervention for individuals at high risks of suicide.