Using machine learning approach to predict depression and anxiety among patients with epilepsy in China: A cross-sectional study

Federated learning-based prediction of depression among adolescents across multiple districts in China

Depression in adolescents is a serious mental health condition that can affect their emotional and social well-being. Detailed understanding of depression patterns and status of depressive symptoms in adolescents could help identify early intervention targets. Despite the growing use of artificial intelligence for diagnosis and prediction of mental health conditions, the traditional centralized machine learning methods require aggregating adolescents' data; this raises concerns about confidentiality and privacy, which hampers the clinical application of machine learning algorithms. In this study, we use federated learning to solve those problems. We included 583,405 middle and high school adolescents from 20 districts in Chengdu China, and collected from three aspects: individuals, families, and schools, containing 11 psychological phenomena to evaluate the status of depressive symptoms. We compared federated and local training frameworks; the results showed the area under the receiver operating characteristic curve for depression increased by up to 20 % (from 0.7544 with local training to 0.9064 with federated training). Moreover, based on the best-performing model, the XGBoost model, we explore the data heterogeneity in federated learning and found that stress, student burnout, and social connection were the three most important predictors of depression symptoms. We then assessed the impact of each subdimension of depression symptoms, results show that sleep was the most impact one which may provide clues to predict depression symptoms in early stage and improve control and prevention efforts.

Identification of a novel disulfidptosis-related gene signature in osteoarthritis using bioinformatics analysis and experimental validation

Osteoarthritis (OA) is a degenerative bone disease characterized by the destruction of joint cartilage and synovial inflammation, involving intricate immune regulation processes. Disulfidptosis, a novel form of programmed cell death, has recently been identified; however, the effects and roles of disulfidptosis-related genes (DR-DEGs) in OA remain unclear. We obtained six OA datasets from the GEO database, using four as training sets and two as validation sets. Differential expression analysis was employed to identify DR-DEGs, and unique molecular subtypes of OA were constructed based on these DR-DEGs. Subsequently, the immune microenvironment of OA patients was comprehensively analyzed using the "CIBERSORT" algorithm for immune infiltration. Various machine learning algorithms were utilized to screen characteristic DR-DEGs, and nomogram models and ROC curves were built based on these genes. The scRNA dataset (GSE169454) was used to classify chondrocytes in OA samples into distinct cell types, further exploring the gene distribution and correlation of characteristic DR-DEGs with specific cell subpopulations. Moreover, the expression levels of four characteristic DR-DEGs were validated through OA cell models and rat models. In our study, we identified 10 DR-DEGs with significant differences in expression within OA samples. Based on these DR-DEGs, two distinct molecular subtypes were recognized (cluster 1 and 2). ZNF484 and NDUFS1 were found to be significantly overexpressed in subtype 1, while the infiltration abundance of activated mast cells was markedly elevated in subtype 2. Moreover, significant differences were observed in the infiltration proportions of 11 immune cell types between OA and control samples, with 9 DR-DEGs demonstrating substantial correlations with immune cell infiltration levels. Further analysis of the scRNA dataset revealed that SLC3A2 and NDUFC1 were predominantly expressed in the preHTC subpopulation. All 10 DR-DEGs exhibited notably higher expression in the EC subpopulation across various cell types. The proportion of EC subgroups with high SLC3A2 expression increased, mainly enriching pathways related to inflammation, such as the IL-17 signaling pathway and TGF-beta signaling pathway. Using machine learning, we identified four characteristic DR-DEGs, which, in combination with the nomogram and ROC models, demonstrated promising performance in the diagnosis of OA. Additionally, in vivo validation confirmed a significant elevation of PPM1F expression in OA models. This study identified DR-DEGs as potential biomarkers for the diagnosis and classification of OA and provided a preliminary understanding of their role in the immune microenvironment. However, further experimental and clinical studies are required to validate their diagnostic value and therapeutic potential.

Machine learning algorithms to predict depression in older adults in China: a cross-sectional study

Objective

The 2-fold objective of this research is to investigate machine learning's (ML) predictive value for the incidence of depression among China's older adult population and to determine the noteworthy aspects resulting in depression.

Methods

This research selected 7,880 older adult people by utilizing data from the 2020 China Health and Retirement Longitudinal Study. Thereafter, the dataset was classified into training and testing sets at a 6:4 ratio. Six ML algorithms, namely, logistic regression, k-nearest neighbors, support vector machine, decision tree, LightGBM, and random forest, were used in constructing a predictive model for depression among the older adult. To compare the differences in the ROC curves of the different models, the Delong test was conducted. Meanwhile, to evaluate the models' performance, this research performed decision curve analysis (DCA). Thereafter, the Shapely Additive exPlanations values were utilized for model interpretation on the bases of the prediction results' substantial contributions.

Results

The range of the area under the curve (AUC) of each model's ROC curves was 0.648-0.738, with significant differences (P < 0.01). The DCA results indicate that within various probability thresholds, LightGBM's net benefit was the highest. Self-rated health, nighttime sleep, gender, age, and cognitive function are the five most important characteristics of all models in terms of predicting the occurrence of depression.

Conclusion

The occurrence of depression among China's older adult population and the critical factors leading to depression can be predicted and identified, respectively, by ML algorithms.

Machine learning algorithms to predict mild cognitive impairment in older adults in China: A cross-sectional study

OBJECTIVE

This study aimed to explore the predictive value of machine learning (ML) in mild cognitive impairment (MCI) among older adults in China and to identify important factors causing MCI.

METHODS

In this study, 6434 older adults were selected based on the data of the China Health and Elderly Care Longitudinal Survey (CHARLS) in 2020, and the dataset was subsequently divided into the training set and the test set, with a ratio of 6:4. To construct a prediction model for MCI in older adults, six ML algorithms were used, including logistic regression, KNN, SVM, decision tree (DT), LightGBM, and random forest (RF). The Delong test was used to compare the differences of ROC curves of different models, while decision curve analysis (DCA) was used to evaluate the model performance. The important contributions of the prediction results were then used to explain the model by the SHAP value.The Matthews correlation coefficient (MCC) was calculated to evaluate the performance of the models on imbalanced datasets. Additionally, causal analysis and counterfactual analysis were conducted to understand the feature importance and variable effects.

RESULTS

The area under the ROC curve of each model range from 0.71 to 0.77, indicating significant difference (P < 0.01). The DCA results show that the net benefits of LightGBM is the largest within various probability thresholds. Among all the models, the LightGBM model demonstrated the highest performance and stability. The five most important characteristics for predicting MCI were educational level, social events, gender, relationship with children, and age. Causal analysis revealed that these variables had a significant impact on MCI, with an average treatment effect of -0.144. Counterfactual analysis further validated these findings by simulating different scenarios, such as improving educational level, increasing age, and increasing social events.

CONCLUSION

The ML algorithm can effectively predict the MCI of older adults in China and identify the important factors causing MCI.

Evaluating predictive artificial intelligence approaches used in mobile health platforms to forecast mental health symptoms among youth: a systematic review

The youth mental health crisis is exacerbated by limited access to care and resources. Mobile health (mHealth) platforms using predictive artificial intelligence (AI) can improve access and reduce barriers, enabling real-time responses and precision prevention. This systematic review evaluates predictive AI approaches in mHealth platforms for forecasting mental health symptoms among youth (13-25 years). We searched studies from Embase, PubMed, Web of Science, PsycInfo, and CENTRAL, to identify relevant studies. From 11 studies identified, three studies predicted multiple symptoms, with depression being the most common (63%). Most platforms used smartphones and 25% integrated wearables. Key predictors included smartphone usage (N=5), sleep metrics (N=6), and physical activity (N=5). Nuanced predictors like usage locations and sleep stages improved prediction. Logistic regression was most used (N=6), followed by Support Vector Machines (N=3) and ensemble methods (N=4). F-scores for anxiety and depression ranged from 0.73 to 0.84, and AUCs from 0.50 to 0.74. Stress models had AUCs of 0.68 to 0.83. Bayesian model selection and Shapley values enhanced robustness and interpretability. Barriers included small sample sizes, privacy concerns, missing data, and underrepresentation bias. Rigorous evaluation of predictive performance, generalizability, and user engagement is critical before mHealth platforms are integrated into psychiatric care.

Using machine learning models to identify the risk of depression in middle-aged and older adults with frequent and infrequent nicotine use: A cross-sectional study

BACKGROUND

Depression is very prevalent in middle-aged and older smokers. Therefore, we aimed to identify the risk of depression among middle-aged and older adults with frequent and infrequent nicotine use, as this is quite necessary for supporting their well-being.

METHODS

This study included a total of 10,821 participants, which were derived from the China Health and Retirement Longitudinal Study Wave 5, 2020 (CHARLS-5). Five machine learning (ML) algorithms were employed. Some metrics were used to evaluate the performance of models, including area under the receiver operating characteristic curve (AUC), positive predictive value (PPV), specificity, accuracy.

RESULTS

10,821 participants (6472 males, 4349 females) had a mean age of 60.47 ± 8.98, with a score of 8.90 ± 6.53 on depression scale. For middle-aged and older adults with frequent nicotine use, random forest (RF) achieved the highest AUC value, PPV and specificity (0.75, 0.74 and 0.88, respectively). For the other group, support vector machines (SVM) showed the highest PPV (0.74), and relatively high accuracy and specificity (0.72 and 0.87, respectively). Feature importance analysis indicated that "dissatisfaction with life" was the most important variable of identifying the risk of depression in the SVM model, while "attitude towards expected life span" was the most important one in the RF model.

LIMITATIONS

CHARLS-5 was collected during the COVID-19, so our results may be influenced by the pandemic.

CONCLUSIONS

This study indicated that certain ML models can ideally identify the risk of depression in middle-aged and older adults, which holds significant value for their health management.

Accuracy of Machine Learning in Detecting Pediatric Epileptic Seizures: Systematic Review and Meta-Analysis

BACKGROUND

Real-time monitoring of pediatric epileptic seizures poses a significant challenge in clinical practice. In recent years, machine learning (ML) has attracted substantial attention from researchers for diagnosing and treating neurological diseases, leading to its application for detecting pediatric epileptic seizures. However, systematic evidence substantiating its feasibility remains limited.

OBJECTIVE

This systematic review aimed to consolidate the existing evidence regarding the effectiveness of ML in monitoring pediatric epileptic seizures with an effort to provide an evidence-based foundation for the development and enhancement of intelligent tools in the future.

METHODS

We conducted a systematic search of the PubMed, Cochrane, Embase, and Web of Science databases for original studies focused on the detection of pediatric epileptic seizures using ML, with a cutoff date of August 27, 2023. The risk of bias in eligible studies was assessed using the QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies-2). Meta-analyses were performed to evaluate the C-index and the diagnostic 4-grid table, using a bivariate mixed-effects model for the latter. We also examined publication bias for the C-index by using funnel plots and the Egger test.

RESULTS

This systematic review included 28 original studies, with 15 studies on ML and 13 on deep learning (DL). All these models were based on electroencephalography data of children. The pooled C-index, sensitivity, specificity, and accuracy of ML in the training set were 0.76 (95% CI 0.69-0.82), 0.77 (95% CI 0.73-0.80), 0.74 (95% CI 0.70-0.77), and 0.75 (95% CI 0.72-0.77), respectively. In the validation set, the pooled C-index, sensitivity, specificity, and accuracy of ML were 0.73 (95% CI 0.67-0.79), 0.88 (95% CI 0.83-0.91), 0.83 (95% CI 0.71-0.90), and 0.78 (95% CI 0.73-0.82), respectively. Meanwhile, the pooled C-index of DL in the validation set was 0.91 (95% CI 0.88-0.94), with sensitivity, specificity, and accuracy being 0.89 (95% CI 0.85-0.91), 0.91 (95% CI 0.88-0.93), and 0.89 (95% CI 0.86-0.92), respectively.

CONCLUSIONS

Our systematic review demonstrates promising accuracy of artificial intelligence methods in epilepsy detection. DL appears to offer higher detection accuracy than ML. These findings support the development of DL-based early-warning tools in future research.

TRIAL REGISTRATION

PROSPERO CRD42023467260; https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023467260.

An explainable predictive model for anxiety symptoms risk among Chinese older adults with abdominal obesity using a machine learning and SHapley Additive exPlanations approach

Background

Early detection of anxiety symptoms can support early intervention and may help reduce the burden of disease in later life in the elderly with abdominal obesity, thereby increasing the chances of healthy aging. The objective of this research is to formulate and validate a predictive model that forecasts the probability of developing anxiety symptoms in elderly Chinese individuals with abdominal obesity.

Method

This research's model development and internal validation encompassed 2,427 participants from the 2017-2018 Study of the Chinese Longitudinal Healthy Longevity Survey (CLHLS). Forty-six variables were defined based on the Health Ecology Model (HEM) theoretical framework. Key variables were screened using LASSO regression, and the XGBoost (Extreme Gradient Boosting) model was further introduced to forecast the risk of developing anxiety symptoms in the elderly with abdominal obesity. SHapley Additive exPlanations (SHAP) was adopted to further interpret and show how the eigenvalues contributed to the model predictions.

Results

A total of 240 participants (9.89%) with anxiety symptoms out of 2,427 participants were included. LASSO regression identified nine key variables: looking on the bright side, self-reported economic status, self-reported quality of life, self-reported health status, watching TV or listening to the radio, feeling energetic, feeling ashamed/regretful/guilty, feeling angry, and fresh fruits. All the evaluation indicators of the XGBoost model showed good predictive efficacy. Based on the significance of the features identified by SHAP (Model Interpretation Methodology), the feature 'looking on the bright side' was the most important, and the feature 'self-reported quality of life' was the least important. The SHAP beeswarm plot illustrated the impacts of features affected by XGBoost.

Conclusion

Utilizing machine learning techniques, our predictive model can precisely evaluate the risk of anxiety symptoms among elderly individuals with abdominal obesity, facilitating the timely adoption of targeted intervention measures. The integration of XGBoost and SHAP offers transparent interpretations for customized risk forecasts.

Felt stigma and associated factors in children and adolescents with epilepsy: a multicenter cross-sectional study in China

Objective

Epilepsy-related stigma is a global problem, yet there has been an inadequate focus on children and adolescents. The purpose of this study was to determine the status quo of stigma and its determinants among children and adolescents with epilepsy in China.

Methods

A multicenter cross-sectional study was conducted across nine hospitals in eight cities within six provinces in China from 10 October 2023 to 15 June 2024. Participants included patients aged 8 to 18 years with epilepsy and their caregivers. Felt stigma was assessed with the Kilifi Stigma Scale for Epilepsy (KSSE). Social support and self-efficacy were collected through the Social Support Rating Scale (SSRS) and the Generalized Self-Efficacy Scale (GSES). The data were analyzed using t-tests, analysis of variance (ANOVA), Spearman correlation analysis, and multiple linear regression analysis.

Results

The study enrolled 281 children and adolescents, with a mean age of 12.25 years (SD = 2.56), including 46.6% females. A total of 35.6% participants had self-reported felt stigma. The mean KSSE score is 9.58 (SD = 7.11). Meanwhile, stigma scores correlated strongly with reduced social support (r = -0.55, p < 0.01) and self-efficacy (r = -0.43, p < 0.01). Place of residence (rural vs. non-rural), academic performance (average and above vs. fair or poor), region (western region vs. non-western region), duration of epilepsy (≤5 years vs. >5 years), drug-resistant epilepsy (yes vs. no), comorbidities (yes vs. no), social support and self-efficacy are major influencing factors among the complex factors influencing the felt stigma among children and adolescents.

Conclusion

Medical staff should be more aware of stigma among children and adolescents with epilepsy, especially those who live in rural and western areas, have poor academic performance, have epilepsy duration of more than 5 years, have drug-resistant epilepsy, and have comorbidities, who are at higher risk of stigma. It is recommended that effective measures be taken to alleviate stigma by improving children and adolescents' self-efficacy and providing more social support for them and their families.

Anxiety in young people: Analysis from a machine learning model

The study addresses the detection of anxiety symptoms in young people using artificial intelligence models. Questionnaires such as the Patient Health Questionnaire-9 (PHQ-9) and Generalized Anxiety Disorder 7-item scale (GAD-7) are used to collect data, with a focus on early detection of anxiety. Three machine learning models are employed: Support Vector Machine (SVM), K Nearest Neighbors (KNN), and Random Forest (RF), with cross-validation to assess their effectiveness. Results show that the RF model is the most efficient, with an accuracy of 91 %, surpassing previous studies. Significant predictors of anxiety are identified, such as parental education level, alcohol consumption, and social security affiliation. A relationship is observed between anxiety and personal and family history of mental illness, as well as with characteristics external to the model, such as family and personal history of depression. The analysis of the results highlights the importance of considering not only clinical but also social and family aspects in mental health interventions. It is suggested that the sample size be expanded in future studies to improve the robustness of the model. In summary, the study demonstrates the usefulness of artificial intelligence in the early detection of anxiety in young people and highlights the relevance of addressing multidimensional factors in the assessment and treatment of this condition.

Microstate-based brain network dynamics distinguishing temporal lobe epilepsy patients: A machine learning approach

Temporal lobe epilepsy (TLE) stands as the predominant adult focal epilepsy syndrome, characterized by dysfunctional intrinsic brain dynamics. However, the precise mechanisms underlying seizures in these patients remain elusive. Our study encompassed 116 TLE patients compared with 51 healthy controls. Employing microstate analysis, we assessed brain dynamic disparities between TLE patients and healthy controls, as well as between drug-resistant epilepsy (DRE) and drug-sensitive epilepsy (DSE) patients. We constructed dynamic functional connectivity networks based on microstates and quantified their spatial and temporal variability. Utilizing these brain network features, we developed machine learning models to discriminate between TLE patients and healthy controls, and between DRE and DSE patients. Temporal dynamics in TLE patients exhibited significant acceleration compared to healthy controls, along with heightened synchronization and instability in brain networks. Moreover, DRE patients displayed notably lower spatial variability in certain parts of microstate B, E and F dynamic functional connectivity networks, while temporal variability in certain parts of microstate E and G dynamic functional connectivity networks was markedly higher in DRE patients compared to DSE patients. The machine learning model based on these spatiotemporal metrics effectively differentiated TLE patients from healthy controls and discerned DRE from DSE patients. The accelerated microstate dynamics and disrupted microstate sequences observed in TLE patients mirror highly unstable intrinsic brain dynamics, potentially underlying abnormal discharges. Additionally, the presence of highly synchronized and unstable activities in brain networks of DRE patients signifies the establishment of stable epileptogenic networks, contributing to the poor responsiveness to antiseizure medications. The model based on spatiotemporal metrics demonstrated robust predictive performance, accurately distinguishing both TLE patients from healthy controls and DRE patients from DSE patients.

Development and validation of a machine learning model for prediction of comorbid major depression disorder among narcolepsy type 1

BACKGROUND

Major depression disorder (MDD) forms a common psychiatric comorbidity among patients with narcolepsy type 1 (NT1), yet its impact on patients with NT1 is often overlooked by neurologists. Currently, there is a lack of effective methods for accurately predicting MDD in patients with NT1.

OBJECTIVE

This study utilized machine learning (ML) algorithms to identify critical variables and developed the prediction model for predicting MDD in patients with NT1.

METHODS

The study included 267 NT1 patients from four sleep centers. The diagnosis of comorbid MDD was based on Diagnostic and Statistical Manual of Mental Disorders fifth edition (DSM-5). ML models, including six full models and six compact models, were developed using a training set. The performance of these models was compared in the testing set, and the optimal model was evaluated in the testing set. Various evaluation metrics, such as Area under the receiver operating curve (AUC), precision-recall (PR) curve and calibration curve were employed to assess and compare the performance of the ML models. Model interpretability was demonstrated using SHAP.

RESULT

In the testing set, the logistic regression (LG) model demonstrated superior performance compared to other ML models based on evaluation metrics such as AUC, PR curve, and calibration curve. The top eight features used in the LG model, ranked by feature importance, included social impact scale (SIS) score, narcolepsy severity scale (NSS) score, total sleep time, body mass index (BMI), education years, age of onset, sleep efficiency, sleep latency.

CONCLUSION

The study yielded a straightforward and practical ML model for the early identification of MDD in patients with NT1. A web-based tool for clinical applications was developed, which deserves further verification in diverse clinical settings.

Fear of disease in patients with epilepsy - a network analysis

Background

Disease-related fear among patients with epilepsy has significantly impacted their quality of life. The Disease-Related Fear Scale (D-RFS), comprising three dimensions, serves as a relatively well-established tool for assessing fear in these patients. However, certain problems potentially exist within the D-RFS's attribution of items, and its internal structure is still unclear. To establish an appropriate dimensional structure and gain deeper comprehension of its internal structure-particularly its core variables-is vital for developing more effective interventions aimed at alleviating disease-related fear among patients with epilepsy.

Methods

This study employed a cross-sectional survey involving 609 patients with epilepsy. All participants underwent assessment using the Chinese version of the D-RFS. We used exploratory network analysis to discover a new structure and network analysis to investigate the interrelationships among fear symptom domains. In addition to the regularized partial correlation network, we also estimated the node and bridge centrality index to identify the importance of each item within the network. Finally, it was applied to analyze the differences in network analysis outcomes among epilepsy patients with different seizure frequencies.

Results

The research findings indicate that nodes within the network of disease-related fear symptoms are interconnected, and there are no isolated nodes. Nodes within groups 3 and 4 present the strongest centrality. Additionally, a tight interconnection exists among fear symptoms within each group. Moreover, the frequency of epileptic episodes does not significantly impact the network structure.

Conclusion

In this study, a new 5-dimension structure was constructed for D-RFS, and the fear of disease in patients with epilepsy has been conceptualized through a network perspective. The goal is to identify potential targets for relevant interventions and gain insights for future research.

Applying contrastive pre-training for depression and anxiety risk prediction in type 2 diabetes patients based on heterogeneous electronic health records: a primary healthcare case study

OBJECTIVE

Due to heterogeneity and limited medical data in primary healthcare services (PHS), assessing the psychological risk of type 2 diabetes mellitus (T2DM) patients in PHS is difficult. Using unsupervised contrastive pre-training, we proposed a deep learning framework named depression and anxiety prediction (DAP) to predict depression and anxiety in T2DM patients.

MATERIALS AND METHODS

The DAP model consists of two sub-models. Firstly, the pre-trained model of DAP used unlabeled discharge records of 85 085 T2DM patients from the First Affiliated Hospital of Nanjing Medical University for unsupervised contrastive learning on heterogeneous electronic health records (EHRs). Secondly, the fine-tuned model of DAP used case-control cohorts (17 491 patients) selected from 149 596 T2DM patients' EHRs in the Nanjing Health Information Platform (NHIP). The DAP model was validated in 1028 patients from PHS in NHIP. Evaluation included receiver operating characteristic area under the curve (ROC-AUC) and precision-recall area under the curve (PR-AUC), and decision curve analysis (DCA).

RESULTS

The pre-training step allowed the DAP model to converge at a faster rate. The fine-tuned DAP model significantly outperformed the baseline models (logistic regression, extreme gradient boosting, and random forest) with ROC-AUC of 0.91±0.028 and PR-AUC of 0.80±0.067 in 10-fold internal validation, and with ROC-AUC of 0.75 ± 0.045 and PR-AUC of 0.47 ± 0.081 in external validation. The DCA indicate the clinical potential of the DAP model.

CONCLUSION

The DAP model effectively predicted post-discharge depression and anxiety in T2DM patients from PHS, reducing data fragmentation and limitations. This study highlights the DAP model's potential for early detection and intervention in depression and anxiety, improving outcomes for diabetes patients.