DeepBiomarker: Identifying Important Lab Tests from Electronic Medical Records for the Prediction of Suicide-Related Events among PTSD Patients

Role of machine learning algorithms in suicide risk prediction: a systematic review-meta analysis of clinical studies

OBJECTIVE

Suicide is a complex and multifactorial public health problem. Understanding and addressing the various factors associated with suicide is crucial for prevention and intervention efforts. Machine learning (ML) could enhance the prediction of suicide attempts.

METHOD

A systematic review was performed using PubMed, Scopus, Web of Science and SID databases. We aim to evaluate the performance of ML algorithms and summarize their effects, gather relevant and reliable information to synthesize existing evidence, identify knowledge gaps, and provide a comprehensive list of the suicide risk factors using mixed method approach.

RESULTS

Forty-one studies published between 2011 and 2022, which matched inclusion criteria, were chosen as suitable. We included studies aimed at predicting the suicide risk by machine learning algorithms except natural language processing (NLP) and image processing. The neural network (NN) algorithm exhibited the lowest accuracy at 0.70, whereas the random forest demonstrated the highest accuracy, reaching 0.94. The study assessed the COX and random forest models and observed a minimum area under the curve (AUC) value of 0.54. In contrast, the XGBoost classifier yielded the highest AUC value, reaching 0.97. These specific AUC values emphasize the algorithm-specific performance in capturing the trade-off between sensitivity and specificity for suicide risk prediction. Furthermore, our investigation identified several common suicide risk factors, including age, gender, substance abuse, depression, anxiety, alcohol consumption, marital status, income, education, and occupation. This comprehensive analysis contributes valuable insights into the multifaceted nature of suicide risk, providing a foundation for targeted preventive strategies and intervention efforts.

CONCLUSIONS

The effectiveness of ML algorithms and their application in predicting suicide risk has been controversial. There is a need for more studies on these algorithms in clinical settings, and the related ethical concerns require further clarification.

Machine learning and the prediction of suicide in psychiatric populations: a systematic review

Machine learning (ML) has emerged as a promising tool to enhance suicidal prediction. However, as many large-sample studies mixed psychiatric and non-psychiatric populations, a formal psychiatric diagnosis emerged as a strong predictor of suicidal risk, overshadowing more subtle risk factors specific to distinct populations. To overcome this limitation, we conducted a systematic review of ML studies evaluating suicidal behaviors exclusively in psychiatric clinical populations. A systematic literature search was performed from inception through November 17, 2022 on PubMed, EMBASE, and Scopus following the PRISMA guidelines. Original research using ML techniques to assess the risk of suicide or predict suicide attempts in the psychiatric population were included. An assessment for bias risk was performed using the transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD) guidelines. About 1032 studies were retrieved, and 81 satisfied the inclusion criteria and were included for qualitative synthesis. Clinical and demographic features were the most frequently employed and random forest, support vector machine, and convolutional neural network performed better in terms of accuracy than other algorithms when directly compared. Despite heterogeneity in procedures, most studies reported an accuracy of 70% or greater based on features such as previous attempts, severity of the disorder, and pharmacological treatments. Although the evidence reported is promising, ML algorithms for suicidal prediction still present limitations, including the lack of neurobiological and imaging data and the lack of external validation samples. Overcoming these issues may lead to the development of models to adopt in clinical practice. Further research is warranted to boost a field that holds the potential to critically impact suicide mortality.

Prediction of adverse events risk in patients with comorbid post-traumatic stress disorder and alcohol use disorder using electronic medical records by deep learning models

BACKGROUND

Identifying co-occurring mental disorders and elevated risk is vital for optimization of healthcare processes. In this study, we will use DeepBiomarker2, an updated version of our deep learning model to predict the adverse events among patients with comorbid post-traumatic stress disorder (PTSD) and alcohol use disorder (AUD), a high-risk population.

METHODS

We analyzed electronic medical records of 5565 patients from University of Pittsburgh Medical Center to predict adverse events (opioid use disorder, suicide related events, depression, and death) within 3 months at any encounter after the diagnosis of PTSD+AUD by using DeepBiomarker2. We integrated multimodal information including: lab tests, medications, co-morbidities, individual and neighborhood level social determinants of health (SDoH), psychotherapy and veteran data.

RESULTS

DeepBiomarker2 achieved an area under the receiver operator curve (AUROC) of 0.94 on the prediction of adverse events among those PTSD+AUD patients. Medications such as vilazodone, dronabinol, tenofovir, suvorexant, modafinil, and lamivudine showed potential for risk reduction. SDoH parameters such as cognitive behavioral therapy and trauma focused psychotherapy lowered risk while active veteran status, income segregation, limited access to parks and greenery, low Gini index, limited English-speaking capacity, and younger patients increased risk.

CONCLUSIONS

Our improved version of DeepBiomarker2 demonstrated its capability of predicting multiple adverse event risk with high accuracy and identifying potential risk and beneficial factors.

DeepBiomarker2: Prediction of Alcohol and Substance Use Disorder Risk in Post-Traumatic Stress Disorder Patients Using Electronic Medical Records and Multiple Social Determinants of Health

Prediction of high-risk events amongst patients with mental disorders is critical for personalized interventions. We developed DeepBiomarker2 by leveraging deep learning and natural language processing to analyze lab tests, medication use, diagnosis, social determinants of health (SDoH) parameters, and psychotherapy for outcome prediction. To increase the model's interpretability, we further refined our contribution analysis to identify key features by scaling with a factor from a reference feature. We applied DeepBiomarker2 to analyze the EMR data of 38,807 patients from the University of Pittsburgh Medical Center diagnosed with post-traumatic stress disorder (PTSD) to determine their risk of developing alcohol and substance use disorder (ASUD). DeepBiomarker2 predicted whether a PTSD patient would have a diagnosis of ASUD within the following 3 months with an average c-statistic (receiver operating characteristic AUC) of 0.93 and average F1 score, precision, and recall of 0.880, 0.895, and 0.866 in the test sets, respectively. Our study found that the medications clindamycin, enalapril, penicillin, valacyclovir, Xarelto/rivaroxaban, moxifloxacin, and atropine and the SDoH parameters access to psychotherapy, living in zip codes with a high normalized vegetative index, Gini index, and low-income segregation may have potential to reduce the risk of ASUDs in PTSD. In conclusion, the integration of SDoH information, coupled with the refined feature contribution analysis, empowers DeepBiomarker2 to accurately predict ASUD risk. Moreover, the model can further identify potential indicators of increased risk along with medications with beneficial effects.

Disentangling the effects of PTSD from Gulf War Illness in male veterans via a systems-wide analysis of immune cell, cytokine, and symptom measures

BACKGROUND

One-third of veterans returning from the 1990-1991 Gulf War reported a myriad of symptoms including cognitive dysfunction, skin rashes, musculoskeletal discomfort, and fatigue. This symptom cluster is now referred to as Gulf War Illness (GWI). As the underlying mechanisms of GWI have yet to be fully elucidated, diagnosis and treatment are based on symptomatic presentation. One confounding factor tied to the illness is the high presence of post-traumatic stress disorder (PTSD). Previous research efforts have demonstrated that both GWI and PTSD are associated with immunological dysfunction. As such, this research endeavor aimed to provide insight into the complex relationship between GWI symptoms, cytokine presence, and immune cell populations to pinpoint the impact of PTSD on these measures in GWI.

METHODS

Symptom measures were gathered through the Multidimensional fatigue inventory (MFI) and 36-item short form health survey (SF-36) scales and biological measures were obtained through cytokine & cytometry analysis. Subgrouping was conducted using Davidson Trauma Scale scores and the Structured Clinical Interview for Diagnostic and statistical manual of mental disorders (DSM)-5, into GWI with high probability of PTSD symptoms (GWIH) and GWI with low probability of PTSD symptoms (GWIL). Data was analyzed using Analysis of variance (ANOVA) statistical analysis along with correlation graph analysis. We mapped correlations between immune cells and cytokine signaling measures, hormones and GWI symptom measures to identify patterns in regulation between the GWIH, GWIL, and healthy control groups.

RESULTS

GWI with comorbid PTSD symptoms resulted in poorer health outcomes compared with both Healthy control (HC) and the GWIL subgroup. Significant differences were found in basophil levels of GWI compared with HC at peak exercise regardless of PTSD symptom comorbidity (ANOVA F = 4.7, P = 0.01,) indicating its potential usage as a biomarker for general GWI from control. While the unique identification of GWI with PTSD symptoms was less clear, the GWIL subgroup was found to be delineated from both GWIH and HC on measures of IL-15 across an exercise challenge (ANOVA F > 3.75, P < 0.03). Additional differences in natural killer (NK) cell numbers and function highlight IL-15 as a potential biomarker of GWI in the absence of PTSD symptoms.

CONCLUSION

We conclude that disentangling GWI and PTSD by defining trauma-based subgroups may aid in the identification of unique GWI biosignatures that can help to improve diagnosis and target treatment of GWI more effectively.

Prediction of Adverse Events Risk in Patients with Comorbid Post- Traumatic Stress Disorder and Alcohol Use Disorder Using Electronic Medical Records by Deep Learning Models

Background

Prediction of high-risk events in mental disorder patients is crucial. In our previous study, we developed a deep learning model: DeepBiomarker by using electronic medical records (EMR) to predict suicide related event (SRE) risk in post-traumatic stress disorder (PTSD) patients.

Methods

We applied DeepBiomarker2 through data integration of multimodal information: lab test, medication, co-morbidities, and social determinants of health. We analyzed EMRs of 5,565 patients from University of Pittsburgh Medical Center with a diagnosis of PTSD and alcohol use disorder (AUD) on risk of developing an adverse event (opioid use disorder, SREs, depression and death).

Results

DeepBiomarker2 predicted whether a PTSD + AUD patient will have a diagnosis of any adverse events (SREs, opioid use disorder, depression, death) within 3 months with area under the receiver operator curve (AUROC) of 0.94. We found piroxicam, vilazodone, dronabinol, tenofovir, suvorexant, empagliflozin, famciclovir, veramyst, amantadine, sulfasalazine, and lamivudine to have potential to reduce risk.

Conclusions

DeepBiomarker2 can predict multiple adverse event risk with high accuracy and identify potential risk and beneficial factors. Our results offer suggestions for personalized interventions in a variety of clinical and diverse populations.

Unveiling the Enigma: Exploring Risk Factors and Mechanisms for Psychotic Symptoms in Alzheimer's Disease through Electronic Medical Records with Deep Learning Models

Around 50% of patients with Alzheimer's disease (AD) may experience psychotic symptoms after onset, resulting in a subtype of AD known as psychosis in AD (AD + P). This subtype is characterized by more rapid cognitive decline compared to AD patients without psychosis. Therefore, there is a great need to identify risk factors for the development of AD + P and explore potential treatment options. In this study, we enhanced our deep learning model, DeepBiomarker, to predict the onset of psychosis in AD utilizing data from electronic medical records (EMRs). The model demonstrated superior predictive capacity with an AUC (area under curve) of 0.907, significantly surpassing conventional risk prediction models. Utilizing a perturbation-based method, we identified key features from multiple medications, comorbidities, and abnormal laboratory tests, which notably influenced the prediction outcomes. Our findings demonstrated substantial agreement with existing studies, underscoring the vital role of metabolic syndrome, inflammation, and liver function pathways in AD + P. Importantly, the DeepBiomarker model not only offers a precise prediction of AD + P onset but also provides mechanistic understanding, potentially informing the development of innovative treatments. With additional validation, this approach could significantly contribute to early detection and prevention strategies for AD + P, thereby improving patient outcomes and quality of life.

DeepBiomarker2: Prediction of alcohol and substance use disorder risk in post-traumatic stress disorder patients using electronic medical records and multiple social determinants of health

Introduction

Prediction of high-risk events amongst patients with mental disorders is critical for personalized interventions. In our previous study, we developed a deep learning-based model, DeepBiomarker by utilizing electronic medical records (EMR) to predict the outcomes of patients with suicide-related events in post-traumatic stress disorder (PTSD) patients.

Methods

We improved our deep learning model to develop DeepBiomarker2 through data integration of multimodal information: lab tests, medication use, diagnosis, and social determinants of health (SDoH) parameters (both individual and neighborhood level) from EMR data for outcome prediction. We further refined our contribution analysis for identifying key factors. We applied DeepBiomarker2 to analyze EMR data of 38,807 patients from University of Pittsburgh Medical Center diagnosed with PTSD to determine their risk of developing alcohol and substance use disorder (ASUD).

Results

DeepBiomarker2 predicted whether a PTSD patient will have a diagnosis of ASUD within the following 3 months with a c-statistic (receiver operating characteristic AUC) of 0·93. We used contribution analysis technology to identify key lab tests, medication use and diagnosis for ASUD prediction. These identified factors imply that the regulation of the energy metabolism, blood circulation, inflammation, and microbiome is involved in shaping the pathophysiological pathways promoting ASUD risks in PTSD patients. Our study found protective medications such as oxybutynin, magnesium oxide, clindamycin, cetirizine, montelukast and venlafaxine all have a potential to reduce risk of ASUDs.

Discussion

DeepBiomarker2 can predict ASUD risk with high accuracy and can further identify potential risk factors along with medications with beneficial effects. We believe that our approach will help in personalized interventions of PTSD for a variety of clinical scenarios.