Investigating the differential impact of psychosocial factors by patient characteristics and demographics on Veteran suicide risk through machine learning extraction of cross-modal interactions
Accurate prediction of suicide risk is crucial for identifying patients with elevated risk burden, helping ensure these patients receive targeted care. The US Department of Veteran Affairs' suicide prediction model primarily leverages structured electronic health records (EHR) data. This approach largely overlooks unstructured EHR, a data format that could be utilized to enhance predictive accuracy. This study aims to enhance suicide risk models' predictive accuracy by developing a model that incorporates both structured EHR predictors and semantic NLP-derived variables from unstructured EHR. XGBoost models were fit to predict suicide risk- the interactions identified by the model were extracted using SHAP, validated using logistic regression models, added to a ridge regression model, which was subsequently compared to a ridge regression approach without the use of interactions. By introducing a selection parameter, α, to balance the influence of structured (α=1) and unstructured (α=0) data, we found that intermediate α values achieved optimal performance across various risk strata, improved model performance of the ridge regression approach and uncovered significant cross-modal interactions between psychosocial constructs and patient characteristics. These interactions highlight how psychosocial risk factors are influenced by individual patient contexts, potentially informing improved risk prediction methods and personalized interventions. Our findings underscore the importance of incorporating nuanced narrative data into predictive models and set the stage for future research that will expand the use of advanced machine learning techniques, including deep learning, to further refine suicide risk prediction methods.
Abstract: Disordered eating is a prevalent and relevant health concern that remains understudied among U.S. military veterans. Avoidant/restrictive food intake disorder (ARFID) is a newly recognized feeding and eating disorder characterized by overly restrictive eating due to (a) picky eating, (b) lack of appetite, and (c) fear of aversive consequences related to eating. The Nine-Item ARFID Screen (NIAS) is a recently developed ARFID screening tool with initial validation studies demonstrating psychometric support. However, the psychometric properties of the NIAS have not been investigated in a veteran sample. To advance our understanding of ARFID screening tools that may be appropriate for use in veterans, the present study examined the factor structure of the NIAS using survey data from a large national sample of recently separated veterans (N = 1,486). Measurement invariance across key subgroups was tested in addition to exploring differential associations between the NIAS and related constructs. Results suggested that a three-factor model provided an excellent fit of the data and demonstrated scalar invariance across self-identified men and women, race and ethnicity, and sexual and gender minority (SGM) identity. Some subgroups had higher latent means on the picky eating (women, SGM, non-Hispanic Black), appetite (women, SGM), and fear (women) factors. The NIAS had some overlap with another measure of disordered eating and was moderately correlated with psychosocial impairment and mental health. Overall, the NIAS may be a useful screening tool for ARFID in veterans, given support for the three proposed subscales and equivalence across diverse identities.