An efficient landmark model for prediction of suicide attempts in multiple clinical settings. (May 2023)

• Record Type: Journal Article
• Authors: Sheu, Yi-han; Sun, Jiehuan; Lee, Hyunjoon; Castro, Victor M.; Barak-Corren, Yuval; Song, Eugene; Madsen, Emily M.; Gordon, William J.; Kohane, Isaac S.; Churchill, Susanne E.; Reis, Ben Y.; Cai, Tianxi; Smoller, Jordan W.
• Is Part Of: Psychiatry research. Volume 323(2023)
• Publication Date: 2023-05
• Subjects: Suicide attempt -- Prediction -- Electronic health record -- Landmark model; Psychiatry -- Periodicals; Psychiatry -- periodicals; Psychiatrie -- Périodiques; 616.89
• DOI: 10.1016/j.psychres.2023.115175
• Languages: English
• ISSNs: 0165-1781

Abstract

Highlights: We developed and validated machine learning models for prediction of suicide attempts using large-scale electronic health record data across three clinical settings (general outpatient, emergency department, inpatient psychiatry). Models used a "landmark model" framework that mirrors the data available to a clinician at the time of risk prediction. The main model achieved high discriminative performance (AUROC = 0.74–0.93) across varying prediction windows and clinical settings, even with relatively short periods of historical data. This strategy can reduce bias and enhance the reliability and portability of suicide risk prediction models. Abstract: Growing evidence has shown that applying machine learning models to large clinical data sources may exceed clinician performance in suicide risk stratification. However, many existing prediction models either suffer from "temporal bias" (a bias that stems from using case-control sampling) or require training on all available patient visit data. Here, we adopt a "landmark model" framework that aligns with clinical practice for prediction of suicide-related behaviors (SRBs) using a large electronic health record database. Using the landmark approach, we developed models for SRB prediction (regularized Cox regression and random survival forest) that establish a time-point (e.g., clinical visit) from which predictions are made over user-specified prediction windows using historical information up to that point. We applied this approach to cohorts from three clinical settings: general outpatient, psychiatric emergency department, and psychiatric inpatients, for varying prediction windows and lengths of historical data. Models achieved high discriminative performance (area under the Receiver Operating Characteristic curve 0.74–0.93 for the Cox model) across different prediction windows and settings, even with relatively short periods of historical data. In short, we developed accurate, dynamic SRB risk prediction models with the landmark approach that reduce bias and enhance the reliability and portability of suicide risk prediction models.