JMIR Publications

ABSTRACT

Background:

Suicide in local jails occurs at a higher rate than in the general population, making it a priority to improve risk screening methods. This article describes a research study that will use administrative data and machine learning modeling to improve suicide risk detection at jail booking.

Objective:

This research study is primarily focused on gathering preliminary information about the feasibility and practicality of using administrative data and machine learning modeling for suicide risk detection, but also incorporates elements of hypothesis testing pertaining to clinical outcomes.

Methods:

The research study validates an existing community suicide risk identification machine learning model – developed and validated by the Mental Health Research Network (MHRN) – on a sample of ~6,000 individuals booked into two diverse jails in a Midwestern state. The model detects suicide risk in jails and post-release by using merged jail, Medicaid, and Vital records data.

Results:

The resulting model will be compared to the jails’ suicide identification practice-as-usual (‘PAU’), to assess risk and detection of identified suicide attempts and deaths from intake through 120 days and 13 months after jail release, and to modeling and PAU together. The research study will also investigate implementation factors, such as feasibility, acceptability, and appropriateness, to optimize jail uptake.

Conclusions:

We hypothesize that a combination of intake screening process-as-usual and the machine learning model will be the optimal approach to be evaluated.