Machine Learning-Based Evaluation of Suicide Risk Assessment in Crisis Counseling Calls.

OBJECTIVE: Counselor assessment of suicide risk is one key component of crisis counseling, and standards require risk assessment in every crisis counseling conversation. Efforts to increase risk assessment frequency are limited by quality improvement tools that rely on human evaluation of conversations, which is labor intensive, slow, and impossible to scale. Advances in machine learning (ML) have made possible the development of tools that can automatically and immediately detect the presence of risk assessment in crisis counseling conversations. METHODS: To train models, a coding team labeled every statement in 476 crisis counseling calls (193,257 statements) for a core element of risk assessment. The authors then fine-tuned a transformer-based ML model with the labeled data, utilizing separate training, validation, and test data sets. RESULTS: Generally, the evaluated ML model was highly consistent with human raters. For detecting any risk assessment, ML model agreement with human ratings was 98% of human interrater agreement. Across specific labels, average F1 (the harmonic mean of precision and recall) was 0.86 at the call level and 0.66 at the statement level and often varied as a result of a low base rate for some risk labels. CONCLUSIONS: ML models can reliably detect the presence of suicide risk assessment in crisis counseling conversations, presenting an opportunity to scale quality improvement efforts.

Current Suicidal Ideation among Treatment-Engaged Active Duty Soldiers and Marines.

We examined suicidal ideation among 399 active duty Soldiers and Marines engaged in mental health treatment. Using a generalized linear model controlling for demographic and military factors, depression, and positive traumatic brain injury screen, we confirmed our hypothesis that self-report measures of current PTSD symptoms uniquely predicted suicidal ideation. The association between PTSD severity and suicidal ideation was moderated by gender with women at higher risk as PTSD severity increased. Female Soldiers and Marines with high levels of PTSD should receive additional monitoring and intervention. Self-report measures may aid with risk assessment and identify symptom-related distress associated with suicide risk.

Geriatric patients improve as much as younger patients from hospitalization on general psychiatric units.

OBJECTIVES: To determine whether geriatric patients aged 65 and older on general adult psychiatric units improve as much as younger patients, over what duration their improvement occurs, and their risk of readmission. DESIGN: Cohort study. SETTING: Inpatient psychiatric unit of an urban, university-affiliated, county hospital from January 1993 through August 1999. PARTICIPANTS: A total of 5,929 inpatients. MEASUREMENTS: Standardized, routine assessments by attending psychiatrists included the Psychiatric Symptom Assessment Scale (PSAS) on admission and discharge. Discharge scores, length of stay (LOS), and risk of readmission within 1 year were modeled for the groups using multiple regression analyses. RESULTS: Geriatric patients constituted 5% (n=299) of the 5,929 admissions. In multivariate analysis, geriatric status was not associated with discharge PSAS scores. Median LOS was longer for geriatric patients (16 days) than younger patients (10 days, P<.001), especially in older women (14 days) and geriatric patients with mild medical illness severity (13 days vs 11 days in those with moderate-to-severe medical illness). Geriatric patients were as likely to be readmitted within 1 year of discharge as younger patients. CONCLUSION: Geriatric patients on general inpatient psychiatry units improved as much as younger patients. Their longer LOS was associated with milder medical illness severity. There may be a role for more specialized care of elderly women or geriatric patients with mild to moderate medical illness to improve the efficiency of their care.