Using Machine Learning to Examine Suicidal Ideation After Traumatic Brain Injury: A Traumatic Brain Injury Model Systems National Database Study.
Lauren B Fisher, Joshua E Curtiss, Daniel W Klyce and 10 others
PMID 35687765WHAT IT FOUND
A machine learning model predicted suicidal ideation one year after TBI with excellent accuracy.
Depressed mood and guilt were the strongest predictors, followed by anhedonia and concentration difficulties. Demographic and injury severity factors added no predictive value. This suggests screening for specific depression symptoms may be more useful than general risk factors.
Key findings
01Depressed mood and guilt were the most important predictors of suicidal ideation, with excellent discrimination, while demographic variables and injury severity performed no better than chance.
02The best performing model (gradient boosting machine) achieved excellent classification performance with an AUC of 0.882, sensitivity of 0.85, and specificity of 0.77.
03Adding demographic variables and injury severity to the final model did not improve performance, actually reducing the AUC to 0.77.
STILL TO COME
How it was doneWhat they foundWhat it means for PTsWhat it means for OTsWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study only predicted suicidal ideation, not suicide attempts or deaths. Most people with ideation do not attempt suicide, and predictors of ideation often do not distinguish who will attempt. The predictors for depression came from the same scale (PHQ-9) as the outcome measure (item 9), which may inflate the association due to method artifact. The sample was limited to individuals who received specialized inpatient rehabilitation, representing only 7% of all persons hospitalized with moderate-to-severe TBI, and less likely to include racial/ethnic minorities or uninsured individuals. Self-reported data may be inaccurate due to deficits in self-awareness common after TBI, though the authors note this might minimize under-reporting of depressive symptoms. The PHQ-9 item 9 has mixed evidence for detecting suicidal ideation and may over-estimate risk compared to more comprehensive tools like the Columbia Suicide Severity Rating Scale.
Declared interests
The authors have no conflicts of interest to disclose. The study used data from the NIDILRR-funded TBIMS National Database.
The easy way to misread this
Do not interpret this model as a clinical tool for predicting suicide attempts or deaths. It predicts suicidal ideation based on self-reported depression symptoms from the same questionnaire, and the authors explicitly state that the clinical impact is limited by the use of a one-item screening tool and the lack of distinction between ideation and behavior.