A machine learning approach to identification of self-harm and suicidal ideation among military and police Veterans

LAY SUMMARY Combat Veterans are vulnerable to suicidal thoughts and behaviour. Many who die by suicide deny having suicidal ideation (SI). Typically, researchers try to find variables indicating the presence of SI using traditional statistical approaches. These approaches do not possess the capacity to detect highly complex multivariable interactions. In contrast, machine learning (ML) is designed to detect such patterns and can consequently yield much higher predictive accuracy. In this study, the authors trained ML algorithms using 192 variables extracted from questionnaires administered to 738 Veterans and serving personnel to detect the presence of self-harm and SI (SHSI). Using the 10 most predictive non-suicide-related items, the ML algorithms could detect SHSI with 75.3% accuracy. Most of these items reflect psychological phenomena that can change quickly over time, allowing repeated risk reassessment from day to day. The study’s findings suggest that ML methods may play an important role in the discovery, within a large data set, of predictive patterns that might be useful in suicide risk assessment. Introduction: Combat Veterans are at increased risk for suicidal ideation (SI). Many who die by suicide deny having SI, so alternative approaches to asking about suicide are needed. Current statistical approaches test whether a hypothesized SI predictor variable is significantly different in groups with and without SI. These group-based methods are of limited value for identifying SI among individuals. The objective of this study was to test the utility and feasibility of machine learning (ML) analysis of the kind of data that could be easily collected in an operational stress injury clinic in order to offer new insights into the detection of SI. Methods: ML algorithms to detect self-harm and SI (SHSI) were trained using 192 variables from questionnaires administered to 738 Veterans and serving members. An autoencoder was used to impute missing data to maximize training sample size. Results: The ML algorithms detected SHSI with an accuracy of 75.3% (area under the receiver operating characteristic curve = 82.7%). Of the 10 items identified, none asked about suicide. Discussion: ML methods can detect patterns predictive of SHSI in large data sets and could aid in early intervention and, ultimately, suicide prevention for individuals.