Identifying probable post-traumatic stress disorder: applying supervised machine learning to data from a UK military...

Area of Science:

• Military health research
• Computational psychiatry
• Machine learning applications

Background:

• Early identification of probable post-traumatic stress disorder (PTSD) is crucial for timely intervention.
• The UK Armed Forces population experiences unique service-related exposures.
• Validated self-report measures are essential for assessing mental health conditions.

Purpose of the Study:

• To evaluate supervised machine learning (ML) classifiers for identifying probable PTSD.
• To assess the feasibility of using ML for probable PTSD detection in UK military personnel.
• To explore the potential of ML in improving early intervention strategies.

Main Methods:

• Applied supervised ML classification techniques to a cohort of 13,690 UK Armed Forces personnel.
• Utilized self-reported service exposures and validated self-report measures for data collection (2004-2009).
• Analyzed the predictive performance and contributing variables of ML classifiers for probable PTSD.

Main Results:

• Supervised ML classifiers demonstrated encouraging predictive performance for probable PTSD detection.
• Key contributing variables identified included alcohol misuse, gender, and deployment status.
• Satisfactory sensitivity was achieved, though low sensitivity indicated a potential for false negatives.

Conclusions:

• Detection of probable PTSD using self-reported data and ML is feasible.
• Early intervention enabled by ML can reduce public health burdens and improve operational efficiencies.
• ML offers a promising tool for proactive mental health management in military populations.