An explainable predictive model for suicide attempt risk using an ensemble learning and Shapley Additive Explanations...

Area of Science:

• Computational psychiatry
• Clinical informatics
• Machine learning in healthcare

Background:

• Machine learning models show promise in predicting suicide attempts but often lack interpretability.
• Understanding the 'why' behind predictions is crucial for clinical decision-making and intervention.

Purpose of the Study:

• To develop an explainable predictive model for suicide attempts.
• To analyze the importance of predictive features for suicide attempts.
• To enhance clinical understanding of suicide attempt risk factors.

Main Methods:

• Implemented two ensemble learning models: Random Forest and Gradient Boosting.
• Integrated SHapley Additive exPlanations (SHAP) for model interpretability.
• Evaluated model performance and feature importance for suicide attempt prediction.

Main Results:

• Both Random Forest and Gradient Boosting with SHAP provided interpretable predictions.
• Gradient Boosting with SHAP demonstrated higher predictive accuracy compared to Random Forest.
• Key predictors identified include history of suicide attempts, suicidal ideation, and ethnicity.

Conclusions:

• Explainable AI, specifically Gradient Boosting with SHAP, can effectively predict suicide attempts.
• The model enhances clinical understanding by highlighting significant risk factors.
• This approach supports more informed clinical decisions regarding suicide prevention.