Shorter, faster, but still accurate: using ant colony optimisation to develop and validate the Athlete Mental Health Screener-18 (AMHS-18) as a brief alternative to Sport Mental Health Assessment Tool 1 (SMHAT-1)

Objective Sport Mental Health Assessment Tool 1 (SMHAT-1) ensures comprehensive and reliable mental health screening in athletes. However, its triage’s predictive validity varies depending on the mental health domain (5%–60% false negative rates), while the full algorithm imposes a substantial time burden for routine screening. This study aimed to develop and validate a brief, clinically accurate alternative using the pool of items included in SMHAT-1. Methods SMHAT-1 and subsequent brief clinical interviews were administered in Polish Olympic athletes over three waves of routine medical check-ups. An ant colony optimisation algorithm that evaluates item combinations holistically rather than sequentially was trained on two datasets (N 1 =1121, N 2 =803) to identify the optimal item subset predicting post-interview mental health recommendations. The solution underwent independent validation in the third dataset (N 3 =1134). Results The best solution comprised 18 items, achieved 81% accuracy in predicting the post-clinical interview mental health evaluation in the training datasets and 80% in an independent dataset, overperforming the SMHAT-1 triage on all performance indicators. The Athlete Mental Health Screen-18 (AMHS-18) demonstrated superior sensitivity (94%) and specificity (72%) at the optimal cut-off (≥19), with strong internal consistency (ω=0.82). Conclusion The AMHS-18 is a time-efficient, valid and reliable tool for accurate mental health screening in athletes. It offers a practical alternative to the full SMHAT-1, particularly in settings where time and resources are limited or when athletes are screened regularly. We recommend using AMHS-18 instead of SMHAT-1’s triage and alternating SMHAT-1 with AMHS-18 as a starting point for routine brief clinical intake interviews.