Abstract

Purpose: Machine learning models (MLM) can be used for analyzing the DAS questionnaire and its impact on mental health of residents.
Methodology: Depression, Anxiety, and Stress (DAS) score of residents, was used to collect relevant information in year 2022 at tertiary hospital. PIFCOM algorithm was then used to find the combinations of features that had the highest accuracy in determining each aspect of the DAS. Results: The questions pertaining to “Modalities of learning prior to pandemic”, “Age”, and “impact of COVID19 on thesis” were significant determinants of all the targets. By Random Forest method, Stress was the best predicted parameter (Accuracy = 0.857), followed by Anxiety (Accuracy = 0.803), and Depression (Accuracy = 0.589). Conclusion: Rather than conventional methods machine learning modalities can accurately predict the score-based parameters with much fewer questions. This can be translated into robust and swift report-generating apps for even self-evaluation.