Assessment of Undergraduate Psychological Wellbeing Using Explainable Machine Learning: A Study from Sri Lanka

Academic, psychosocial, financial, environmental, and interpersonal factors are widely recognized as important contributors to the overall wellbeing of students. This study aimed to identify factors associated with the psychological wellbeing of university undergraduates and compare the performance of machine learning models for binary and multiclass classification of wellbeing status using self-reported survey data from undergraduates at the University of Peradeniya, Sri Lanka. The findings may facilitate the early identification of students with lower wellbeing and inform targeted interventions to foster a healthier and more supportive academic environment. Cross-sectional questionnaire data from 2,913 undergraduates were analyzed to classify psychological wellbeing using overall happiness scores. Scores were grouped into binary (1-5 and 6-9) and multiclass (1-4, 5-6, and 7-9) categories. Data preprocessing included missing value imputation, one-hot encoding, and feature scaling where appropriate. Support Vector Machine (SVM), Random Forest (RF), XGBoost (XGB), and Multilayer Perceptron (MLP) models were trained using repeated stratified 10-fold cross-validation and evaluated using macro F1-score, ROC-AUC, and classification reports. SHAP analysis was employed to interpret feature contributions and enhance model interpretability. XGB achieved the highest macro-F1 score (70.5%) for binary classification, with a test accuracy of 72.5%. The model correctly classified 63.6% and 80% of participants in the lower and higher psychological wellbeing groups, respectively. For multiclass classification, RF achieved the highest macro-F1 score (54.6%), with a test accuracy of 56.5%. The model correctly classified 48%, 57%, and 60% of participants in the lower, intermediate, and higher psychological wellbeing groups, respectively. SHAP analysis identified academic coping, health and wellbeing, financial circumstances, and interpersonal relationships as key predictors of psychological wellbeing across both models. These findings demonstrate the potential of machine learning to classify psychological wellbeing groups using self-reported data, supporting early identification of students at risk of poor wellbeing and informing targeted university-level mental health interventions.

Authors

Institutions

Publication Details

Journal
medRxiv
Published
2026-11-12
DOI
https://doi.org/10.64898/2026.11.10.26367798
Primary Topic
Artificial Intelligence in Healthcare
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Assessment of Undergraduate Psychological Wellbeing Using Explainable Machine Learning: A Study from Sri Lanka

Damayanthi Herath, Kanthi Hettigoda, Shalika Kumburegama, Ruchika Fernando et al.
medRxiv
Artificial Intelligence in Healthcare
preprint

Assessment of Undergraduate Psychological Wellbeing Using Explainable Machine Learning: A Study from Sri Lanka

Damayanthi Herath, Kanthi Hettigoda, Shalika Kumburegama, Ruchika Fernando, Thushara Kamalrathne, Asanka Bulathwatta, S. R. Perera, Anoma Perera, Poornika Seelagama, Githmi Paranawithana
preprint en

Abstract

Academic, psychosocial, financial, environmental, and interpersonal factors are widely recognized as important contributors to the overall wellbeing of students. This study aimed to identify factors associated with the psychological wellbeing of university undergraduates and compare the performance of machine learning models for binary and multiclass classification of wellbeing status using self-reported survey data from undergraduates at the University of Peradeniya, Sri Lanka. The findings may facilitate the early identification of students with lower wellbeing and inform targeted interventions to foster a healthier and more supportive academic environment. Cross-sectional questionnaire data from 2,913 undergraduates were analyzed to classify psychological wellbeing using overall happiness scores. Scores were grouped into binary (1-5 and 6-9) and multiclass (1-4, 5-6, and 7-9) categories. Data preprocessing included missing value imputation, one-hot encoding, and feature scaling where appropriate. Support Vector Machine (SVM), Random Forest (RF), XGBoost (XGB), and Multilayer Perceptron (MLP) models were trained using repeated stratified 10-fold cross-validation and evaluated using macro F1-score, ROC-AUC, and classification reports. SHAP analysis was employed to interpret feature contributions and enhance model interpretability. XGB achieved the highest macro-F1 score (70.5%) for binary classification, with a test accuracy of 72.5%. The model correctly classified 63.6% and 80% of participants in the lower and higher psychological wellbeing groups, respectively. For multiclass classification, RF achieved the highest macro-F1 score (54.6%), with a test accuracy of 56.5%. The model correctly classified 48%, 57%, and 60% of participants in the lower, intermediate, and higher psychological wellbeing groups, respectively. SHAP analysis identified academic coping, health and wellbeing, financial circumstances, and interpersonal relationships as key predictors of psychological wellbeing across both models. These findings demonstrate the potential of machine learning to classify psychological wellbeing groups using self-reported data, supporting early identification of students at risk of poor wellbeing and informing targeted university-level mental health interventions.

medRxiv
University of Peradeniya (LK)
Artificial Intelligence in Healthcare
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon