College student mental health assessment: Predictive models based on machine learning and feature importance analysis
To assess and forecast the mental health conditions of university students utilizing machine learning methodologies, focusing particularly on the influence of the nine psychological symptom dimensions encompassed by the Symptom Checklist-90 (SCL-90). The prevalence of mental health issues among coll...
Uložené v:
| Vydané v: | Molecular & cellular biomechanics Ročník 22; číslo 3; s. 1477 |
|---|---|
| Hlavní autori: | , , , , |
| Médium: | Journal Article |
| Jazyk: | English |
| Vydavateľské údaje: |
21.02.2025
|
| ISSN: | 1556-5297, 1556-5300 |
| On-line prístup: | Získať plný text |
| Tagy: |
Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
|
| Shrnutí: | To assess and forecast the mental health conditions of university students utilizing machine learning methodologies, focusing particularly on the influence of the nine psychological symptom dimensions encompassed by the Symptom Checklist-90 (SCL-90). The prevalence of mental health issues among college students is a significant concern. Traditional methods for assessing mental health may lack the precision required for early detection and intervention. Machine learning offers advanced tools to analyze complex data and predict outcomes based on multiple variables. The primary objective is to construct and evaluate predictive models for the mental health status of college students using various machine learning algorithms, optimize their performance, and identify the most impactful psychological symptom dimensions. Psychological health data from 11,943 college students were gathered via an online questionnaire platform. Multiple machine learning algorithms were utilized to develop predictive models. Hyperparameter optimization was achieved through K-fold cross-validation and the northern goshawk optimization algorithm. To tackle class imbalance, the synthetic minority over-sampling technique was employed to create synthetic samples for underrepresented classes. Model performance was assessed using metrics such as accuracy, recall, and f1 score. The light gradient boosting algorithm demonstrated superior performance, with only 6 misclassifications out of 2,388 test samples. Tree-based ensemble methods like random forest and extreme gradient boosting consistently outperformed non-ensemble methods such as k-nearest neighbors, multi-layer perceptron, and kernel discriminant analysis. A detailed analysis using Shapley additive explanations values indicated that features such as obsessive-compulsive symptoms and anxiety were the most influential in the model’s predictions. This study underscores the efficacy and potential of machine learning in mental health assessment. The results provide a robust scientific foundation for the development of early warning systems and targeted intervention strategies to enhance the mental well-being of college students. |
|---|---|
| ISSN: | 1556-5297 1556-5300 |
| DOI: | 10.62617/mcb1477 |