Predicting quality of life based on mental health state: A machine learning approach using Urban-HEART 2

Abstract

Introduction: Quality of life (QoL) is a complex and multifaceted concept often used as an indicator in evaluating public policy. This study aimed to predict QoL based on mental health state using a machine learning approach. The analysis was conducted using data from the Urban Health Equity Assessment and Response Tool (Urban-HEART 2) survey conducted in Tehran, Iran. Methods: This secondary analysis utilized data from the second round of the Urban-HEART 2 survey, which included 117,839 participants. Various machine learning (ML) algorithms were employed, including Random Forest, Decision Tree, Support Vector Machine (SVM), Naive Bayes, and Logistic Regression. Additionally, an unsupervised learning method, specifically k-means clustering, was used. Results: Following data preparation, the k-means clustering algorithm identified five clusters based on mental health features. ML algorithms were then utilized to predict each participant's QoL label through distinct scores. The top-performing ML algorithms based on high scores were found to be Random Forest (0.994), Decision Tree (0.991), SVM (0.990), Naive Bayes (0.935), and Logistic Regression (0.934), respectively. Conclusions: By implementing k-means clustering, we identified distinct clusters based on mental health features and assigned labels to each participant accordingly. Machine learning models accurately predicted the QoL label for each participant. All models achieved high scores (above 0.93), indicating that mental health features can reliably predict QoL labels with high accuracy.

Keywords

• academic performance
• metacognitive learning strategies
• nursing students
• planning control and regulation