A Comparative Study of Traditional Machine Learning, Deep Learning, and Large Language Models for Mental Health Forecasting using Smartphone Sensing Data
Kaidong Feng, Zhu Sun, Roy Ka-Wei Lee, Xun Jiang, Yin-Leng Theng, Yi Ding
TL;DR
This study tackles forecasting mental health changes from passive smartphone sensing, benchmarking traditional ML, DL, and LLM methods on the College Experience Sensing (CES) dataset. It introduces a unified evaluation framework that varies temporal windows, feature granularity, personalization, and class-imbalance handling to understand each approach's strengths and limitations. The results show Transformer-based DL models achieve the best overall performance (Macro-F1 around 0.58), while LLMs provide contextual reasoning but lag in numeric temporal modeling; personalization substantially boosts forecasts for severe states. The findings inform the design of next-generation, adaptive mental health technologies that are data-efficient, person-centered, and capable of early, proactive intervention in real-world settings.
Abstract
Smartphone sensing offers an unobtrusive and scalable way to track daily behaviors linked to mental health, capturing changes in sleep, mobility, and phone use that often precede symptoms of stress, anxiety, or depression. While most prior studies focus on detection that responds to existing conditions, forecasting mental health enables proactive support through Just-in-Time Adaptive Interventions. In this paper, we present the first comprehensive benchmarking study comparing traditional machine learning (ML), deep learning (DL), and large language model (LLM) approaches for mental health forecasting using the College Experience Sensing (CES) dataset, the most extensive longitudinal dataset of college student mental health to date. We systematically evaluate models across temporal windows, feature granularities, personalization strategies, and class imbalance handling. Our results show that DL models, particularly Transformer (Macro-F1 = 0.58), achieve the best overall performance, while LLMs show strength in contextual reasoning but weaker temporal modeling. Personalization substantially improves forecasts of severe mental health states. By revealing how different modeling approaches interpret phone sensing behavioral data over time, this work lays the groundwork for next-generation, adaptive, and human-centered mental health technologies that can advance both research and real-world well-being.