# TimeSRL: Generalizable Time-Series Behavioral Modeling via Semantic RL-Tuned LLMs -- A Case Study in Mental Health

arXiv: https://arxiv.org/abs/2605.21295 · HTML: https://mlsanity.com/paper/2605.21295/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 16 of its 72 references that are in the map.
- First posted on arXiv 20 May 2026.

## Builds on (most cited first, 16 of 16)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Concept Bottleneck Models](https://mlsanity.com/paper/2007.04612/) (2020)
- [SWIFT:A Scalable lightWeight Infrastructure for Fine-Tuning](https://mlsanity.com/paper/2408.05517/) (2024)
- [Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs](https://mlsanity.com/paper/2506.14245/) (2025)
- [Mental-LLM: Leveraging Large Language Models for Mental Health Prediction via Online Text Data](https://mlsanity.com/paper/2307.14385/) (2023)
- [MindScape Study: Integrating LLM and Behavioral Sensing for Personalized AI-Driven Journaling Experiences](https://mlsanity.com/paper/2409.09570/) (2024)
- [LLM-based Conversational AI Therapist for Daily Functioning Screening and Psychotherapeutic Intervention via Everyday Smart Devices](https://mlsanity.com/paper/2403.10779/) (2024)
- [SleepLM: Natural-Language Intelligence for Human Sleep](https://mlsanity.com/paper/2602.23605/) (2026)
- [OSF: On Pre-training and Scaling of Sleep Foundation Models](https://mlsanity.com/paper/2603.00190/) (2026)
- [HEARTS: Benchmarking LLM Reasoning on Health Time Series](https://mlsanity.com/paper/2603.06638/) (2026)
- [MoodCapture: Depression Detection Using In-the-Wild Smartphone Images](https://mlsanity.com/paper/2402.16182/) (2024)
- [ProMind-LLM: Proactive Mental Health Care via Causal Reasoning with Sensor Data](https://mlsanity.com/paper/2505.14038/) (2025)
- [Daily Stress Recognition from Mobile Phone Data, Weather Conditions and Individual Traits](https://mlsanity.com/paper/1410.5816/) (2014)
- [DailyLLM: Context-Aware Activity Log Generation Using Multi-Modal Sensors and LLMs](https://mlsanity.com/paper/2507.13737/) (2025)
- [AWARE Narrator and the Utilization of Large Language Models to Extract Behavioral Insights from Smartphone Sensing Data](https://mlsanity.com/paper/2411.04691/) (2024)

## Built on by (most cited first, 2 of 2)

- [Evidence-Grounded Mapping of Multimodal Human Sensing Psychological Transdiagnostic Dimensions](https://mlsanity.com/paper/2608.24903/) (2026)
- [LineupRL: Verifiable Reinforcement Learning for Time Series Captioning via Caption-to-Series Identification](https://mlsanity.com/paper/2610.01800/) (2026)

## Abstract (arXiv)

Longitudinal passive sensing enables continuous health prediction, yet models often fail under cross-dataset distribution shifts. Traditional ML overfits cohort-specific artifacts, while Large Language Models (LLMs) struggle to reason reliably over long, heterogeneous time-series. We introduce TimeSRL, a two-stage LLM framework that routes predictions through an explicit semantic bottleneck. The model first abstracts raw signals into high-level natural language, then predicts behavioral outcomes from these abstractions alone. This forces the model to reason over semantic concepts that we argue generalize better than raw numbers. We optimize this process end-to-end using Group Relative Policy Optimization (GRPO) with Reinforcement Learning from Verifiable Rewards (RLVR), learning outcome-aligned abstractions without gold intermediate annotations. Instantiated on mental-health prediction, TimeSRL achieves state-of-the-art performance on a benchmark designed to stress-test cross-cohort generalization under a rigorous leave-one-dataset-out (LOSO) protocol, reducing mean absolute error (MAE) over strong non-LLM ML and LLM baselines by 3.1--10.1% and 9.5--44.1% for anxiety, and 3.2--9.6% and 27.4--57.6% for depression (all ps<0.05). TimeSRL significantly outperforms prior methods in cross-benchmark transfer across different sensing pipelines, rivaling its own within-domain performance without target-domain fine-tuning. These results demonstrate that semantic abstractions are reusable and point to a new direction for generalizable behavior modeling via RL-tuned LLMs.