PulseAugur
EN
LIVE 20:40:47
research · [2 sources] ·

TimeSRL uses RL-tuned LLMs for generalizable mental health predictions

Researchers have developed TimeSRL, a novel two-stage LLM framework designed for generalizable time-series behavioral modeling, particularly in mental health applications. This framework first abstracts raw data into natural language concepts, then predicts outcomes solely from these semantic abstractions, aiming to improve cross-dataset generalization. Optimized using Group Relative Policy Optimization (GRPO) and Reinforcement Learning from Verifiable Rewards (RLVR), TimeSRL demonstrates state-of-the-art performance in predicting anxiety and depression, significantly outperforming existing ML and LLM baselines. AI

Summary written by gemini-2.5-flash-lite from 2 sources. How we write summaries →

IMPACT Introduces a novel approach for improving LLM generalization in time-series analysis, with potential applications beyond mental health.

RANK_REASON Publication of a new research paper detailing a novel framework and its performance on specific benchmarks.

Read on arXiv cs.AI →

COVERAGE [2]

  1. arXiv cs.AI TIER_1 · Yuang Fan, Lilin Xu, Millie Wu, Jingping Nie, Qingyu Chen, Yuzhe Yang, Zhuo Zhang, Xin Liu, Subigya Nepal, Xiaofan Jiang, Xuhai "Orson" Xu ·

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

    arXiv:2605.21295v1 Announce Type: cross Abstract: 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 re…

  2. arXiv cs.AI TIER_1 · Xuhai "Orson" Xu ·

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

    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…