Reproduce “Physiological World Models for Human State Transitions” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chongyang Zhang, Rendong Wang, Hao Zheng, Hanwen Zhang, Yang Liu, Xiaolong Wei, Bin Chong
Researchers propose the Physiological World Model (PWM) framework, defining a human-centered, event-conditioned approach to model how an individual's integrated physiological state dynamically transitions. This framework outlines a structured representation for physiological data, introduces the HumanState Transition Token for event-anchored learning, and establishes progressive capability levels, data acquisition protocols, and benchmark tasks for its development and evaluation.
Key insights
- The framework shifts the focus from static physiological analysis to dynamic, event-conditioned state transition modeling, moving beyond mere state recognition to understanding how and why states change.
- The 'HumanState Transition Token' provides a critical mechanism for structuring diverse, fragmented data into actionable units, allowing for learning from specific, high-quality event-response sequences.
- Establishing progressive capability levels (L1-L4) and corresponding data acquisition protocols (P1-P4) offers a structured pathway for developing and rigorously validating complex physiological world models.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.