HCPI-HRL: Human Causal Perception and Inference-driven Hierarchical Reinforcement Learning.

The dependency on extensive expert knowledge for defining subgoals in hierarchical reinforcement learning (HRL) restricts the training efficiency and adaptability of HRL agents in complex, dynamic environments. Inspired by human-guided causal discovery skills, we proposed a novel method, Human Causal Perception and Inference-driven Hierarchical Reinforcement Learning (HCPI-HRL), designed to infer diverse, effective subgoal structures as intrinsic rewards and incorporate critical objects from dynamic environmental states using stable causal relationships. The HCPI-HRL method is supposed to guide an agent's exploration direction and promote the reuse of learned subgoal structures across different tasks.

Linking guidelines to Electronic Health Record design for improved chronic disease management.

The promise of electronic decision support to promote evidence based practice remains elusive in the context of chronic disease management. We examine the problem of achieving a close relationship of Electronic Health Record (EHR) content to other components of a clinical information system (guidelines, decision support and workflow), particularly linking the decisions made by providers back to the guidelines. We use the openEHR architecture, which allows extension of a core Reference Model via Archetypes to refine the detailed information recording options for specific classes of encounter.