Search papers, labs, and topics across Lattice.
MediSkill-Evo is an advanced clinical agent designed to evolve process knowledge while adhering to evidence and care-process constraints without requiring backbone fine-tuning. By organizing experience into four distinct banks and utilizing a Process-Constrained Preference Harness, it effectively enhances diagnosis accuracy from 61.33% to 69.00% and treatment-intent coverage from 33.62% to 66.44%. The agent also significantly reduces critical failures from 31.00% to 16.33% across various stress conditions, demonstrating its robustness in clinical interactions under pressure.
MediSkill-Evo not only boosts diagnostic accuracy but also ensures adherence to clinical processes, achieving a remarkable 93.61% recovery rate under patient-behavior pressure.
Interactive clinical agents must gather decisive evidence and convert it into grounded actions under partial observability. A correct final diagnosis alone does not show that an agent respected evidence and care-process constraints. We introduce MediSkill-Evo, a clinical agent that evolves governed process knowledge without backbone fine-tuning. It separates experience into four typed banks for clinical skills, process rules, symbolic schemas, and measurement procedures. Provenance, support, replay, and controller-defined safety checks govern publication to a frozen test-time snapshot. A Process-Constrained Preference Harness binds evidence to its source, rejects controller-invalid candidates, and ranks actions with a safety-prioritized Clinical Process Critic. We evaluate complete agent systems across two backbone endpoints and six controlled stress dimensions under the same Doctor-turn limit. On 300 held-out Qwen encounters, MediSkill-Evo improves diagnosis accuracy from 61.33 percent to 69.00 percent and treatment-intent coverage from 33.62 percent to 66.44 percent, while reducing automatically scored critical failures from 31.00 percent to 16.33 percent relative to AgentClinic. On 180 hard-isolation conditions derived from 30 cases, target recovery reaches 93.61 percent under patient-behavior pressure, 100.00 percent for temporal evidence, and 92.22 percent for triage red flags. An exploratory 100-case MedSAM comparison evaluates request-gated tool-interface feasibility. These results provide descriptive end-to-end evidence for the complete system on fixed evaluation suites, not causal evidence for an individual bank or clinical validation of the automatic judge.
