clinical-reasoning

Clinical Quality Language (CQL) is an HL7 specification for the expression of clinical knowledge that can be used within both the Clinical Decision Support (CDS) and Clinical Quality Measurement (CQM) domains. This repository contains complementary tooling in support of that specification.

CQF Clinical Reasoning on FHIR for Java

Official Codebase for "ER-Reason: A Benchmark Dataset for LLM-Based Clinical Reasoning in the Emergency Room"

Project COLON-X (Shaping the neXt frontier in intelligent COLONoscopy)

Tooling for CQL and IG Authors

Claude Code skill for multi-persona Bayesian diagnostic reasoning using NEJM Clinical Problem-Solving format with likelihood ratios, evidence-based medicine, and medical AI education

Graph-based self-evolving memory for experience-augmented clinical reasoning with applicability-aware retrieval and online memory evolution.

Clinical Reasoning, Allocation, and the Mathematics of Moral Injury — Part of the Geometric Series (Bond 2026)

This experimental application serves an open source large language model Llama-2 that is connect to a streamlit user interface. The user can define a list of custom symptoms or ICD-10 symptoms as input to the LLM. A two-step chain of prompts will output a list of differential diagnoses followed by a list of examinations to workup these diagnoses.

AI medical case generator for medical education: LLM-built synthetic patient cases, anamnesis and blinded diagnostic workup from ICD-11. REST + NATS.

AI-Powered Medical Simulation Platform with 10k+ USMLE Cases. Interactive LIS & Clinical Decision Support.

🏥 Clinical Decision Support System for Pulmonary Embolism - An agent-based AI system using LLMs and RAG to assist healthcare professionals with evidence-based PE diagnosis and treatment recommendations.

🩺 RAGnosis — An AI-powered clinical reasoning assistant that retrieves real diagnostic notes (from MIMIC-IV-Ext-DiReCT) and generates explainable medical insights using Mistral-7B & FAISS, wrapped in a clean Gradio UI. ⚡ GPU-ready, explainable, and open-source.

MedAgent-Pro reimplementation + AdaptiMed (guideline-grounded adaptive depth). CAGED coming soon.

Does increasing GPT-5.2 reasoning effort improve diagnosis accuracy enough to justify the token/latency cost? Ablation study on 897 paired medical cases.

Claude Agent Skill for clinical reasoning: differential diagnosis, GRADE evidence appraisal, a 13-perspective clinical board, and treatment planning. Every citation is resolved and checked for retraction. Calibrated confidence, traceable output, bilingual EN/ES. Decision-support drafts, never a diagnosis.

PBL medical teaching toolkit | 7 functions: 3-act case generation, session design, basic-clinical bridging, tutor guide, student worksheet, assessment, flowchart

Uncertainty & Confidence Management (UCM): A healthcare AI benchmark suite for uncertainty recognition, justification boundaries, confidence calibration, proportionate action, and reassessment.

Enterprise-grade AI Clinical Intelligence Platform for evidence-based clinical reasoning, differential diagnosis, medical research, and decision support.

Research fork of KnowGuard: A* interactive clinical reasoning on ioMEDQA (PrimeKG+Hetionet, clinical RAG, NIM Ultra). Partial run ~71% ACC vs paper 70.98%.