Research map / Large language models / AI and society
Towards Trustworthy Automatic Diagnosis Systems by Emulating Doctors' Reasoning with Deep Reinforcement Learning
Arsene Fansi Tchango, Rishab Goel, Julien Martel, Zhi Wen, Gaetan Marceau Caron, Joumana Ghosn
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Train to predict the differential instead of only the ground truth pathology | CASANDE · DDXPlus test | DDF1: 30.84 -> 94.24 | Improved |
| Use CASANDE instead of Diaformer when trained with differential | CASANDE · DDXPlus test | DDF1: 83.3 -> 94.24 | Improved |
| Use CASANDE instead of the best baseline when trained with differential | CASANDE · DDXPlus test | DSHM: 69.43 -> 73.88 | Improved |
| Use CASANDE instead of AARLC | CASANDE · DDXPlus test | PER: 54.55 -> 98.39 | Improved |
| Use CASANDE instead of BASD, which has shorter interactions | CASANDE · DDXPlus test | interaction length: 17.86 -> 19.71 | Hurt |
| Train AARLC with differential versus without | AARLC · DDXPlus test | PER: 32.78 -> 54.55 | Improved |
| Train BASD with differential versus without | BASD · DDXPlus test | PER: 88.45 -> 88.18 | No clear effect |
| Add all rewards but remove classification reward R_Cl (ablation) | CASANDE · DDXPlus validation | DSHM: 73.55 -> 74.19 | Improved |
| Add exploration, confirmation and severity rewards (ablation) | CASANDE · DDXPlus validation | DDF1: 93.11 -> 94.26 | Improved |
| Add only classification reward R_Cl (ablation) | CASANDE · DDXPlus validation | DDF1: 93.11 -> 93.48 | No clear effect |
| Add exploration and confirmation rewards (ablation) | CASANDE · DDXPlus validation | interaction length: 17.56 -> 19.71 | Mixed |
| Add severity reward R_Sev (ablation) | CASANDE · DDXPlus validation | DSHM | Improved |
Closest papers by meaning
- MedExAgent: Training LLM Agents to Ask, Examine, and Diagnose in Noisy Clinical Environments (NeurIPS 2026 · similarity 0.86)
- Doctor-R1: Mastering Clinical Inquiry with Experiential Agentic Reinforcement Learning (ICLR 2026 · similarity 0.85)
- Simulating Viva Voce Examinations to Evaluate Clinical Reasoning in Large Language Models (NeurIPS 2025 · similarity 0.85)
- Patho-R1: A Multimodal Reinforcement Learning-Based Pathology Expert Reasoner (AAAI 2026 · similarity 0.85)
- DDx-TRACE: A Benchmark for Medical Diagnostic Trajectories in VLMs (NeurIPS 2026 · similarity 0.84)
- MedEyes: Learning Dynamic Visual Focus for Medical Progressive Diagnosis (AAAI 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
