Sharing "Neuropathic Pain Diagnosis Simulator for Causal Discovery Algorithm Evaluation", one of the most popular datasets used for evaluating causal discovery algorithms! 🩺🧠 @RuiboTu @kunkzhang 📄 arxiv.org/abs/1906.01732 💻github.com/TURuibo/Neuropath… 1/n

🏥 The Motivation: In Causal Discovery, evaluating if X -> Y is correct is difficult because we lack gold-standard labels for complex systems. This simulator uses well-studied biomedical knowledge of neuropathic pathophysiology to create that "Gold Standard."

🛠️ The Solution: A high-fidelity simulator built on clinical expertise (e.g., Radiculopathy, Carpal Tunnel). - 200+ variables (symptoms, physical exams, diagnostics). - 800+ expert-validated causal edges. - Realistic noise and complex non-linear relationships.

🧪 Testing the SOTA: The authors show that even top algorithms (PC, FCI, GES) struggle with this level of clinical logic, highlighting the need for more robust discovery methods.