Reproduce “Training language models in simulation for psychiatric intake” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Samuel Schmidgall, Tim Strother, Alex Bijamov, Min Woo Sun, Anil Palepu, Akshay Goel, Ulysse Marteau-Ferey, Théo Moutakanni and 48 more
The simulation-trained model improves psychiatric intake conversations by practicing multi-turn encounters with simulated patients and receiving rewards for clinical reasoning, communication, and safety. In a randomized, double-blind crossover study, it received higher psychiatrist transcript ratings than its identically prompted base model in 46 of 60 matched scenario pairs (76.7%). This supports improvement in controlled text encounters, not real-patient safety or effectiveness.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.