Phesi built an AI digital twin from real-world data to replace the standard-of-care control arm in a cGvHD trial.
Control arms in conditions like cGvHD are hard to recruit because half of participants get the comparator. Phesi built a digital twin from its Trial Accelerator platform (108 million patients, nearly 367,000 cohorts). The proof of concept replicated outcomes on prednisone, using 2,042 patients from 32 cohorts, and was published in Bone Marrow Transplantation.
Phesi is a clinical development analytics firm whose Trial Accelerator platform and digital patient profiles are used by sponsors for trial design and site selection
This sits in the Build stage of the AI use-case lifecycle: Named examples from the foundation-model labs and platforms pharma teams use to actually construct, train and generate the model or application.
Published by Phesi on its own site — read the original case study on technologynetworks.com ↗
Note: Proof of concept reported in trade press, not a customer deployment.
Related use cases
- 01OpenAI's dedicated healthcare vertical page/launch post.
- 02Frontier reasoning model built for biology, drug discovery and translational medicine; named qualified customers include Amgen, Moderna, Allen Institute, and Thermo Fisher Scientific.
- 03follow-up update to the life-sciences model.
- 04"How Amgen uses GPT-5": Amgen applies GPT-5/frontier models to help accelerate delivery of potential new therapies; found GPT-5 met scientific-accuracy standards.
- 05Moderna and OpenAI partner to accelerate development of treatments; ChatGPT Enterprise deployed to thousands of employees company-wide.
- 06Longevity biotech collaborated with OpenAI to build and research GPT-4b micro, a version of GPT-4o specialized for protein engineering.