The MIT–MGB AI Cures Conference
The MIT–MGB AI Cures Conference, organized by the MIT Jameel Clinic for Machine Learning in Health with MIT and Mass General Brigham (MGB), brings together the two communities to showcase collaborative clinical-AI projects and their effect on patient care. It also looks at how the data used to train clinical algorithms should be collected, with attention to privacy, diversity, and quality. The 2023 conference was held in person on April 24 at MIT’s Samberg Conference Center, with sessions on the clinical translation of AI tools, fairness and equity, algorithmic and regulatory challenges, secure access to patient data, and new MIT–MGB collaborations, and a poster session.
Clinical AI, explained
Clinical AI refers to machine-learning models that are trained on health records, images or other patient data to support decisions such as estimating a patient’s risk or predicting how a treatment may work. Moving a model from a research paper into a hospital raises questions that the program reflects: whether it works equally well for different groups of patients, whether the data behind it were gathered with appropriate privacy protection, and how regulators and clinicians can check its behavior. The audience photograph on this page shows a full conference room during a talk, with a slide on domain randomization, a technique for training models to cope with variation in their input data.
Dr. Costilla Reyes at the conference
On the day of the conference, Dr. Omar Costilla Reyes wrote that he was ready to start the day and invited colleagues to learn more about his research in the space at the poster session. He also worked with Massachusetts General Hospital’s Center for OCD and Related Disorders on interpretable models for predicting treatment outcomes. He returned the following year, writing that he was glad to explore synergies and collaborations.