Deutsch 
Participants will gain a comprehensive understanding of SPEs, including their role in enabling safe and efficient use of sensitive health data and supporting FAIR principles. Perspectives from three key stakeholder groups — data users, data holders, and technology providers — are shown to offer insights into their roles, challenges, and priorities. Real-world case studies, practical guidance, and interactive discussions provide insights into the effective implementation and use of SPEs and can initiate a community building process. The sith event in the series will take place on 18 October 2026 at 4.00 pm.
Mapping How Disease Unfolds, Where the Data Live
Electronic health records hold the most detailed longitudinal account of human illness available to research, and the institutions that hold them cannot release them. In this talk I will describe how our group at Massachusetts General Hospital built the first instantiation of the phylophenome, a directed, confidence-ranked map of how clinical phenotypes follow one another over time, entirely inside the protected research environment of Mass General Brigham. Working from de-identified records of 295,678 patients, we mined 435,240 time-ordered disease associations. We then adjudicated them with a medical language model that ran on local servers, and compared the result against genetic correlations from the UK Biobank. Only aggregate transition statistics left the environment, and they now power ClaiD, an open browser of disease progression. I will walk through how each of the five safes applied to this project and what the work looks like from the analyst's chair. I will also show how the same strategy, running a computable phenotype locally at each site and keeping what agrees across sites, carried a Long COVID algorithm across 58 hospitals.
Date & Time: 18 October 2026 | 16:00–17:00
Lecturer: PhD Hossein Estiri | Massachusetts General Hospital & Harvard Medical School
Format: online via zoom
Registration: [here]
