BILH Hosts AI and Machine Learning System-Wide Symposium
Advancing Medicine: Highlights from the BILH 2025 AI and ML Symposium
BOSTON — The Beth Israel Lahey Health (BILH) 2025 Artificial Intelligence (AI) and Machine Learning (ML) Symposium brought together leading experts in the field of AI and ML, and researchers and clinicians for a thought-provoking event held at Harvard Medical School’s Joseph B. Martin Center. Through a program of speakers, moderated panels and poster presentations, more than 450 registrants from across BILH explored the growing role of AI and ML in medicine, their potential uses in research, patient care and medical education, and the complex ethical considerations surrounding their equitable application as well as issues of patient privacy and data encryption.
“At BILH we are particularly excited about the opportunities to bridge the gaps between technology, research, and patient care,” said Gyongyi Szabo, MD, PhD, Chief Academic Officer of BILH and Beth Israel Deaconess Medical Center (BIDMC). “This symposium serves as a platform for us to learn from each other—to share ideas, push boundaries, and, most importantly, to collaborate.”
“Today’s event is a great example of how individuals, departments and organizations across BILH are coming together to share their knowledge and expertise with the goal of advancing our academic mission and further strengthening the care we provide to our patients and the communities we serve,” said Kevin Tabb, MD, President and CEO of BILH, in welcoming remarks via video.
The keynote speaker, Peter Szolovits, PhD, Professor of Computer Science and Engineering and head of the Clinical Decision-Making Group of MIT’s renowned Computer Science and Artificial Intelligence Laboratory (CSAIL), provided an overview of his five decades in the field. A pioneer in using AI to enhance clinical decision-making, he recalled that in 1970, computer scientists predicted artificial intelligence, capable of providing a “built-in second opinion for every decision,” would completely revolutionize medicine and medical education by the year 2000. Only the timing, he said, was not correct. A professor at MIT since 1974, Szolovits has been a major contributor to the field, helping to develop AI models capable of diagnosing single acute diseases in the 1970s to models capable of integrating electronic health records by the 1990s, and now aiming to incorporate speech-to-text medical documentation and enhance ICU data analysis, all with the goal of improving patient outcomes and advancing personalized medicine.
The first of two expert panels moderated by Jennifer Stevens, MD, MS, Director of BIDMC’s Center for Healthcare Delivery Science, focused on how to responsibly implement AI and ML in healthcare. The discussion featured insights from Lori Dutcher, JD, BILH’s Chief Compliance Officer; Steve Horng, MD, Clinical Lead for Machine Learning at BIDMC’s Center for Healthcare Delivery Science; David Sontag, JD, BILH Senior Associate General Counsel and Director of Ethics; Venkat Jegadeesan, Vice President of BILH Technology and Innovation; and MIT’s Szolovits. The panelists addressed issues such as data privacy, bias mitigation, and regulatory challenges, underscoring the need for careful oversight as AI’s role in medicine expands.
Stevens pressed the panelists about the well-known danger of human biases finding their way into the models.
“At the end of the day, what we actually care about is whether it is going to perform well,” said Horng. “Have we done enough testing to make sure that it isn’t missing some key information, that it isn’t biased against some populations?”
“What’s your approach to evaluate these models, what are best practices to mitigate bias?” Stevens asked Horng.
“We have to have a concrete plan ahead of time to come up with an objective way of measuring bias,” said Horng. “If we don’t measure it, how do we know we have it, and how can we know we’ve eliminated it? We need to develop a test set of objective criteria among a large set of attributes.”
BILH’s Information Technology department was a major participant in the Symposium. “As a developer and not a clinician, my team focuses on building AI models to optimize hospital operations and enhance the overall patient experience,” said Jegadeesan.
“Responsible AI starts with a fundamental question: how will it be used? We developed a tool that predicts whether a given patient will show up. But we have two options with how we use the information that a patient is likely to miss an appointment; we can double book, but it’s more responsible to send patient reminders and address other barriers to coming in to try to reduce healthcare disparities. We want to make sure our decisions lead to patient-centered outcomes.”
The second panel discussion focused on how members of the BILH community can find support to launch their own AI/ML projects. The discussion featured members of the BILH Innovation Lab, a team dedicated to revolutionizing healthcare delivery through strategic AI integration. Their mission includes enhancing patient care, advancing scientific discovery, improving employee experience, and optimizing operations. Panelists including the Innovation Lab’s John Shang, Director of Data Engineering and EHR Services, and Enterprise Architects Si Wong and Phanidhar Adusumilli, shared their experiences with, and practical insights on, initiating AI and ML projects within BILH, highlighting available resources and key considerations for successful implementation.
BIDMC interventional radiologist Seth Berkowitz, MD, and transplant surgeon Jennifer Ho, MD, shared their experiences working with BILH’s Innovation Lab to improve patient care. Berkowitz wanted a better way to determine the optimal timing to recall patients for follow-up colonoscopies; Ho is currently working with the group on a model to better identify patients with complex heart failure and the best course of treatment for them.
The event concluded with a forward-looking keynote by Rowland Illing, MD, Chief Medical Officer and Director of Global Healthcare at Amazon Web Services (AWS). His talk focused on the importance of the cloud to AI and ML healthcare technologies. The cloud, he emphasized, offers scalability and agility that other technological platforms cannot compete with.
The day also featured a competitive poster competition. The BILH-wide multi-disciplinary scientific program committee judged fifty digital poster presentations across four categories. A top entry and a runner-up were selected from each category. Congratulations to the following poster presenters and principal investigators (PI).
First place: Aman Mohapatra, MD (presenter), and Joseph D. Feuerstein, MD (PI) System-wide implementation of a large language model workflow for colonoscopy recall inference
Runner up: Shreyas Puducheri, Olivia T. Zhou, BS; Krish Kapadia, BS (presenter), and Juan E. Small, MD and Vijaya Kolachalama, PhD (PI) for AI-Augmented Assessment of Multisequence MRIs for Alzheimer’s Disease and Related Dementias
First place: Konstantinos Stefanakis, MD (presenter), and Christos S. Mantzoros, MD DSc (PI) for Accurate non-invasive detection of Metabolic Dysfunction-Associated Steatohepatitis with fibrosis stages F2-F3 using lightweight clinical and metabolomics-based categorical gradient boosting models: a first-in-class approach in the new FDA guidelines era
Runner Up: Haadi Mombini, PhD (presenter), and Venkat Jagadeesan for Radiology Prediction with AI
First Place: Jean Filo, BS (presenter), and Christopher S. Ogilvy, MD (PI) for Development of the First Externally Validated Brain Aneurysm Detection AI Platform for CT Angiography Scans Reporting Generalizable Performance
Runner Up: Kristyn Beam, MD (presenter/PI) for NeoCLIP: A Self-Supervised Foundation Model for the Interpretation of Neonatal Radiographs
First place: Diego Trujillo, BS (presenter), and Jeanne-Marie Guise, MD, MPH, MBA (PI) for SAFE-AI: A medically grounded AI method to identify patient safety events in healthcare
Runner Up: Mohammed Yamin, BS (presenter), and Ryan Cauley, MD, MPH (PI) for Machine Learning to Predict the Risk of Postoperative Wound Complications in Open Spine Surgery: A Prediction Model for High-risk Patients
About Beth Israel Lahey Health
Beth Israel Lahey Health is a health care system that brings together academic medical centers and teaching hospitals, community and specialty hospitals, more than 4,700 physicians and 39,000 employees in a shared mission to expand access to great care and advance the science and practice of medicine through groundbreaking research and education.
