MEDIA INQUIRIES
Oops! Something went wrong while submitting the form.
When the first sequence of the human genome project was drafted 25 years ago, few could have predicted how transformative it would be for biomedicine. "Today's young researchers cannot imagine doing biomedical research without having the foundation of the human genome," said Todd Golub, Director and Founding Core Institute Member of the Broad Institute of MIT and Harvard, in his opening remarks. Golub compared that turning point from the human genome project to the present moment of AI and the life sciences: while the full potential of AI in biomedical research is not fully known yet, the field could look remarkably different a generation from now.
The Eric and Wendy Schmidt Center Symposium on Biomedical Science and AI showcased research at the forefront of this interdisciplinary field and offered a glimpse of the exciting paths ahead.
Launched in 2021, the Schmidt Center is enabling a new field of research at the intersection of machine learning and biology, aimed at improving human health. The Center's inaugural symposium, held on April 30 and May 1 at the Koch Institute For Integrative Cancer Research at MIT, brought together researchers and industry leaders who are advancing the foundations of machine learning and using these tools to understand the programs of life across scales, from proteins to cells to tissues and organisms.
More than 250 in-person and 400 virtual attendees joined 27 speakers and panelists from across the Broad, MIT, Harvard, and beyond to learn how machine learning is uncovering insights into today’s most pressing biological questions, and in turn, how these questions are inspiring new directions in AI.
“We strongly believe that the biomedical sciences are not only well-positioned to benefit from machine learning, but they also offer some of the most exciting inspiration for foundational advances in ML,” said Caroline Uhler, Director of the Schmidt Center and Andrew and Erna Viterbi Professor of Engineering at MIT.
At the start of the symposium, Uhler noted that there had already been great energy – “I’m really looking forward to two exciting days of stimulating research and discussions,” she said.
The symposium included a wide range of panels, talks, and poster presentations. The talks represented the breadth of the field, including using imaging and AI to understand the spatial organization of a tissue, molecular dynamics and functions at the subcellular level, and active drug discovery.
A common theme throughout the symposium was that science is at a pivotal moment, where the AI revolution will undoubtedly transform how researchers study the fundamental laws of life. Several discussions centered on whether and how a holistic foundation model for biology across scales, similar to ChatGPT in the language domain, could eventually be created, and what biological insights could be gained in the meantime from available data and simpler, more cost-effective machine learning frameworks.
The two panel discussions that concluded each of the days were complementary: one showcased experimentalists working on large-scale biological data generation, and the other offered insights from computational experts on foundation models across all scales (molecules, cells, tissues, organisms).
The symposium brought together well-established research leaders with researchers in the early stages of their independent careers, postdoctoral fellows, and PhD students.
The 11 invited talks featured Jennifer Lippincott-Schwartz (HHMI), Jean-Philippe Vert (Owkin, Bioptimus), GV Shivashankar (ETH Zurich), Susanne Rafelski (Allen Institute for Cell Science), Dana Pe’er (Memorial Sloan Kettering Cancer Center), Richard Bonneau (Genentech), Emma Lundberg (Stanford), Maria Brbic (EPFL), Anshul Kundaje (Stanford), Eric Xing (GenBio AI, CMU, MBZUAI), and Patrick Hsu (Arc Institute, UC Berkeley).
During the panels, discussions were led by Shantanu Singh (Broad), Xiaowei Zhuang (Harvard, HHMI), Fei Chen (Broad), Mark Daly (Broad, HMS, MGH), Sergey Ovchinnikov (MIT), Faisal Mahmood (HMS), Shirley Liu (GV20 Therapeutics), and Marzyeh Ghassemi (MIT), moderated by Eric Lander (Broad, MIT) and Caroline Uhler (Broad, MIT).
Joining them were 10 early career researchers – Athanasios Litsios (University of Toronto), Bo Xia (Broad, Harvard), Xinyi Zhang (Broad, MIT), Yakir Reshef (BWH, Broad), Yichen Si (Broad), Joey Bose (University of Oxford), Sandeep Kambhampati (Harvard, Broad), Pinar Demetci (Broad), and Michelle M. Li (HMS).
On the first day, Jennifer Lippincott-Schwartz illustrated how cutting-edge imaging technologies, combined with AI, allows researchers to study molecular dynamics in live cells at sub-cellular resolution, opening new doors to understanding cell function at multiple scales.
Dana Pe'er presented computational methods developed by her lab to extract insights from spatial imaging data, define tissue niches, and help map how tissue architectures relate to function in health and disease.
On the second day, Eric Xing presented the idea of AI-driven “digital organisms” (AIDO) for simulating all biological phenomena. He noted that this could potentially be achieved by integrating and optimizing several modality-specific foundation models that are currently being developed at GenBio.
During the last session of the symposium, a panel on Foundation models in biology: DNA, Protein, Cells, Tissues, Organisms, Shirley Liu shared how AI can help identify targets for personalized cancer immunotherapy. She started from the idea that the immune system naturally produces antibodies to fight cancer, and by analyzing tumor-infiltrating immune cells, scientists can identify the tumor antigens that could amplify the endogenous immune response.
The symposium also gave early-career researchers an opportunity not only to present their posters, but to also participate in flash talks on both days, helping to share their work more widely.
The posters were judged by a committee that included Lindsay Edwards (Relation Therapeutics), Jean-Philippe Vert (Owkin, Bioptimus), and Orr Ashenberg (Broad).
Three poster winners emerged:
The poster presentations reflected the innovative, interdisciplinary nature of the field – blending computational rigor with biological insight.
“The level of science in the last couple of days has been absolutely phenomenal,” said Edwards. “We chose poster winners on the basis of three things: 1) originality; 2) scientific quality; and 3) scientific communication.”
Edwards expanded on the need for strong science communication skills, which includes compelling storytelling, clear visuals, and confident presentation, adding that those who communicate well are more likely to be invited to speak, ultimately amplifying the impact of their work.
During hallways conversations in between sessions, attendees reconnected with old colleagues and forged new collaborations, creating a lively, dynamic environment for idea-sharing across generations of researchers.
“This symposium brought together a lot of different perspectives to illuminate my research,” said Yue Qin, postdoctoral fellow at the Schmidt Center. “Just talking to people during the coffee breaks and lunch chats helped me figure out what is the next area that I should look into.”
Another Schmidt Center postdoctoral fellow – Viki Schuster – shared similar sentiments. “It was great to talk to people, get interesting questions, and discuss my work from different perspectives,” she said.
Looking forward, speakers emphasized the importance of a coordinated global effort to standardize and aggregate datasets that are both consistent and representative of a broad spectrum of diseases. Future research should prioritize gaining novel biological insights, which may sometimes be achieved with simpler models rather than complex and resource-intensive foundation models. Continued work is required to better understand what these models learn, identify and mitigate potential biases, and recognize their limitations.
As the Schmidt Center continues to advance biomedical discoveries and expand the community at the intersection of biology and AI, it looks forward to hosting this symposium annually, and continuing these conversations and collaborations.
“There’s that level of potential to really disrupt and transform how we think about biomedical research in the setting of machine learning and AI,” said Golub. “What a remarkable moment to be in science when, over the next decade, we are all going to figure out what the new world of biology and medicine is going to look like.”
Watch the symposium talks online at broad.io/ewsc-watch-symposium.
To stay updated with the Schmidt Center, follow us on LinkedIn and Twitter and join our mailing list.
