BRIDGE-SIC is a pilot study to determine the accuracy and feasibility of applying LLMs to improve identification, summarization and communication around goals of care. Upon successful completion, a larger study will determine whether this intervention has an impact on the care that patients receive during and after hospitalization.
Christopher Manz, MD, is the sponsor-investigator and study Co-Principal Investigator. Dr. Manz is an Assistant Professor in Medical Oncology at Dana-Farber in the Gastrointestinal Cancer Center and Division of Population Sciences. He has experience implementing pragmatic clinical trials to improve serious illness conversations at the University of Pennsylvania Health System and Dana-Farber. Charlotta Lindvall, MD, PhD, is the Co-Principal Investigator. Dr. Lindvall is a palliative care physician and researcher at Dana-Farber and Harvard Medical School, in Boston. She leads a cross-disciplinary research team of physicians, nurses, and computer scientists to develop AI to extract patient-centered outcomes from clinical narratives and conversations, and has research experience applying LLMs to summarize SICs. She additionally serves as Senior Advisor to Clinical Informatics at Dana-Farber and as technical expert on AI for the National Quality Forum. Manuel Fanarjian, MD, is a palliative care physician and clinical informaticist at Dana-Farber, as well as an Instructor in Medicine at Harvard Medical School. He provides clinical informatics leadership, in concert with the hospital’s Information Services team.
Dr. Manz noted that, “Patients with cancer often receive care near the end-of-life that may not be beneficial or accordant with their preferences. Serious Illness Conversations (SIC) between patients and clinicians improve well-being and reduce aggressive care near the end-of-life by identifying and matching care to patient goals and preferences. Because SIC documentation is often buried in voluminous free-text EHRs, a key challenge in ensuring high-quality, goal-concordant care in acute hospitalization settings is effectively identifying and communicating patient preferences to clinical teams providing inpatient care. LLMs are helping us to drastically reduce the time and effort needed to search for documentation of these conversations.”
“We believe deeply in the goal of every patient having their wishes front and center,” said Dr. Lindvall. “We are on track to show that LLMs for real-time identification of Serious Illness Conversations work and represent a major breakthrough in our field.”
Dana-Farber has implemented HDAI’s HealthVision platform, viewable in the EHR, applying validated prediction models to stratify patients at highest risk of 90-day mortality for inclusion in the trial and a secure, performant LLM infrastructure that creates SIC summaries in seconds, not minutes. For eligible patients in the trial, these summaries are sent to the inpatient and outpatient clinical teams for patients being admitted to the hospital.
“As I’ll share on the stage at HLTH this week, academic medical centers like Dana-Farber are home to some of the most advanced clinical research in the world, yet they often face significant challenges when trying to connect patients with palliative care resources – challenges rooted in care complexity, clinician time and fragmented data,” said Nassib Chamoun, founder and CEO of HDAI. “We are truly honored to be part of this important work that will pave the way for better care alignment with patients with serious illness and as an exemplar for other institutions to emulate.”
The results will be compiled in early 2026 and presented at conferences and in publications by the Fall of next year.
About previous studies from this team
Feasibility Study for Using Large Language Models to Identify Goals-of-Care Documentation at Scale in Patients With Advanced Cancer.
Agaronnik ND, Davis J, Manz CR, Tulsky JA, Lindvall C.JCO Oncol Pract. 2025 Apr 10:OP2400992. doi: 10.1200/OP-24-00992.
Large Language Models to Identify Advance Care Planning in Patients With Advanced Cancer.
Agaronnik ND, Davis J, Manz CR, Tulsky JA, Lindvall C.J Pain Symptom Manage. 2025 Mar;69(3):243-250.e1. doi: 10.1016/j.jpainsymman.2024.11.016.
About Health Data Analytics Institute (HDAI)
HDAI is focused on addressing the US’s looming healthcare crisis with actionable, responsible AI tools that help improve care outcomes, financial results and clinician satisfaction for healthcare systems and value-based care organizations. HDAI’s HealthVision™ is software that combines hundreds of high-performing, pre-built predictive models with LLM-enabled chart summarizing capabilities and a powerful, cost-effective, scalable cloud computing platform. HealthVision is embedded in the EHR and provides AI-enabled, use-case specific, patient stratification in real-time across the continuum of care. These intelligent rosters make it easier to align scarce resources with the patients at the time they are most needed. At the patient level, HealthVision synthesizes hundreds of encounters into one-page AI Summaries that help clinicians quickly understand patient needs, replacing tedious, time-consuming searches. Recognized in the 2025 TIME Top 100 HealthTech companies list.
Media Contact: Carola Endicott, HDAI, 617-699-0725
SOURCE Health Data Analytics Institute
