One of the biggest challenges in healthcare today isn’t the lack of data, it’s what happens when that data fails to come together when it matters most.
At the Society of Physician Entrepreneurs’ first-ever Physician AI Hackathon hosted at UC San Diego, that challenge became the foundation for an award-winning idea.
Led by Poorva Bedmutha, a Ph.D. student in Computer Science at UC San Diego and a machine learning researcher at the Jacobs Center for Health Innovation (JCHI) and UC San Diego Health, a multidisciplinary team built Manifesto AI, a prototype of a perioperative coordination platform designed to transform fragmented clinical information into real-time, team-based workflow intelligence.
The result: a first-place hackathon win, and a compelling vision for the future of surgical care.
During surgery, clinicians are constantly synthesizing patient history, medications, labs, procedural risks, and team communication, often across disconnected systems and under intense time pressure. This leads to cognitive overload, fragmented communication and avoidable delays.
Manifesto AI tackles this differently. Instead of adding more alerts or dashboards, it transforms existing clinical information into a real-time, patient-specific coordination layer for the entire surgical team.
“We needed a way to bring the right information together, at the right time, for the right people. Instead of adding more alerts or more data, we could reduce cognitive burden by coordinating existing information in a patient-specific way,” Bedmutha says.
The idea took shape during early brainstorming, when the team, coming from different clinical and technical backgrounds, was still exploring directions.
“Amidst all of the other ideas being still discussed, about creating new systems in healthcare, I was looking for solutions that we can integrate into existing workflows,” she says.
For Bedmutha, inspiration came from The Checklist Manifesto by Atul Gawande, which highlights how simple checklists can dramatically improve outcomes in complex, high-stakes environments.
“That got me thinking about healthcare workflows. Hospitals already use checklists and standardized protocols because they work. But real patients are rarely ‘standard,’” Bedmutha says. “For a single operation, surgeons, nurses, anesthesiologists, etc., all of them still have to mentally integrate patient history, medications, risks, lab results, and team communication, often under significant time pressure.”
That insight led to a key realization: instead of redesigning healthcare workflows, AI should help coordinate them more intelligently.
“We needed a way to bring the right information together, at the right time, for the right people,” she says. “Instead of adding more alerts or more data, we could reduce cognitive burden by coordinating existing information in a patient-specific way. That’s when Manifesto AI really started to take shape.”
For Bedmutha, the motivation is clear: apply computer science to problems where impact is immediate and human.
“Healthcare generates enormous amounts of information, but clinicians still spend significant effort synthesizing it under time pressure. That’s where AI can help, not by replacing expertise, but by making it easier to apply,” she says. “Manifesto AI is one example of how I think about healthcare innovation: not creating more data, but helping people make better use of the information they already have.”
