Med Student Builds AI Lifeline for Vulnerable Patients

By Dawn M. Cwierley

There, the second-year student helps providers manage patients’ clinical issues, like stabilizing blood pressure and treating infection. But, addressing patients’ non-medical needs, such as finding an open shelter or a nearby food pantry, is the bigger challenge.

“Throughout the first year and a half of medical school, I saw that there’s a gap in the referral process. We have resources available, but connecting people to the right ones is the challenging part,” noted Fox.

So, he created a simple spreadsheet of local resources that clinicians could keep in exam rooms. As he updated it, he saw how quickly the information became obsolete as phone numbers, staffing and program capacity shifted.

Fox explained that traditional resource lists like these are static and often outdated soon after they’re created. In contrast, AI systems can draw from multiple sources in real time, providing results that are more current and complete. In this context, the role of AI is less about generating content and more about efficiently gathering, updating and organizing information across sites.

He realized there could be a much more powerful solution; he developed, an AI-driven, web-based Community Resource Finder that searches multiple sites in real time, organizes information, and delivers tailored options based on a patient’s needs and zip code. 

Fox noted that the Community Resource Finder is deliberately bare bones. It is a mobile-friendly website that any clinician or patient can open on a phone at the bedside or during an office visit.

Users can search by category, such as food assistance, housing and shelter or medical care, and by ZIP code. Behind the scenes, a large language model sifts through information from numerous public sources, then presents it in a concise, usable format that includes location, hours and eligibility details.

“AI condenses that searching so clinicians can focus on the human part, connecting with patients,” said Fox.

The system was designed to capture only anonymous search queries that can’t be traced back to individual users; the queries are then logged to a secure spreadsheet for quality improvement.

Early analytics, gathered in partnership with the Great Lakes Health System of Western New York’s social work team, show approximately 150 searches spanning 40 unique ZIP codes in the pilot phase.

At Friends of Night People and in similar community settings, clinicians are increasingly turning to the tool. They report that it has streamlined the referral process and made it easier to match patients with services with less delay.

Fox was seeing in real time what medical students are learning nationwide: that a lack of safe shelter, food or transportation can undermine even the best clinical treatment plan and derail a patient’s overall health.

At the Jacobs School, students are trained to recognize how these social realities shape health and to approach patient care through a whole‑person lens, an approach that aligns deeply with Fox’s own commitment to service.

He credits his undergraduate years at Boston College with grounding him in the principles of service and whole-person care, values he continues to strengthen during his training in Buffalo.

For Fox, this holistic perspective also shapes how he views new technology. He believes this is where AI belongs in medicine: handling the administrative, search-heavy work so clinicians can focus on what matters most, empathy, counseling, shared decision-making and delivering truly holistic care.

His mentor, Kenneth V. Snyder, MD, PhD, associate professor of neurosurgery, radiology and neurology at the Jacobs School, praised Fox’s approach, noting that “Brendan built a scalable tool that brings resources directly to patients and allows feedback on the quality of services.”

He added that “This work is an exceptional contribution to helping us all improve population health outcomes.”

The referral tool is one component of what Fox calls a broader “HEAL AI” initiative — Health, Equity and Accessibility through Language and AI. With support from a UB Community Engagement Fellowship, he and fellow medical student Camryn Imperiale are expanding the work into health literacy by translating dense preoperative notes and consent forms into plain, sixth–grade–level language across multiple languages, including English, Spanish, French Bengali, Arabic, Mandarin and Hindi. The intention is to make information easier to grasp, reduce confusion and deepen the trust patients place in their care.

The fellowship supports projects that confront social, educational or health care inequities and helps trainees grow as leaders in community‑engaged, equity‑focused research. Fellows work with a faculty mentor for 12 months to design, implement or analyze a project and present their findings at the annual Fellowship Symposium in April.