A Los Angeles-based health care technology company is introducing artificial intelligence (AI) into street medicine programs in the Bay Area.
Akido Labs has started working with street medicine teams in San Mateo County to help diagnose and treat people experiencing homelessness. The company created ScopeAI, an AI tool that increases access to care for unhoused patients.
ScopeAI allows outreach workers to visit patients, even if they lack formal medical training. The system guides them through structured questions, records the conversation, and organizes the information for a doctor to review. It then suggests possible diagnoses, tests, and medications.
“ScopeAI is the first AI technology capable of running a comprehensive patient visit,” said Jessica Greene, an Akido spokesperson. “It acts as an intelligent teleprompter during a patient visit.”
Greene said that the tool gathers symptoms, medical history, and other details in real time, but the final medical decisions on diagnosis and treatment plans are always made by a licensed provider.
In street medicine settings like encampments and other nontraditional locations, Greene said the structure has been valuable.
“Many unhoused patients have multiple chronic conditions, behavioral health needs, and fragmented medical histories,” Greene said. “Having a system that consistently gathers and organizes information helps reduce the chance that something important is overlooked.”
Homeless people often have limited access to regular medical care and depend heavily on emergency services.
According to The Guardian, the Bay Area’s unhoused population has grown by 6%, with nearly two-thirds living on the streets. A 2023 San Francisco Fire Department report also found that 25% of ambulance trips involve people experiencing homelessness.
Health outcomes for unhoused Californians are far worse than those of the general population, and many die at younger ages, according to CalMatters. Many face complex medical and mental health conditions, including diabetes, substance use disorders, and serious mental illness, yet there are often not enough providers to give them consistent care and support.
Justin Szlaza, the director of homeless initiatives at the Future Communities Institute (FCI), a nonprofit partnering with Akido Labs, said this AI device will be highly effective.
“People living on the street don’t have access for all kinds of reasons, because they can’t get into the doctor’s office, because they don’t have the ability to pay, because there aren’t enough physicians that are able to treat them. So I think it can really benefit some people who don’t normally have access to health care in a profound way,” Szlaza said.
According to Greene, ScopeAI has enabled providers across its broader medical network to see more patients by up to four times and achieve a patient Net Promoter Score (NPS) greater than 90. The tool has also surfaced the correct diagnosis in its top three suggestions 92% of the time.
The technology is currently being used across six specialties, but the team is expanding it across Akido’s medical network of 26 specialties. The tool has helped shorten treatment timelines, especially for unhoused Medicaid patients.
“Many of our patients who are experiencing homelessness can now receive a comprehensive visit and treatment plan within 24 hours of initial engagement, rather than weeks later,” Greene said.
Because trust is important in street medicine, patients are clearly informed, both verbally and in writing, that AI is part of the visit and that their provider remains in charge. Providers can see how the system arrived at its suggestions through a comprehensive justification log, which helps them review the AI’s reasoning and full visit transcripts.
This was an intentional design choice because most large language models are known as “black boxes,” meaning even developers cannot always explain how they reach decisions. The system is also continuously improved using real-world feedback.
“ScopeAI is designed around transparency, which is key to building trust with providers, who in turn can spend more time focused on patients. Human oversight is built into the system by design,” Greene said.
However, there are also concerns about AI bias.
“Bias in health care is a long-standing issue that predates AI. Our goal is to reduce variability and blind spots rather than reinforce them,” Greene said.
Clinicians are trained not to depend only on AI. ScopeAI organizes patient information, including medical history and social factors, so providers can look at the whole picture. This helps make care more consistent for vulnerable patients.
“We see AI’s role in safety net care as two-fold: massively expanding the capacity of every provider and addressing systemic barriers that vulnerable populations face,” Greene said. “By allowing providers to see more patients while increasing meaningful face-to-face time, we can extend high-quality care to more people.”
While ScopeAI is still expanding, Akido says early results show improved efficiency and high patient satisfaction.
“ScopeAI is still in the early stages of scaled deployment, and we are building longitudinal data to evaluate long-term outcomes,” Greene said.
A nonprofit’s take on technology
Experts outside of Akido see AI as part of a larger trend in healthcare technology. Szlaza compared AI in healthcare to the data tools already used by major medical systems.
“I’m a member of Kaiser, and Kaiser uses a lot of information gathering tools when I schedule an appointment, or when I schedule an appointment for my kids, or they’re constantly analyzing the information, the data they have with us,” Szlaza said. “So I think that doing a better job with managing that information, by applying AI-type tools to try to spot trends or pick things up, is probably going to help deliver better outcomes for patients and make people’s lives happier and more comfortable.”
Recently, hundreds of millions of people have turned to chatbots to better understand their symptoms, according to Sermo, an online community for physicians. Some physicians say that trend shows a larger issue in the U.S. healthcare system: the shortage of physicians.
“Is it going to be perfect? No. But the main problem with our healthcare system is cost, and that’s related to its accessibility to physicians,” Szlaza said.
According to a study published in the World Journal of Advanced Research and Reviews, staffing in specialized fields, especially geriatric care, is inadequate as the population ages. Additionally, bedside nursing roles experience high turnover and burnout, resulting in ongoing shortages of registered nurses (RNs) and licensed practical nurses (LPNs).
However, AI can help address these growing medical needs.
“AI can respond to changes in demographics and the need for different types of specialties,” Szlaza said.
Szlaza added that as more people track their own health data, they will want tools to help them interpret it.
“It’s a way to get around that supply and demand problem where we just don’t have enough physicians to properly care for, at a very high level, all the people who require care. So we have to find some way to increase the supply of care,” Szlaza said.
Caring beyond the clinic: the human side of street medicine
Kevin Lagor is a certified family nurse practitioner who works in street medicine and overdose response in San Francisco. He recently served as Associate Medical Director at Philadelphia FIGHT’s Broad Street Love clinic, where he led and expanded programs for people experiencing homelessness and housing insecurity.
Lagor said the traditional health care system often does not work for people living on the streets.
“Much of the traditional medical system is rigid and scheduled. While this may work for the population at large, people experiencing homelessness deal with barriers that make successfully making a scheduled appointment near impossible,” Lagor said.
Not having money for transportation, the risk of belongings being stolen, and the need to choose between food and a doctor’s visit all contribute to missed appointments.
Street medicine, Lagor said, is built differently.
“Street medicine brings the clinic to the person and does it with kindness and dignity,” Lagor said. “Sometimes people are too sick to even make it to the clinic, so having the ability to provide care in the community ensures they get the help they need, where and when they need it.”
While some health systems are now turning to AI, Lagor believes there are limits to what technology can do in this setting.
“AI is capable of taking in data, such as symptoms or needs, and generating an answer based on a slew of research, news articles, and personal blogs from people who really don’t know what they’re talking about,” Lagor said. “So first, I would question the quality of data that goes into the answer.”
Treating people living outside goes beyond entering information into a system or following a checklist.
“People experiencing unsheltered homelessness need an experienced, caring individual to ask the right questions and observe things that might not be entered into a query but are the most important factors. This takes a conversation and insight, which is impossible with artificial intelligence,” Lagor said.
Through his experience, Lagor has seen repeated breakdowns in the health care system for people living on the streets.
“You see the need for compassionate care in the community when you see the system fail someone multiple times,” Lagor said. “We see incredibly sick people who need help, and the traditional expectation is that this person who is suffering does the work to be seen. When you think about it, how could this possibly work? It doesn’t, which is why street medicine exists.”
Many people need support just to reach a point where they can work through the traditional system — and even then, some still struggle.
A county perspective
Since 2016, Christopher King has spent nearly every Thursday providing care to people experiencing homelessness through the San Mateo County Mobile Clinic. Over the years, he has watched the program grow from one team covering the entire county to five teams divided by geography.
Each team is usually two to three people and travels in a fully stocked van with medications, diagnostic tools, and wound care supplies. They provide comprehensive care on the street, including physical exams, STI testing, vaccines, and medication dispensing.
Although working outside a traditional clinic may seem limiting, King says medical care itself is not the hardest part.
“Surprisingly, diagnosis and treatment come fairly naturally in our setting,” King said.
Instead, one of the biggest challenges is instability. Encampment sweeps can displace patients, making it difficult to find them and continue their treatment plans.
For King, the foundation of street medicine is consistency and respect.
“Frequent engagement is critical. Showing up when we say we’re going to show up is everything. It demonstrates that you care, that you keep your word, that you can be counted on,” King said. “We respect the wishes of the people we meet and follow a creed of non-judgment and acceptance.”
King recalls one patient living with severe mental illness and alcohol use disorder who first allowed the team to treat a lice infestation. Later, when the patient reported urinating blood, he trusted the team enough to be examined and undergo testing. He was diagnosed with advanced bladder cancer. Because of that established trust, the team helped guide him through treatment, address his addiction, and connect him to hospice care, where he spent his final months in stability and peace.
As AI tools begin entering health care, King is cautious about how they might fit into street medicine.
“Because relationship-building is the foundation of street medicine, I’m not sure AI would be particularly helpful at this juncture — though I suppose it would depend on how the technology is deployed,” King said. “Using AI to help build a differential diagnosis or develop a thorough workup could certainly be useful. But replacing a provider with AI in the field? I have a hard time wrapping my head around that.”
Years of working on the streets have shaped King’s view of the public’s perception of homelessness.
“I think people misunderstand that no one came out of the womb wanting to be homeless,” King said. “They want health care and deserve health care.”
His team has treated HIV, cured hepatitis C, and helped patients manage serious mental illness and substance use disorders. But he says the everyday care matters just as much — vaccines, STI screenings, wound treatment, chronic pain, and seasonal allergies.
“Homelessness does not suspend a person’s basic human needs,” King said. “People experiencing homelessness still want, need, and deserve health care — all of it, just like anyone else.”
From the front lines: a student volunteer’s view
Dheekshitha Sathyanarayanan, a junior at Carlmont High School, volunteers at Mills-Peninsula Sutter Health in Burlingame. Through her work, she has seen how important face-to-face interaction is in patient care — something AI cannot provide.
“For example, I met a woman who enjoyed crocheting as I did. That conversation led to her talking about things she liked and her granddaughter, and it was obvious to see that my company made her visibly happier,” Sathyanarayanan said. “So it is definitely important to have face-to-face interaction in patient care.”
Sathyanarayanan believes what makes patients feel safe, especially when they are vulnerable, is compassion and reassurance.
“So far, what I’ve realized is the importance of conveying that you are there for someone and that you’re willing to help,” Sathyanarayanan said. “There are many older patients there who don’t even know why they are there, so during times like that, it is important to stay calm and convey that they will be okay.”
Sathyanarayanan said her hospital is preparing to introduce a new surgical machine, though it has not yet been used. With many patients and fewer doctors and nurses, she believes technology could help ease the workload. Still, she understands why some patients may feel unsure about relying too much on machines, especially in serious situations.
“It’s a life-and-death situation, and I can see why people would be skeptical about having a robot operate on them rather than a human,” Sathyanarayanan said.
She also worries about accuracy and trust when it comes to AI tools.
“From personal experience, I’ve seen AI like ChatGPT give out wrong information. Getting a diagnosis from AI does seem skeptical in that sense and possibly even dangerous,” Sathyanarayanan said.
As ScopeAI reaches more street medicine programs, providers, and nonprofit leaders hope it can help fill some of the gaps in a health care system that’s already struggling to keep up. But some believe that compassion, not code, will continue to define what real care looks like.
“Most importantly is that humans need humans. We all need to feel cared for or seen. Support and understanding from a human cannot be replaced, even by the most advanced computer programs,” Lagor said.
