“Our focus is really on what we can do to remove the doctor from the visit,” says the CTO of medical startup Akido.
See also: ShadowLeak: AI-powered data theft 'undetectable'
Picture this: You’re feeling unwell, so you call your doctor’s office to make an appointment. To your surprise, they schedule you for the next day. At the appointment, you’re not rushed into describing your health concerns. Instead, you have half an hour to share your symptoms and concerns, as well as the exhaustive details of your medical history, with someone who listens carefully and asks thoughtful questions. You leave with a diagnosis, a treatment plan, and the feeling that, for once, you were able to discuss your health with the attention it deserves.
That’s the new reality for patients at a small number of clinics in Southern California run by medical startup Akido Labs. These patients—some of whom are on Medicaid—can access appointments with specialists on short notice, a privilege usually enjoyed only by the wealthy who visit concierge clinics.
The key difference is that Akido patients spend relatively little, if any, time with their doctors. Instead, they see a medical assistant, who can provide a listening ear but has limited clinical training. The work of formulating diagnoses and creating a treatment plan is done by a proprietary LLM-based system called ScopeAI, which transcribes and analyzes the dialogue between the patient and assistant. A doctor then approves or corrects the AI system.
See also: RevengeHotels leverages AI to distribute VenomRAT
According to Prashant Samant, CEO of Akido, this approach allows doctors to see four to five times more patients than before. There’s good reason to want doctors to be much more productive. Americans are getting older and sicker, and many are struggling to access adequate medical care. The upcoming 15% in federal funding for Medicaid will only make things worse.
But experts aren’t convinced that shifting so much of the cognitive work of medicine to AI is the right way to address the doctor shortage. There’s a big gap in expertise between doctors and AI-enhanced physician assistants, says Emma Pierson, a computer scientist at UC Berkeley. Jumping into such a gap can introduce risks.
AI is already everywhere in medicine. Computer vision tools detect cancers during screening scans, automated research systems allow doctors to quickly sort through medical literature, and medical secretaries with LLMs can take appointment notes on behalf of a clinician. But these systems are designed to support doctors as they go about their standard medical routines.
What sets ScopeAI apart, Goodner says, is its ability to independently complete the cognitive tasks that make up a medical visit, from recalling a patient's medical history to generating a list of possible diagnoses and identifying the most likely diagnosis and suggesting appropriate next steps.
See also: LangChainGo: Critical vulnerability allows access to sensitive files
Behind the scenes, ScopeAI is a set of large language models, each of which can perform a specific step in the visit—from generating appropriate follow-up questions based on what a patient has said to populating a list of possible situations. In most cases, these LLMs are adapted versions of Meta’s open-source Llama , though Goodner says the system also uses Anthropic’s Claude models.
