Revolutionizing Healthcare Access with AI-Driven Medical Assistance
Picture this scenario: You’re feeling under the weather and call your healthcare provider to schedule a visit. To your surprise, you’re offered an appointment the very next day. During the consultation, you’re given ample time-about 30 minutes-to thoroughly explain your symptoms, medical history, and concerns to a compassionate listener who asks insightful follow-up questions. You leave with a clear diagnosis, a tailored treatment plan, and a genuine feeling that your health has been prioritized.
Here’s the twist: the person you spoke with might not have been a licensed physician.
Introducing AI-Augmented Clinics in Southern California
This innovative model is currently being piloted by Akido Labs, a healthcare startup operating several clinics in Southern California. Their patients, including many covered by Medicaid, gain rapid access to specialist consultations-an advantage traditionally reserved for affluent clients of concierge medical services.
Instead of spending most of the appointment time with a doctor, patients interact primarily with a medical assistant who, while empathetic, has limited clinical training. The complex tasks of diagnosing and crafting treatment strategies are handled by ScopeAI, a proprietary system powered by large language models (LLMs). This AI tool transcribes conversations between patient and assistant, analyzes the data, and generates diagnostic suggestions and care plans, which are then reviewed and approved by a licensed physician.
“Our goal is to minimize the doctor’s direct involvement during the visit,” explains Jared Goodner, Akido’s Chief Technology Officer.
Boosting Physician Efficiency Amid Growing Healthcare Demands
According to Akido’s CEO, Prashant Samant, this approach enables doctors to manage four to five times more patients than traditional methods allow. This efficiency is critical as the U.S. faces a growing shortage of healthcare providers and increasing rates of chronic illnesses such as diabetes and heart disease. Moreover, the anticipated 15% cut in federal Medicaid funding threatens to exacerbate access issues for vulnerable populations.
Balancing Innovation with Caution: Expert Perspectives
Despite the promise, some experts urge caution. Emma Pierson, a computer scientist at UC Berkeley, highlights the significant expertise gap between physicians and AI-supported medical assistants. “While AI has tremendous potential to broaden access to medical expertise, this particular model raises concerns about patient safety and care quality,” she notes.
AI technologies are already embedded in healthcare workflows-ranging from cancer detection in imaging to literature review tools and automated clinical note-taking. However, these systems typically assist doctors rather than replace their cognitive functions.
ScopeAI: An Autonomous AI System for Medical Visits
ScopeAI distinguishes itself by autonomously performing key cognitive tasks during patient visits. It gathers detailed medical histories, generates relevant follow-up questions, compiles differential diagnoses, identifies the most probable condition, and recommends next steps such as referrals or prescriptions.
Technically, ScopeAI comprises multiple fine-tuned LLMs-primarily based on Meta’s open-source Llama models and Anthropic’s Claude models-each specialized for different stages of the clinical encounter. Medical assistants use an interface that prompts questions generated by ScopeAI, which adapts dynamically to patient responses.
After the appointment, physicians receive a concise summary including the patient’s history, the leading diagnosis, alternative possibilities, and suggested treatments, along with justifications for each recommendation.
Expanding Care to Underserved Communities
ScopeAI is currently deployed in cardiology, endocrinology, primary care, and Akido’s street medicine program serving Los Angeles’ homeless population. Led by addiction medicine specialist Dr. Steven Hochman, this outreach team leverages ScopeAI to facilitate timely treatment for substance use disorders.
Previously, Dr. Hochman had to meet patients in person to prescribe medications for opioid addiction. Now, caseworkers equipped with ScopeAI conduct interviews independently, with Dr. Hochman remotely reviewing and authorizing treatment plans. “This technology lets me extend my reach to multiple patients simultaneously,” he says.
Since adopting ScopeAI, the team has reduced the time to initiate medication-assisted treatment to under 24 hours-a remarkable improvement in care delivery.
Insurance and Regulatory Challenges
This model’s success hinges partly on Medicaid’s flexibility, which permits asynchronous physician approval of AI-generated treatment plans. However, many private insurers mandate direct doctor-patient interactions before authorizing care, raising concerns about potential disparities in access.
Samant acknowledges these concerns but emphasizes that rapid AI-assisted visits offer a valuable alternative to long wait times common in Medicaid populations. Patients can still opt for traditional doctor appointments if they prefer.
Regulatory frameworks present additional hurdles. Harvard Law professor Glenn Cohen points out that AI systems functioning as “virtual doctors” may require FDA approval and could conflict with medical licensing laws that restrict practice to credentialed professionals.
California’s Medical Practice Act prohibits AI from replacing a doctor’s diagnostic and treatment responsibilities but allows physicians to use AI tools and to diagnose patients remotely. While regulatory bodies have not definitively ruled on ScopeAI’s compliance, Akido maintains that its system operates within legal boundaries by ensuring all AI recommendations receive physician oversight.
Patient Experience and Ethical Considerations
Patients interact solely with medical assistants, unaware of the AI’s role behind the scenes. This design aims to create a familiar and comfortable environment. However, bioethicists like Zeke Emanuel from the University of Pennsylvania caution that this setup may obscure the extent of algorithmic influence on care decisions.
Pierson concurs, stating, “This model challenges traditional notions of the human touch in medicine.”
Medical assistant DeAndre Siringoringo explains that while patients are informed that AI is involved in gathering information, they are not briefed on how ScopeAI formulates diagnostic suggestions.
Automation Bias and Quality Assurance
Although physicians make the final decisions, studies have shown that doctors often over-rely on AI recommendations-a phenomenon known as automation bias. This risk may be heightened when doctors are not physically present during patient interactions.
Akido acknowledges this concern and has implemented measures to mitigate bias. “ScopeAI is designed to counteract common blind spots in clinical decision-making by supplementing physician intuition with data-driven insights,” an Akido spokesperson states. Physicians receive training to use the system judiciously, maintaining accountability and avoiding overdependence.
Performance monitoring includes retrospective testing on historical patient data and tracking how frequently doctors amend AI suggestions. Before launching ScopeAI in any specialty, Akido ensures the system’s top three diagnostic recommendations include the correct diagnosis at least 92% of the time.
However, Akido has yet to conduct comprehensive clinical trials comparing patient outcomes between ScopeAI-assisted visits and conventional care models. Such studies would be crucial to validate the system’s safety and effectiveness and to assess the real-world impact of automation bias.
Looking Ahead: The Future of AI in Healthcare
Expanding affordable and timely medical care through AI is an admirable objective. Yet, as Pierson emphasizes, “Robust, evidence-based evaluations are essential to ensure that these innovations truly benefit patients without compromising quality.”
As AI continues to reshape healthcare delivery, balancing technological advancement with ethical responsibility and regulatory compliance will be paramount to building trust and improving outcomes for all patients.
