Visiting your healthcare provider, even in a difficult time, should absolutely not be a frustrating experience. Of all the spaces we visit and inhabit, the doctor’s office is one of the most important to be comforting, where you can feel warmth and care.
Too often, this doesn’t turn out to be the case. You have probably had the experience of waiting hours in a doctor’s office for a fifteen minute appointment. After that wait you may have felt disappointed and muted, speaking briefly with a physician who spends more time looking at a computer screen than at you, typing as you speak. This is not a conversation. It’s not personal. It might make you feel more like a faceless chart than a patient.
I don’t blame doctors or their staff for this dynamic. Even as each person is a patient with thoughts and feelings who wants to feel seen and considered, they also present a huge depth of information. Doctors see many patients every day. Helping them all requires efficiency and copious note taking for each chart, and it takes time outside of working hours – some call it “pajama time” – to digest those notes. Extended time doing documentation in the electronic health record (EHR) causes burnout for physicians.
“For every hour any of us as a provider, a physician, spends with the patient, we spend two hours at the keyboard trying to capture all of it.”
- Dr. Nishit Patel, the vice president and chief medical informatics officer at Tampa General Hospital, to CNBC.
Our clinical system would feel more compassionate if we could format our appointments as direct, personal conversations and leave note taking aside. It would be more efficient if doctors could spend more time talking to patients and less time scrambling to get everything written down.
There are potential solutions that are already making a difference. Some clinics are integrating Ambient AI into their workflows. This tech can record transcripts of appointments and synthesize them in real time, creating structured notes that physicians can later check and edit. This replaces the manual note taking that keeps doctors watching screens while speaking to patients. It allows appointments to be more like real conversations, structured in a more free form, letting doctors be more present.
The Permanente Medical Group (TPMG) has given this technology a try. In October 2023, they enabled an ambient AI system for 10,000 physicians and staff, using it to assist in hundreds of thousands of patient interactions, and studied the results.
Both physician and patient feedback showed promise in the technology. In preliminary data, 81% of patients reported that their physician spent less time looking at the computer screen than in their previous visits, while all patients reported feeling “neutral” or “very comfortable” about the AI tool being used.
Physicians raved. Said one user: “I use it for every visit I can and it is making my notes more concise and my visits better. I know I’m gushing, but this has been the biggest game changer for me[…] I even had a patient praise the fact that I could listen instead of type during the visit.” AI scribes are still evolving in their accuracy and quality, but the potential benefits to our clinical system are obvious. When AI can handle the tedious parts of clinical work that distract doctors from the patients in front of them, the resulting conversations are more authentic and compassionate even in long appointments.
Stanford Medicine has been working on this too. They’ve developed ChatEHR, an AI-powered software to streamline clinical workflows and enhance patient care. This makes notes and transcripts even more accessible by enabling healthcare providers to ask questions directly of patient records in natural language, the way you might ask a question to an LLM. This reduces the amount of manual searching that is usually required when navigating a patient’s health records.
Stanford Medicine outlines many potential use cases here, especially as the technology continues to improve. In the future, programs like ChatEHR could predict and highlight patient risks based on data in their medical records. AI is already showing the capability to find insights based on images like MRIs and mastectomies, sometimes even catching risks before human analysts do. It could even recommend treatments based on patient histories and medical literature, taking advantage of AI’s capability to synthesize massive amounts of data. There’s still more it can do in the clinic, too. Programs like ChatEHR can help optimize resource allocation and pre-visit information collection, letting clinics and hospitals run more smoothly and reducing wait times for patients.
There are compelling concerns about the use of AI in the medical field. One is the presence of “hallucinations,” when AI models make up information. Sometimes it makes up references or studies, or generates inaccuracies. Ambient AI scribes have made key errors in their appointment summaries such as recalling exams being performed that were simply spoken about, or diagnoses of diseases that did not happen. It’s critical that physicians rigorously check its work. We can use generative AI to help us with tedious tasks, but we cannot use it to replace our knowledge and experience.
And in order to integrate AI successfully into healthcare, we need to know that patients still come first. In all the trials mentioned here, patients were given the option to decline the use of ambient AI in their appointments. Surveys show patients’ fear of AI taking the physician out of the loop, of an algorithm making decisions around their health.
Transparency is key here. AI must not be a mysterious entity to users and patients. Clinics and hospitals must make sure everyone knows exactly where in their processes AI is being used, and what its limitations are. In medical settings where people share sensitive information out of trust in their doctors, doctors must respect that trust by securing consent before using any type of AI service.
Still, this technology is exciting because it’s improving so quickly. We can likely expect to see iterations of ChatEHR and tools like it with fewer hallucinations, that save more time for doctors and make more time for patients. In clinics that adopt ambient AI as it develops, the frustrating stereotypical experience of impersonalized healthcare can be done away with.
