Robin Healthcare’s co-founder, Noah Auerhahn, says the company is “problem obsessed.” This is what makes it different from other young healthcare companies that all too often start with a focus on technology first (e.g., “let’s do something cool with Alexa”). The problem the Robin team wants to solve is big and it’s important: the documentation burden that is contributing to inefficient and costly care as well as to the now pervasive issue of physician burnout.
The documentation problem
According to an article published on the website Physicians Practice, there is not much formal research linking EHR use and burnout, however it quotes the deputy chief health officer at IBM Watson Health, Paul DeChant MD, as stating that.
“…there is data that shows for every hour physicians are directly with a patient, they are spending two hours doing administrative work. Many physicians are spending one to two hours at home working in the EHR — known as pajama time.”
He went on to say that the issue of burnout is
“eating away at the heart of [medical] practice.”
This is due, in part to the clunky nature of most EHR systems that require physicians to use a keyboard to enter data forcing them to turn their backs on patients while they try to capture all of the information the EHR requires. These systems were simply not designed to fit into a doctor’s practice, rather they were designed to meet the business needs of the healthcare organization (e.g., billing) no matter how disruptive they are to the physician-patient relationship.
How Robin is different
Instead of making physicians alter their workflow to accommodate the requirements of the Electronic Health Record (EHR), Robin fits seamlessly into the way doctors have been taking care of patients for ages. The device sits in the exam room unobtrusively recording audio and video of the patient encounter. Using artificial intelligence, the device can translate the doctor-patient conversation into a physician note. The note is reviewed by a medical scribe and then is ported into the doctors EHR where it becomes a part of the patient’s medical record. Unlike EHRs, it doesn’t require that the doctor spend most of the visit looking at a computer screen and unlike medical scribes, it doesn’t require the presence of someone else in the room.
Now, I have made it seem simple, but, of course, it is not. Conversations in medical exam rooms are often unstructured and can wander all over the place although most visits will consist of taking a patient history, doing a physical exam, and then formulating a treatment plan. Robin and its scribes must turn the recording into a highly structured note, using “doctor talk,” that can easily be read and understood by the examining doctor as well as other health professionals that have a medical need to access the record.
The Robin AI is not meant as a replacement for scribes any more than autopilot is a replacement for pilots. Rather, it is able to capture the entire conversation and create the first draft for a medical scribe to edit and a physician to review and sign before it becomes a part of the official record. Noah tells me that sometimes 80% of the note is generated by the machine with a 20% contribution by the scribe and other times only 20% is created by the machine with the scribes listening to the recording and doing most of the note creation. As with all systems that use AI, it is anticipated that it will get better and better as it learns from its experience (i.e., machine learning).
A conversation with Emilio
I was so intrigued by the promise of Robin that I invited Robin’s other co-founder, Emilio Galan to join me in a video interview. You can see the entire video, titled “Robin is Helping Doctors and Patients Eye to Eye” by clicking here. What follows is a condensed version edited for readability.
Pat: Emilio Galan is a serial entrepreneur. His latest venture that we’re going to talk about today is Robin Healthcare. Emilio, I’ll just let you describe what it is, it’s really very cool.
Emilio: Robin is a device placed in the room with the physician and the patient. Instead of the physician typing and, every once in a while looking over his shoulder at the patient, he can maintain eye contact and actually just have a conversation.
Pat: That’s the way we used to do it.
Emilio: Exactly and that’s what we are trying to create again – just place this device in the room and you get to forget about the documentation part of the encounter and focus on patient care.
Pat: You use artificial intelligence to capture the conversation but how does that turn into a note? What are the steps involved in going from me examining the patient to ending up with a note that gets ported into my patient’s electronic health record?
Emilio: For the physician, it’s kind of like this magical experience. You just get to focus on the patient and the note shows up in your EMR. On our end, you can imagine there’s a ton of work that goes into it. We are capturing audio and there is optional video so when a patient says “it hurts here” or “it hurts there” or if he comes in with a wheelchair, all that is captured.
The first thing the machine does is try to extract all of the clinical information from the conversation so if you’re talking about Game of Thrones or the weather or a recent trip to Hawaii, it is filtering all that out and it’s looking for the knee pain, the severity, tingling, medications. It identifies what is medical and extracts it out of the out of the conversation to place within the note.
One of the hard parts of the process is determining what is relevant and what is not. We also need to sort out confusing parts of the conversation (for example, previous aspirin use is different than current aspirin use which is different than I recommend aspirin use -Knowing exactly where to place that mention of aspirin is really important.
Pat: A lot of people are do think about AI as magic or probably the word I hear the most is it’s a black box. In fact, if you go into that black box it’s all about the people – people develop the algorithms. Who developed your algorithms? How do you tweak them? How do you keep them current? How do you do this thing they call machine learning which is, I think, really people teaching the machine.
Emilio: That’s an important point. Machines learn best from repetition, things that a human can do again and again and again – it picks up those patterns. We have a fantastic technical team that includes people from UC Berkeley and some former folks from Google who are helping build out those algorithms. The real key is having the data and breaking down clinical visits into something that’s repeatable, little chunks that we’ve seen again and again that the machine can learn from. For example, the first time I say “it’s your left knee that hurts, right?” the machine might say is it left or right? Or I’m depressed because my mother recently was diagnosed with cancer, who has the depression, who has cancer? It is difficult to tease these things out but when you see it again and again, the machine can learn that right is used differently here, they’re actually talking about the left knee and it can capture that and document it.
Pat: And how long has it taken you to build this out?
Emilio: We’ve been working on Robin for a year and a half.
Pat: Which is which is really a very short period of time.
Emilio: There’s been a lot of development in Natural Language Processing and Machine Learning. But the challenge is that unlike Alexa or Google home that may play the wrong song, documentation of a medication or a diagnosis needs to be correct. There is a very low tolerance for error in medicine, so an extra piece that we have to do with Robin is to ensure the accuracy of everything.
Pat: I understand that you use some human beings for that.
Emilio: Absolutely. There are humans in the loop as the machine learns more. Also, every note we create gets QA’d by a human. And then, the machine learns from any corrections the humans made on the algorithm.
Pat: Who are these humans and how does this relate to what a lot of doctors are doing right now by having an actual human, called a scribe, in the room. This person is capturing the information and creating the note.
Emilio: About 5% of the market now hires a human to follow them around and document as they talk with patients. There are challenges: (1) you have another human in this small room kind of invading that space and (2) it’s expensive to have a human follow you around all day to type (3) the human gets sick and even more so they’re doing the job to go off to medical school.
Pat: So you train one and then hopefully he gets into medical school and then you have to train another one.
Emilio: Yes, they typically leave after nine months – you train them which takes a few months and then they’re with you for another six months. The clinician’s experience with Robin is very similar except Robin never gets sick, never goes off to medical school and it’s much cheaper than having a human.
Pat: And, there isn’t another person in the room with you.
Emilio: Exactly, it’s like a sanctuary of patient and doctor. On our end, the humans we use to review recordings have very similar demographic to in-office medical scribes. The difference here is that our scribes are super-powered by the AI algorithms – they are reviewing and guiding the algorithms.
Pat: So if I got it right, what you’re saying is that Robin does the first draft of the note, the scribe reviews it and does the next draft and then, I’m assuming, the physician does the final draft since he or she has the ultimate responsibility for the note.
Emilio: Exactly so no matter what solution physicians use now whether it be dictation with Dragon, an overseas transcription service or an in-person scribe, the physician always does the final review.
Pat: How much of an individual note does Robin generate versus how much is actually created by the scribe who listens to the recording?
Emilio: It depends on the complexity of the case, whether it’s a new patient or a follow-up. It also depends how many notes we’ve done for that physician and how much of their behaviors have been learned. The simple cases get automated much faster, the complex cases may always need some level of human review.
Pat: How much would you say it cost to produce a simple note?
Emilio: If you’re coming back after an operative visit, we most likely know you’re going to say. It’s very cheap to do those notes, it might just take 30 seconds for that note to be QA’d and sent to the EHR. A complex patient is another story. We’ve had physicians say Robin saved them 15 hours in a single week.
Pat: And, 15 hours of an orthopedist time is a lot of money!
Emilio: It is a lot of money. There are so many winners here: physicians are saving hours of time with Robin. They can see more patients and make more money if they want but more than that they get to go home on time see their family. One physician told me that he finally got to put his new son to bed for the first time after Robin. That’s the kind of story I love to hear. Patients, on the other hand, are saying, I love that the physician gets to look at me and maintain eye contact. This is instead of dictating in front of you or going in and out of the room to chart.
Pat: That’s fantastic because both doctors and patients say that a part of the joy of medicine is the physician-patient interaction.
Emilio: Every physician should have something like this, right? Whether it be Robin or any other service, physicians deserve to be able to just focus on medicine. Patients deserve physicians that are completely focused on them and the system at large deserves to have really great documentation. The other piece to Robin is having these great notes that are correct, that you can refer to and rely on – that’s what we’re working towards…I want to see this in every Doc’s office.
Pat: Well I want to thank you very much, not only for joining us and telling Robin’s story but for inventing this amazing device that could really be a game-changer – actually, it will be like going back to the kind of game that I experienced when I was first in training before we had electronic health records.
Emilio: We are trying to get back to the heart of medicine which is to focus on the patient. We are not just trying to throw some new technology at physicians, that’s not the point. The point is technology should get out of the way of practice and let Doc’s do medicine and let technology happen in the background.
Pat: We’ll let that be the last word. Thank you very much, Emilio.
The transcribed interview has been significantly condensed as well as edited for readability. We encourage you to learn more about Robin by viewing the entire interview here.