By Peter Yellowlees, MD
As physicians, we all remember certain patients or clinical events with great clarity as they have a special impact on us for many reasons. One of these happened to me in 2009 when I was working in the UC Davis outpatient psychiatric clinic.
For several years, I had been treating a pleasant elderly lady with quite severe longstanding bipolar disorder. As a retired couple, she and her husband spent California winters in a warmer state and midway through one of these trips she became depressed. Rather than seek out a local physician whom they did not know, her husband, who was technologically quite savvy, took a video of her with his phone, which he texted to me. I can still recall her describing to me how she felt and what were her symptoms, and how her husband finished the video by showing me the view from the balcony they were sitting on.
He asked me to review the video and respond to them via the Epic MyChart. I could see that she was depressed, which she confirmed with her description of her symptoms, so I went online making some changes to her medication and sending the script to their local pharmacy in the other state. Because the text communication was not HIPAA compliant I deleted the video but described her in my notes, although I rather wish now that I had not deleted the video as this turned out to be, I believe, likely the first documented asynchronous video psychiatric consultation ever recorded. In the coming weeks I followed up with my patient via MyChart and was pleased to see that she recovered well from her depressive episode, and that both she and her husband were delighted to have been able to be treated by their own doctor in this manner.
By 2009 I had already been practicing telepsychiatry for many years, seeing my first patient on an Aboriginal reservation in Australia in 1991, and had thought about video-recording patients many times. It has always seemed ridiculous to me that we see patients on video and then write out lengthy details of their mental state when we could have easily inserted a video clip to more accurately show most of the mental state components. Unfortunately, every lawyer that I spoke to up until then told me not to make such recordings, although their advice has since changed with the widespread advent of smart phones.
So, I had an “Aha!” moment with this patient. Two ideas were clarified at once. Firstly, this was proof that taking videos of patients could be very helpful clinically, with lots of other potential use cases, and likely with high satisfaction from patients. Secondly, taking video is really just collecting data, as we do with questionnaires, but with my decades-old interest in artificial intelligence from research in the 1990’s I knew that these recordings are a rich source of video and audio file data that could be enhanced by AI and other technologies to improve our clinical care. But what to do about this?
Now began a decade of research at UC Davis into asynchronous telepsychiatry. With a strong team of clinicians and researchers, including about 10 faculty psychiatrists at UCD, we demonstrated the feasibility, good clinical outcomes, high patient and provider satisfaction, and cost-effectiveness of the asynchronous telepsychiatry approach in both primary care environments, and with my colleague, Glen Xiong MD, in nursing homes. During this time, we received about $7 million in funding from NIH, conducted three clinical trials and treated over 500 patients.
Then came the pandemic. The recent surge in the popularity of telepsychiatry, when at times 50-60% of all psychiatric outpatient visits were being conducted on video or phone, has enabled patients to receive care in their homes and clinicians to have much more flexible schedules, working when and where they wish. It is hardly surprising that in the two years after the pandemic mental health has emerged as the clinical discipline that has continued to have the highest levels of real-time consultations by video, at over a third of consultations nationally, while other clinical disciplines such as primary care, pediatrics and internal medicine have settled at about 10% overall. These numbers do not include asynchronous (or “store and forward”) telemedicine, where the disciplines of radiology, pathology, dermatology and ophthalmology, in particular, feature many specialists working 100% asynchronously, often across state and national boundaries.
Given the enormous changes driven by the pandemic, and more recently by the clinical emergence of artificial intelligence, what is the next digital step that will impact my own discipline of psychiatry? I think there is little doubt that it will be the development of asynchronous mental health consultations, likely eventually making up about 20% of all mental health visits, where AI tools will be added to the recorded videos of patients to make increasingly flexible and responsive assessment, monitoring and treatment programs.
Why is this? Well, we have actually been using asynchronous consultations in psychiatry for many years. Think of the traditional “curbside” consultation that every psychiatrist undertakes at the request of colleagues; this is an asynchronous consultation based on a conversation with a colleague. Asynchronous telehealth is already widely used in training residents, where the resident presents recordings of patient assessments to psychiatrist mentors for review. Similar asynchronous consults occur commonly in collaborative care programs and with e-visits, where there is no real-time interaction between a patient and a psychiatrist.
But what does AI add to asynchronous video recordings of patients? The two tasks that take most time in many psychiatric visits are patient interviewing and documentation. Both of these can now be done by AI, using highly realistic AI avatars that a patient can select to look like them, and by using language tools to summarize histories and write much of the medical note prior to the psychiatrist seeing the patient. AI can also potentially quantify the patients’ speech, facial and movement patterns, analyze language, and examine voice tone and sentiment all using the recorded voice and video files, giving the psychiatrist more information to make diagnostic and management decisions when they review the patients’ interviews. So, AI can turn asynchronous recorded interviews, perhaps used for triaging patients, into powerful decision support environments that can assist psychiatrists in their work while giving them more time to focus on their patient relationships and diagnostic and treatment issues.
Having said this, I do not believe that AI is yet ready to provide therapy or to make formal psychiatric diagnoses. The main current use cases are in the decision-support realm to triage patients and provide specialist opinions, perform screenings and intakes, including using avatar-driven versions of popular standardized interviews, and monitor medication management.
So, with all the benefits of asynchronous mental health visits, I believe it is inevitable that patients and doctors will take them up rapidly as an extra choice of consultation beyond in-person or real-time telehealth consultations. These benefits include:
1. Improved Access to Care: Patients can receive consultations without the need for appointments, reducing wait times and increasing accessibility.
2. Flexibility for Providers: Psychiatrists can review patient interactions at their convenience, allowing for better work-life balance and reduced burnout.
3. Efficiency in Consultations: Asynchronous methods streamline the consultation process, enabling providers to manage more patients and focus on data analysis and treatment planning, rather than spending so much time on interviewing and documentation.
4. Comparison over time: Psychiatrists can compare videos of patients over time to better judge treatment impacts.
5. Enhanced Patient Engagement: Patients can engage more easily with psychiatrist assistants, whether they be human or avatar interviewers, who look and sound like them, and who they may choose by age, gender and racial group.
6. Cost-Effectiveness: Asynchronous telehealth reduces travel costs for patients and operational costs for health care facilities, making it a financially viable option for both parties.
7. Scalability: The ability to handle a larger volume of consultations without compromising quality makes asynchronous telehealth a scalable solution for addressing the growing demand for mental health services.
With all these advantages, it is inevitable that a number of software companies will emerge offering the latest in telepsychiatry — asynchronous video visits. Not surprisingly, with a history of over a decade of research at UC Davis, the first company, Asynchealth, which I have co-founded, is already implementing these consultations in private psychiatric practices and with county mental health providers here in Sacramento. In these projects the main aims are to be able to assess and monitor more patients more effectively, while at the same time increasing psychiatrist and therapist flexibility and reducing burnout. This is an exciting time to be able to pioneer a new type of mental health consultation and is a credit to the innovative health care environment in our region.
