The following is a guest article by Erica Jain, Co-Founder and CEO at Healthie
ChatGPT has been dominating headlines since its release in November. While the AI Chatbot has already led to some truly whimsical poetry, comedic greetings, and helpful travel suggestions, it has also started to raise questions about its application in healthcare.
ChatGPT is revolutionizing the world – and this will hit healthcare, albeit a bit slower than other industries. Healthcare is inherently conservative and resistant to change, plus liability in a care setting is a massive issue. All that being said, in the 5 to 10-year time horizon, Large Language Models (LLMs) will displace trillions of dollars in waste and eventually revolutionize how care is delivered.
The Revolution Comes Slowly, Then All at Once
OpenAI’s timing is uncanny for digital health as an industry. We recently emerged from digital health’s giant leap forward in 2021 and 2022 – not just in terms of funding, but in terms of adoption and innovation. This timing matters because ChatGPT’s innovation will be layered upon this new digital frontier of healthcare: Virtual-first companies stand to benefit the most. We’ve seen other countries like China and India, apply Internet 2.0 technologies directly to Mobile and skip the Web, thus leapfrogging consumer adoption on a massive scale. Similarly, ChatGPT will be best applied to Virtual-first companies as opposed to traditional, in-person-first healthcare companies.
Virtual-first companies will bake all of their innovation into their platforms. Digital Health, as an industry, already made the mistake in the 2010s of “death by one thousand cuts” – creating a point solution for every problem. Instead, platforms will encompass this new tech. What used to be a “movement” in the industry will become single companies. What used to necessitate new companies will now be product launches. What used to be product launches will now become feature sets of software platforms.
Let’s dive into what this change might look like.
Transforming the Way Patients Find Care
Overall, asynchronous communication is set to skyrocket. Study after study shows the potential for asynchronous care to reduce time and increase efficiency – but with ChatGPT we finally now have the tools to accelerate that innovation.
Patients can now find care and interact with the front-office pre-visit in a way that they never could before. Searching for care, scheduling appointments, picking certain providers, providing insurance information, and giving medical history – this can all be done via a smart Chatbot. The days of “googling for doctors” are dwindling. In the beginning, of course, both parties will be skeptical. But over time the efficiency gains and technological advancements will breed trust. Surprisingly, clinician adoption is already increasing, as 46% of practices surveyed are using Chatbots. While that number may be high, it’s directionally indicative.
Reducing the Burden of Care for Providers
During the visit itself, ChatGPT (and similar tools) will allow the provider to focus on the patient rather than the administration layer around the visit. To start, studies show that scribing for doctors can save millions of dollars. The translation of that scribing directly into the EMR can do the same – this will also reduce provider burnout related to EMR fatigue.
LLMs won’t just augment the administrative layer – they’ll help providers with the burden of care as well. ChatGPTtool recently passed the U.S. Medical Licensing Exam (USMLE) without clinician input; 95% of the time it was also able to tell why the other answers were wrong (not just why the one answer was right). This level of sophistication lends itself naturally to suggestions of care pathways – long sought after as a holy grail in the industry.
Facilitating Better, More Personalized Care
After the visit – the management of “ongoing care” as we know it will slowly change. The lines between reminders, health coaching, and care delivery will all blur. This is because it will become difficult to distinguish between talking to a robot and talking to a human. Many interactions will now be “smart” as well as automated, meaning they will take into account patient input, previous history, and newly available data from the internet and other tools. Let’s look at an example to make this more clear.
Suppose there is a diabetic patient that requires ongoing management and care. Post-visit, the patient can be reminded to take certain blood samples by a Chatbot. They can input their results and symptoms, and the bot will recommend the next steps based on their medical history and personalized preferences. If a novel symptom is occurring, the bot can recommend a test for the patient, which the provider can approve within the chat flow. The patient can click “order” within the chat window, the ChatGPT will use an API to communicate with an at-home lab company to order the test to the patient’s home. The results get input, informing the provider and furthering the cycle of care.
Preserving the Most Wasted Resource in Healthcare: Time
Healthcare is a roughly $4 Trillion industry and studies show that between $200B-$600B of that spending is wasted. In addition to that, the administrative layer of healthcare is contributing to burnout. Studies show that care providers spend 2 hours in their EHR for every 1 hour they spend with patients, and this high burden is destroying their autonomy and love of the job. These tasks often require routine, automated, and programmable actions which are perfect for a bot.
The biggest area where AI makes its initial administrative impact is billing. Billing is a 20+ step process in which providers get paid 3-6 months after the visit. ChatGPT can help with revenue cycle management, claims submission, prior authorization, billing, denials and so much more. At least one physician made a ChatGPT bot for prior authorizations and it’s’ already reducing the admin workload. Expect to see much more clinician-driven innovation as utilization of these tools quickly becomes an “edge” for providers to find time in their overburdened schedules.
The AI Curve in Healthcare is Just Beginning
Ultimately innovation and adoption take time, especially in Healthcare. There are several staunch barriers to adoption which include trust in technology and workforce displacement as top reasons. However, we’re at or near a breaking point in the industry in terms of cost: physician burnout, consumer debt, and frustration are at all-time highs. Desperation, not the need for innovation, will drive adoption in the short run. Ultimately this may be the bridge we need to move the industry forward.
