Integrating AI Advancements Into Clinical Care

Featuring Josh Reischer, MD, CEO, Co-Founder, Health Note

Please share your name, title, and a brief overview of your professional history.

Reischer: My name is Josh Reischer, MD, and I’ve been in health care for nearly 20 years, completing my medical studies at the Ross University School of Medicine, a residency in Internal Medicine at the University of Arizona, and Biodesign and Innovation Fellowship at Texas Medical Center. 

One of my first experiences working in health care was as a hospitalist treating adult patients during inpatient stays. I was inspired by my colleagues and the care teams I was a part of, as their devotion to the profession made me believe in the fundamentals of what we were doing. As I branched out into outpatient primary care, working for Iora Health, I realized the challenging business dynamics that applied outside pressure to the operation of a health system. I knew there were opportunities for innovation, specifically around technology that could improve the way the EHR served doctors. 

My entrepreneurial journey began when I met my technical co-founder, Aaron Rau, at Y Combinator, and we developed Health Note as a platform to improve documentation and patient intake. The goal: improve the health care experience for patients and providers.

What inspired you to shift from clinical care to focus on the potential of integrating AI and other tech tools into health care?

Reischer: I am reminded of why I got into this line of work every time I make a doctor’s appointment for myself or a family member. We are stuck in the past, as it relates to how we communicate with and collect information from patients. 

In 2018, I stepped away from my role at Iora Health, and internal medicine, because it was becoming obvious to me that digitization was about to swallow the industry whole. While it may take a generation to properly vet, test, implement, and see the impact of these technologies, we will look back at this period (the early 2020s) as a turning point for innovation and adoption. 

Meeting my co-founder and chief technology officer helped to solidify some of my hypotheses. Over 6 years, we’ve evolved and expanded our platform several times, as technology continues to mature at an accelerated pace. We are also in a unique position to deliver innovation to remote parts of the country, and places that don’t always have the financial resources to pay and scale ‘big ticket’ generative AI tools, especially when they are already paying for an EHR.

While we began by focusing our intellectual capital on patient intake – all those key interactions at the beginning of a medical visit – it’s becoming apparent that those exam room conversations are where providers need more support.

Please share some practical examples of how AI and other tech tools integrated into clinical settings can support the provider and patient experience.

Reischer: When patients are sick, and in their most vulnerable state, the last thing they want is an endless game of phone tag trying to schedule an appointment. That can be done almost instantly today, with an online scheduler. This is not a threat to administrative roles in health care – it’s an asset and morale booster to help refocus time on more meaningful patient interactions.

Beyond this, I envision a process where all the surveying and questionnaires are delivered before an appointment on a tablet, smartphone, or computer and are relevant to that patient and that specific visit. At the very least, we hope that paperwork is minimized considerably to improve speed and accuracy. By collecting a brief medical history, patient medications, allergies, and other important insurance information automatically – and one-time only – the entire experience is simplified and tolerable. 

Finally, every exam room interaction with a nurse or doctor will be translated into bite-sized summaries, both for the doctor and the patient. Generative AI is being trained in dozens of languages and can interpret complex medical jargon in seconds. This software allows doctors to be present, engaged, and better informed before they walk into the door. Transcription tools are not new to health care, but the accuracy, speed, and user experience of the software have been “leveled up” significantly in recent years.

When a patient walks away from an encounter, they are finished and free to go home. No more check-out or additional homework. An abbreviated summary with to-dos and follow-up support can be delivered to a smartphone or via email. AI will touch every facet of the experience in some way, refining the way we interact with the system.

What are some potential limitations or challenges you see with implementing AI in clinical settings?

Reischer: I think we are going to need more extensive training, a measured roll-out, and advocacy for artificial intelligence in bipartisan circles. We are seeing major health systems award contracts to technology vendors almost weekly, but what you read in the media is not always a clinical reality. For generative AI to operate at full capacity and be adopted by a critical mass of users, it will take months and years of trust, iteration, and process development.

We cannot let the AI “run free”, either. Guardrails must be put in place to make sure that advanced analytics and critical thinking are aligned with modern medical training. In some of our earlier tests, we noticed that AI chatbots lack empathy and can consume research and studies that are inherently biased. These are areas that need to be ironed out if we are to continue to build on the foundation that’s already in place.

A recent American Medical Association study of medical professionals revealed two-thirds of physicians see advantages to using AI, but only 38% of those were actually using it in practice (this was conducted in 2023). 

More education is necessary, and technologists need to work even more closely with the medical community to get this right. If the design fails, utility strays from the original intent of the technology, and patient care can suffer in the end.

It’s an especially critical moment for underserved patients and ‘care deserts’ – places that are still struggling with adequate internet connectivity and workforce shortages. We need to shore up these vulnerabilities before the “AI gold rush” widens the chasm between the haves and have-nots of health care.

What advice do you have for other health care professionals looking to integrate AI into their practice?

Reischer: The most important thing you can do is to poll your staff and frontline workers to see what their impressions of artificial intelligence are. Having small focus groups demo and test the technology can help to create evangelists and ease the transition. 

What we know is that patients are experiencing this technology in every facet of their lives – from the Netflix account that recommends shows to the financial products that predict retirement savings. Artificial intelligence has the potential to personalize the health care experience for patients and allow doctors to be more effective in their work.

Artificial intelligence is not going to suddenly render all clinical training obsolete. It’s also not going to change how entrenched the EHR is in your workflow. But it will bring a new set of benefits that help reframe patient interactions and reduce things like “pajama time” (hours of post-office manual coding and note logging).

The use of AI in health care is evolving rapidly. What do you project to be the next big opportunities to advance health care with AI and similar tools in the next few years?

Reischer: We already have seen tremendous advances in the way notes are organized and presented in clinical care settings. We are now beginning to see an AI-driven shift to preventive care, which could create a more stable and viable system in the future.

AI is becoming increasingly adept at spotting trends among populations and identifying the risk criteria to surface patients who may be due for certain testing, like an age-appropriate colonoscopy. This can be generated by using data on family history, blood tests, or other presented symptoms that get recorded into a master database.

From there, AI can act as a reminder and guide to help schedule screening, panel testing, or other measures to anticipate these individual conditions. 

We all want to see a more unified health care system that allows patient data to travel between specialties and sites of care, regardless of who is delivering the care. AI has the potential to reduce chaos, shorten lines, and release the pressure faced by a reactive system built around the emergency department.

The future is already here, it’s just going to take a coalition of stakeholders and financial incentives to make it universally available across the country.