Artificial intelligence has undergone a remarkable evolution. In the time since ChatGPT was released in November 2022, a revolution has been playing out in the public square. But what is its role in healthcare? Dr. Neil Kudler, Chief Medical Officer at Pixel Health, recently shared his perspectives on the roots of AI in healthcare and now discusses where sees current possibilities for advancing patient care and optimizing providers’ processes in the clinic.
What is your interest with AI in healthcare?
As a physician for close to 30 years, I am intimately aware of the hurdles faced by healthcare providers as they balance the adoption of technology against spending quality time with a patient. Despite my years as an informaticist and technology evangelist, I would argue that “electronic solutions” have made it more difficult to create a comfortable and productive provider-patient relationship and contributed to provider burnout. Compound this with the documentation requirements for compensation. While good documentation correlates with good care, much of what is required is of questionable meaning in the context of a well-populated EMR.
I’m excited by the possibility of the provider in the exam room no longer having to stare at monitor and keyboard and sharing more time and eye contact with their patient.
What applications of AI do you see as supporting healthcare providers in the future?
I believe AI will shine in how it improves the experience of the provider in the exam room. I am following the work being done to implement AI-supported ambient voice recognition (AVR) documentation systems. Providing passive assistance for documentation and medical decision making (MDM) will not only ease the time pressure, but augment standardization in care delivery and improve clinical outcomes.
With exam rooms equipped with high-resolution microphones to capture conversations, which are then ingested by ML-AI platform, notes and orders can be drafted with intelligent support. The time saved, the care provided, and the improvement in outcomes and experience will finally make real the expectations we have long had of the electronic health record era.
How would this technology comply with HIPAA regulations?
When using AVR documentation tools, there is a potential risk of unintentional capture of sensitive information or conversations that are not directly related to the patient visit. Healthcare organizations must ensure that the technology is configured and used in a way that prevents unauthorized access to protected health information (PHI) and engenders trust by its stakeholders. Just like electronic medical reports are stored in a HIPAA-protected cloud, so would these patient-provider conversations. Surely, the role of the office of information security will be critical to decision-making.
I think these AI-enabled tools will have to allow a certain level of control to make sure the right privacy settings are in place. Ideally, only medically relevant information is captured, and other sensitive or medically irrelevant details can be deleted.
How will AI help support diagnoses?
I’ve already alluded to ambient voice recognition documentation tools that leverage AI for MDM. The opportunities for AI in the interpretation of images, specimens, and lab data are limitless. We’ve already seen data suggesting that AI can have a greater positive predictive value than humans, especially with pattern recognition. Humans are excellent at pattern recognition, but we are plagued by bias. There is also data pointing out that AI engines are also plagued by bias. My hope and assumption are that generative AI can be tuned to reduce or eliminate bias, at least while processing and comparing millions of images to reliably interpret a patient’s images for diagnostic use.
What does AI mean for healthcare organizations?
It seems to me that AI can be applied today, right away, to solve the problems of clinical documentation improvement and coding optimization. Either or both activities would serve both the fee-for-service and value-based environments. But my bigger hope is the near-time implementation of ambient voice recognition tools. With burnout rates at historic levels, I am certain that relieving providers in practical ways—note creation, medical decision-making support, writing prescriptions—will radically improve the day-to-day for my colleagues.
How is Pixel Health connected this these efforts?
What excites me most right now is our partnership with T-Mobile. We are now facilitating the introduction of private 5G networks to streamline access to proprietary platforms. The volume and velocity of data transmission will facilitate real-time AI-driven activities with unforeseen reliability. I am certain that the future of healthcare holds fewer clicks and more eye contact!