The Application of Artificial Intelligence (AI) in Healthcare; An Interview With MayaMD.AI Founders

Featured in Volume 76 of the Nevada State Board of Medical Examiners Newsletter. View original article.

Overview

By now, you’ve heard the term Artificial Intelligence (“AI”). Simply stated, AI is intelligence demonstrated by machines, whereby natural intelligence (typically referred to as “IQ”) is displayed by humans and animals. According to John McCarthy, a leading authority at Stanford, “[i]t is the science and engineering of  making intelligent machines, especially intelligent computer programs. It is related to the similar task of using computers to understand human intelligence, but AI does not have to confine itself to methods that are biologically observable.”1

So, what is the potential for AI in healthcare? First and foremost, it should be noted that AI is a compliment to and not a substitution for traditional person-to-person medical care. According to a publication posted by the U.S. Department of Health and Human Services (“HHS”): 

Artificial intelligence can help transform healthcare by improving diagnosis, treatment, and the delivery of patient care. Researchers in academia, the private sector, and government have gained increasing access to large amounts of health data and high-powered, AI-ready computing systems. These powerful tools can greatly improve doctors’ abilities to diagnose their patients’ medical issues, classify risk at a patient level by drawing on the power of population data, and provide much-needed support to clinics and hospitals in under-resourced areas.2  

Additionally, as noted in its February 2020 publication, A.I. Application and Security Implications in the Healthcare Industry, HHS cites data from Accenture, which estimates AI in healthcare will be a $6.6 billion market by 2021.3 Needless to say, AI is here to stay.

In light of the increasing utilization of AI in healthcare, I contacted Christian Haberman, MBA, a business school classmate of mine, as well as Dr. Vipindas Chengat, his fellow co-founder of MayaMD.AI,4 a Nevada corporation, in order to gain a practical understanding of AI in healthcare. The purpose of this article is not to address various regulatory considerations, nor is it an endorsement of any particular AI platform; rather, it is to give physicians and other providers an understanding of how AI works and how it is being deployed from a practical standpoint in a variety of clinical settings and disease management scenarios.  

Q&A 

RR: The use of AI appears to have “exploded” in recent years. From your perspective, what led to its increased use in healthcare and why did you co-found MayaMD? 

VC/CH: AI has been utilized in healthcare for many decades actually, but more recently, and especially this past decade, it’s taken off for a few reasons. There’s so much excitement around its potential, and along with that has come a tsunami of money invested in this space. The EHR has been a real game changer, as well as the processing speed of computers today. All of this momentum and data collection has created a kind of perfect storm for growth. We believe this is a new chapter and an exciting time in the healthcare industry. 

RR: What are the different types of solutions that various AI companies like MayaMD offer? 

VC/CH: This is where AI gets so exciting and shows unlimited potential. Right now we are witnessing some major breakthroughs, like in radiology, how AI is assisting lab evals and in some cases actually out performing human efforts. At MayaMD, we use AI to help patients and providers receive timely and appropriate triage and care coordination insight. In turn, patients receive the appropriate care and providers can use their time and financial resources more efficiently and effectively. Way too much time and money are spent on patients receiving care at facilities where it’s not efficient for either the patient or the provider. We hope to alter this behavior with our AI solution acting like a digital front door for triage, creating a more convenient and user-friendly patient experience. Our AI engine, MayaMD, can collect over 25 decision points in less than 90 seconds from the patient via our app. With this data it creates a secure, shareable clinical note with differentials sorted by probability, suggested labs, physical signs, and triage with maps to the nearest provider. This e-note can be instantly sent to the patient's provider, so that on the other end, the provider has the intake done in seconds -- saving documentation time, which has been the bane of clinics today. 

Locally, here in Las Vegas, we are just starting some AI projects with a nephrologist group, helping to educate and manage their patients with chronic kidney disease (CKD). Our clinical engine is versatile and customizable; therefore, we can create chronic disease management algorithms very quickly for just about any condition. Working with this local nephrology group, we are creating treatment, as well as health literacy algorithms, that can be used not only just for patients with CKD in Las Vegas, but anywhere. It’s exciting indeed.  

RR: How can AI be utilized in value-based or patient-centered care, which is the cornerstone of Accountable Care Organizations (“ACOs”), as well as the Affordable Care Act (“ACA”)? 

VC/CH: AI can be a real game changer and positive asset for ACOs since the key objectives include trying to optimize performance and improve patient outcomes by delivering high quality care. By utilizing AI machine learning to better understand a patient population, ACOs can structure their programs and infrastructure to support patients in an effective and efficient manner. For example, using data insight to reduce unnecessary Emergency Room visits and costly readmissions, as well as implementing preventive treatment programs to minimize expensive reactionary procedures, which may have been avoided. With more data insight, clinicians can get ahead of the treatment curve to ideally manage their patients’ health more effectively. AI data insight can allow a clinician to behave more proactively, which is ideal for all of us.  

RR: It is my understanding that the American Medical Association is providing grants for companies to work with universities. How does MayaMD hope to utilize the grant to explore AI in relation to diagnosis error rates? 

VC/CH: We actually had four leading universities and a national organization, the Society to Improve Diagnosis in Medicine (SIDM), support us for an AMA grant last year. This grant enabled us to create a clinical reasoning curriculum for students, which integrates into our platform. With this educational program we hope to help students understand cognitive bias and how it relates to diagnosis error. We work with some top medical universities and try to improve students’ clinical reasoning and diagnostic accuracy. It’s energizing, and it is just in its infancy. 

RR: During COVID-19, telehealth utilization became necessary to continue care when possible. How can AI integrate with and be utilized in tandem with telemedicine? And, how does this differ from its application to in-person treatment? 

VC/CH: AI applications during the pandemic have been quite useful, from simple symptom checking algorithms used in chatbots to more complex things where patient data is being analyzed using machine learning to help us better understand risk factors and safety protocols. The engine captures appropriate patient history in less than a minute, which can improve efficiency of telehealth visits. The tool can also perform a “pre-triage” and determine if a particular symptom or symptoms are appropriate for a teleconsultation or whether the patient would require an Emergency Room visit.  

RR: What is some practical advice for providers who want to integrate AI either into their practice or enable their patients to utilize it to potentially improve disease outcomes?  

VC/CH: One of the biggest challenges is the lack of structure in EHR data. Most of the EHRs are modified billing systems and are not designed to assist with clinical decision making. A lot of times, there is too much noise in the data or lack of completeness. MayaMD can change that. AI can help to capture data in a structured format, which can subsequently be used in an efficient manner by the machine learning algorithms. 

But if a physician is uncertain as to where to start incorporating AI into a practice, then start small, learn to crawl, walk, and then eventually run. It is suggested to review one’s health system or practice and select a few areas where a physician feels AI could help improve performance. Then, form a team to manage the research, vendor selection, and cost benefit analysis, as well as the crucial workflow integration. It’s definitely beneficial to have some fantastic data insights that AI can allow, but you need internal buy-in or utilization to get a return. We recommend that clinics take an incremental approach for implementation. 

Conclusion 

While AI is not a panacea and is not perfect, it can be used as another tool for both providers and patients to improve clinical outcomes. Additionally, because of its predictive nature (depending on the sample size and variables), it may be helpful in managing large-scale issues, such as COVID-19. Hopefully, this article provided a bit more insight into this evolving area of technology and its intersection with healthcare.  

1 Stanford University, What is AI?/Basic Questions, http://jmc.stanford.edu/artificial-intelligence/what-is-ai/index.html (last visited Nov. 14, 2020).  

2 Center for Open Data Enterprise, Sharing and Utilizing Health Data for AI Applications – Roundtable Report (2019), 

https://www.hhs.gov/sites/default/files/sharing-and-utilizing-health-data-for-ai-applications.pdf.  

3 HHS, A.I. Application and Security Implications in the Healthcare Industry (Feb. 6, 2020), https://www.hhs.gov/sites/default/files/ai-application-and-security-implications-in-healthcare-industry.pdf?language=en 

4 MayaMD.AI, https://mayamd.ai/about-us/ (last visited Nov. 14, 2020).  

Rachel V. Rose - Attorney at Law, PLLC (Houston, Texas) - advises clients on healthcare, cybersecurity and qui tam matters.  She also teaches bioethics at Baylor College of Medicine. She has consecutively been named by Houstonia Magazine as a Top Lawyer (Healthcare) and to the National Women Trial Lawyers - Top 25. She can be reached at rvrose@rvrose.com.  

‍

Disclaimer: The opinions expressed in the article are those of the authors, and do not necessarily reflect the opinions of the Board members or staff of the Nevada State Board of Medical Examiners.

‍