Q & A with Joe Marks
Marks brings his extensive experience in artificial intelligence (AI) and human-in-the-loop automation—both integral components of Augmedix’s mission to facilitate improved clinical excellence through technological innovation.
Marks holds degrees in Applied Mathematics and Computer Science from Harvard University. He has run highly-acclaimed research labs at Mitsubishi Electric and The Walt Disney Company, worked on DARPA-sponsored research projects for the US government, co-founded two startups, and served as an advisor and consultant to the American Medical Association—all in addition to his current position at Carnegie Mellon.
Marks shared his thoughts regarding innovations in digital healthcare.
“Innovation happens in multiple settings: universities, big companies, small startups, government labs, etc. And in some way or another, each has contributed to the computational wonders of our time, such as personal computing, cloud computing, mobile computing, the Internet, GPS, perceptual user interfaces, and so on. My experience working in all these different innovation cultures has given me the ability to recognize each organization’s strengths and weaknesses, too, and glean best practices from each culture.
At Carnegie Mellon and in my work with various startups, I have worked on projects that utilize data collected in electronic health records (EHR) to ensure better diagnostics, improve operational efficiencies, and pinpoint areas for continuing education for healthcare providers.
While there are numerous benefits of EHR, there is also a considerable burden on providers to create and maintain an EHR for each patient. Augmedix’s innovative approach lightens that burden, makes EHR practical and useful, and facilitates better patient care. I find the possibilities exciting.”
Electronic Health Records
“Although I am a relative newcomer to the field of healthcare, as Executive Director of the University of Pittsburgh Medical Center-sponsored Center for Machine Learning & Health at Carnegie Mellon University, I am exposed to lots of healthcare-centric problems and technologies.
The Center currently funds 40+ projects in all areas of digital health and has helped launch seven healthcare-related startups in Pittsburgh. These opportunities have given me a broad perspective on the challenges and opportunities related to digital health—a very dynamic field.”
Artificial Intelligence and Machine Learning
“I have been working in artificial intelligence (AI) since the mid-1980s. One thing that drew me to Augmedix is the use of AI in capturing the nuances of the patient/provider relationship.
Data entry for EHR is complex. First, it requires an understanding of real-world, human-to-human communication. Second, it requires an understanding of the world of medicine, which is inherently complex and continuously evolving. Third, one hundred percent accuracy is crucial, and last, efficiency matters.
An all-AI, fully automated solution to this problem is not realistic considering those aspects, at least not for the foreseeable future. But that’s where machine learning (ML) comes into play. An application of AI, ML automatically learns and improves from experience without being explicitly programmed.
But it is essential to understand that ML is just one tool in the AI toolkit. Sub-areas of AI, including heuristic search, rule-based systems, logic programming, knowledge representation, planning, robotics, natural language processing, speech recognition, computer vision, intelligent user interfaces, each have their own rich set of concepts and techniques. I believe most real-world problems require a sophisticated hybrid of all these techniques.
Semi-automated solutions in which facets of AI augment human experts are very plausible. But they are tricky to design well. But I believe that Augmedix is in a unique position to lead in the development of realistic hybrid AI systems for EHR data entry, and also for EHR data analysis. And that excites me.”