The COVID-19 pandemic has propelled health tech innovation globally. While new technologies continue to emerge, the pandemic has also made the disparities in the health system evident.
Digital health tools have been pitched as a way to help fight health disparities and improve access, however, the technologies could also widen the care gap if not deployed correctly.
MobiHealthNews asked Paul Cerrato, senior research analyst at Mayo Clinic Platform, about the future of digital tools in healthcare, and his upcoming presentation at HIMSS22 with Dr. John Halamka.
MobiHealthNews: How do you think digital health tools could help address challenges of health equity?
Cerrato: To address health equity, developers and researchers need to start by improving the data sets upon which their algorithms are based. Those data sets have to be more representative of the patient population being served by the algorithms. As Dr. [John] Halamka and I explain in a soon to be published article in BMJ Health and Care Informatics, algorithmic bias is common because the data sets being used by insurers and healthcare providers often misrepresent people of color, women and patients in lower socioeconomic groups.
MobiHealthNews: Could digital health potentially create more disparities in health? How could that be remedied in the development process?
Cerrato: Yes, digital health tools have the potential to create disparities for several reasons, not the least of which is many come to market without strong scientific evidence to support them. An analysis of 130 FDA-approved AI devices, for instance, revealed that the vast majority had been approved based solely on retrospective studies. That’s rarely enough to justify their use in patient care. Prospective observational studies, and ideally randomized controlled trials, are needed to avoid subjecting patients to ineffective, biased digital tools.
MobiHealthNews: What validation needs to be conducted on digital health products in terms of equity?
Cerrato: At Mayo Clinic, we have created a suite of validation tools to improve accuracy, fitness of purpose and equity of digital tools currently being developed. Mayo Clinic Platform Validate, enabled by a large volume of de-identified data, can accurately and impartially evaluate the efficacy of a model and its susceptibility to bias. It helps measure model sensitivity, specificity and bias, and enables the breaking of racial, gender and socioeconomic disparities in the delivery of care. Validate lends credibility to models, accelerates adoption into clinical practice and enables meeting regulatory requirements for approval.
MobiHealthNews: Is there anything you would like to add about your presentation at HIMSS22?
Cerrato: Here’s the rationale behind choosing the theme for our presentation – Digital Health 3.0: Innovation > Validation > Equity. The industry is experiencing a three stage progression in healthcare AI. Initially, we have had enthusiasm for technologists and clinicians about all the innovative new AI fueled diagnostic tools. Now we are entering phase two, in which folks are questioning the value of these tools and looking for ways to separate useful technology from marketing hype — the validation phase. And we are slowly entering phase three, in which the tools are being reevaluated to make sure they are also equitable.
Cerrato’s session is entitled “Digital Health 3.0: Innovation > Validation > Equity.” It’s scheduled for Thursday, February 17, from 10:00–11:00 a.m. in Orange County Convention Center W414A.