Melissa Bianchi, a partner in Hogan Lovels’ health practices and leader of the company’s digital health initiative, spurred discussions at the summit on healthcare AI regulatory issues about the potential of artificial intelligence to revolutionize the healthcare industry. I guessed. Its ability to help reduce waste, streamline payments and better diagnose patients. Still, the question remains as to how the industry can maximize its potential and whether there are barriers that prevent it from being fully realized.
What is “artificial intelligence”?
Panelists will first discuss how to define AI, and the American Medical Association (AMA) senior advisors Kathleen Blake, MD, MPH, and AI product sponsors are obliged to show the population on which the evidence was created. I started by emphasizing. Population-specific evidence is needed to safely extend the use of AI to new communities, representing communities around the world where technology is deployed. Dr. Blake emphasized that AI needs to be shown to increase fairness for all and to suggest meaningful patient outcomes.
Following Dr. Blake’s remarks, Kelly Hallezin, Director of Technology & Standards at the Consumer Technology Association (CTA), has published two of her organizations examining the definition of AI and the importance of AI reliability. Mentioned AI standards. Haresign warns that “trying to define AI broadly will get bogged down”, distinguishing between “supported intelligence” and “autonomous intelligence”, the latter category requiring no human intervention. To define AI, Dr. Blake states that AMA considers AI to be “extended” intelligence and recommends focusing on incremental gains from new technologies.
Bianchi highlights the relevance of standard settings in AI, with some of the goals of the standard being “building towards something that can facilitate regulated, more efficient and faster approval.” explained. Dr. Blake urged industry stakeholders to include patients early in the design of the study.
Overcoming data challenges in AI development
Moving on to the challenges associated with getting the large, high-quality datasets needed to build AI, Bianchi said HIPAA was drafted long before many of today’s innovations, so access to datasets. He pointed out that challenges would arise. Reflecting this concern, Dr. Blake described the competing goals of HIPAA’s “almost chimera.” This is intended to facilitate broader access to the data by the patient while ensuring the privacy of the data. Dr. Blake promoted more automated capture of data, data entered by patients, and increased access for patients to examine the data and correct errors.
Haresign said at the CTA that the industry recognizes the importance of proper handling of health data and has published industry practices to address this. Pointing out the challenges associated with relying on the data used by healthcare providers in AI algorithms, Bianchi asked the panel what the problems with accountability and bias would be. Dr. Blake replied, “Trust is comparable to accountability,” explaining how clear labeling can help solve this dilemma. Haresign said the level of trust required for an AI product corresponds to the level of risk associated with the drug or device.
