[Re: https://www.hifa.org/dgroups-rss/health-policy-watch-gates-and-openai-te... ]
Hi Neil, we agree that reliable health information is at the basis of any safe and effective intervention, including this one. Our team at Global Strategies adapts clinical guidelines into clinical decision support applications for LMICs. One of the most surprising aspects of this work is how often the guidelines themselves have gaps, ambiguities and even significant errors that are only revealed when you try to apply them at the bedside.
This led to an approach that is equal parts excitement about the potential and caution about the risks. The framework that guides us is hazard analysis, and I think that will be all the more important for outputs that have an element of unpredictability. Even in decision support that is predictable, hazard analysis led us to add logic to ensure users’ inputs make sense and were intended, especially around pediatric dosing — so that a miskeyed number won’t result in dangerous advice. AI is an exciting new technology, but we agree with you that its clinical applications should be carefully vetted and shaped, always keeping in mind the completeness and applicability of the health information that it will draw on.
--
Elon Danziger
Director of Software Design and Development
Keep In Touch With Global Strategies!
www.globalstrategies.org
Office: 415-451-1814
Facebook:
https://www.facebook.com/GloblStrategies/#
Instagram:
https://www.instagram.com/globalstrategies/
HIFA profile: Elon Danziger is director, software design & development, with Global Strategies in United States and has a professional interest in neonatal health. Email address: elon.danziger AT globalstrategies.org