I am a pediatric cardiologist and Chief Intelligence and Innovation Officer of CHOC Sharon Disney Lund Medical Intelligence, Information, Investigation and Innovation Institute (Mi4). I am the founder of Artificial Intelligence in Medicine (AI-Med), the Medical Intelligence Society (MIS), the American Board of Artificial Intelligence in Medicine (ABAIM), the Alliance of Centers in Artificial Intelligence in Medicine (ACAIM) and the Pediatric Centers of Artificial Intelligence in Medicine (PCAIM). My book, Intelligence-Based Medicine is the first of its kind textbook on AI in medicine and is used at colleges and universities around the globe.
“What is this thing that has happened to us? It’s a virus, yes. In and of itself it holds no moral brief. But it is definitely more than a virus. Some believe it’s God’s way of bringing us to our senses.”
Arundhati Roy, author of Azadi: Freedom, Fascism, Fictions
The COVID-19 pandemic seems to be better under control these wintry days, but the execution of the vaccinations seems to be almost as unpredictable as the pandemic itself. The algorithms used to help facilitate the distribution of the vaccines in the United States failed to a large degree, with media showing horrific pictures of the elderly in wheelchairs waiting over 10 hours in the arctic cold for their turn. The lessons learned from this real life deployment of simple algorithms are worth noting for the future….
The health agencies at the local, state, and federal levels have all developed their own allocation algorithms so that there was no consistency or coherence in the allocation schemes. In addition, the Tiberius algorithm used for nationwide vaccine distribution seemed to lack transparency and has been criticized.
Lesson: Algorithms need to generalize somewhat independent of venue and to be transparent.
The myriad of formulas used by hospitals and health organizations to allocate vaccines in some ways exacerbated the inequities that already existed in morbidity and mortality in African- Americans and Latinos. These vulnerable populations have a fraction of the vaccination rates as whites and this is complicated by the at-risk populations being undercounted.
Lesson: Algorithms need to in place to neutralize inequities as much as possible.
At Stanford, an early vaccine distribution algorithm was severely criticized for the very low number of frontline residents that were assigned slots for vaccinations (while senior administrators with little front line exposure managed to get to the front of the line). This prioritization strategy should have been observed directly and proactively by human administrators to detect any potential bias.
Lesson: Algorithms need to have human oversight with their common sense to detect any bias.
One state implemented an interesting incentive to gather the elderly for vaccinations: the accompanying person can concomitantly get a vaccination. Perhaps an algorithm would not be able to come up with that creative solution as it involves a solution with the complexity of human group behavior.
Lesson: Algorithms cannot yet be a source of creative solutions to existing complex problems.
It appears that we should take these lessons learned during the vaccination process and remember these valuable lessons for future deployment of algorithms in other clinical scenarios.