I am participating in a debate Thursday at Melbourne Connect where Robert Osler and I will be opposing the proposition that The Risks Outweigh the Benefits for AI in Healthcare (come if you can). In preparing for the debate, I’ve been talking to colleagues and thinking about the risks innovation inevitably presents. Daniel Capurro brought up the analogy of surgery and the scientific method. We start with a theory and devise interventions based on the theory. Only through experimentation can we learn whether the intervention actually results in better outcomes and how to get the best outcomes. If we are collecting data from our practice and learning from that data, we will revise our theories accordingly. We used to operate on many conditions we now know do not require surgery, and the act of surgery is very high risk for patients. We accept risk for the potential benefits.
It reminded me of the history of liver transplant surgery:
In 1963, Starzl et al. performed the first liver transplantation. In the first five liver transplantations no patient survived more than 23 days. In 1967… Starzl began a successful series of liver transplantation. Until 1977, 200 liver transplantations were performed in the world. In that period, technical problems were overcome.
A colleague and friend, Wallis Marsh, was in the second generation of liver transplant surgeons under Starzl in Pittsburgh (see the city to the right). He told us that they would line the doorways of the surgical theatre with towels, because the patients lost so much blood it would run into the hallways. A patient having a liver transplant now will hardly lose any blood.
We are at the beginning of learning how to leverage AI for the benefit of patients, and we should expect risk and therefore design our interventions to minimize that risk while continuously learning to ultimately reap the benefits. We must go down that path, because as Brian Chapman pointed out in a conversation, to not pursue AI in healthcare is to say healthcare will not use data, which is unthinkable. I’ve stolen our thunder for the debate, but you can come see how well we address the other side’s arguments!
Regional Linguistic Quirks (RLQ): In a meeting with Chuck Friedman last week regarding the Mobilizing Computable Biomedical Knowledge symposium (still time to register!), he said, “Let’s deep six that idea.” Philip Scott, from the UK, had never heard the phrase, so I wondered if it was new to Australians as well. Chuck thought it came from the Watergate era, but in fact it is a North American saying originating from the 1920’s: as the deep six ‘the grave’: perhaps from the custom of burial at sea at a depth of six fathoms (which I also had to look up - 72 inches)..
photo from https://au.hotels.com/go/usa/pittsburgh
Comments