A paper published in the Lancet Digital Health found that less than 1% of clinical AI studies reviewed were of sufficient quality to have confidence in their results. This is a huge problem. Dr Xiao Liu and Professor Alastair Denniston are part of the team who published that study — and to tackle this this problem — they have today published the first international standards for the reporting of clinical AI trials. These are called SPIRIT-AI and CONSORT-AI and have today been published in Nature Medicine, the Lancet Digital Health and the BMJ. The goal of these guidelines is to improve the quality of clinical AI studies and increase their transparency. We talk about the general quality problem in clinical AI research, we talk about their guidelines and some of the recommendations within them — and also about what the future of clinical AI validation looks like. Papers Nature Medicine CONSORT-AI: https://doi.org/10.1038/s41591-020-1034-x SPIRIT-AI: https://doi.org/10.1038/s41591-020-1037-7 The Lancet Digital Health CONSORT-AI: http://www.thelancet.com/journals/landig/article/PIIS2589-7500(20)30218-1/fulltext SPIRIT-AI: http://www.thelancet.com/journals/landig/article/PIIS2589-7500(20)30219-3/fulltext BMJ CONSORT-AI: https://www.bmj.com/content/370/bmj.m3164 SPIRIT-AI: https://www.bmj.com/content/370/bmj.m3210

Ask this episode

“What did the council decide on item 42?” — answered with the speaker and the moment.

Transcript

Related