So what ” most AI models described in reviews, and praised in press releases, are never used in the clinic “, and that ” the rare exceptions fall far short of their revolutionary expectations (cf. Artificial intelligence and medicine: real progress?), several organizations announced last Wednesday a plan to change this state of affairs. There ” Coalition for Healthcare AI brings together university hospitals, government agencies and private companies. Among them: the mayo-clinicBerkeley and Stanford Universities, the FDA, and the National Institutes of Health (NIH), or even GAFAM: Google Y Microsoft.
Understand the benefits and risks
” Today we don’t have the tools to understand whether machine learning algorithms and these new technologies being implemented are good or bad for patients. says John Halamka, president of Mayo Clinic Platform. ” The only way to change thatadded, is to study its impacts more rigorously and make the results transparent, so that users can understand the benefits and risks “.
the Coalition for Healthcare AI ask for the creation of independent testing bodies and a national registry of clinical algorithms to allow clinicians and patients to assess their suitability and performance “. EITHER ” from multiple labs across the country to allow developers to test their algorithms on more diverse data sets and audit them for bias “.
Framing the inevitable?
The group hopes its many members will help spark a national debate and “ concrete steps to start regulating the use of AI in medicine (cf. Health data: the patient turned into a product). He is also considering the idea of creating an accreditation organization that would certify the suitability of an algorithm for use in a given task or set of tasks “, with the objective of ” provide some level of quality assurance “.” We need to establish that AI-guided decision-making is useful “says Nigam Shah, professor of biomedical informatics at Stanford (cf. Digital health: patient first?).
 Artificial intelligence
Source: Stat news, Casey Ross (07/12/2022)