An interesting analogy in this article!
“I think of machine learning kind of as asbestos,” he said. “It turns out that it’s all over the place, even though at no point did you explicitly install it, and it has possibly some latent bad effects that you might regret later, after it’s already too hard to get it all out.”
“AI is just as likely to perpetuate bias as it is to eliminate it”
https://www.statnews.com/2019/06/19/what-if-ai-in-health-care-is-next-asbestos/
I think that the high profile use cases or clinical use cases are the most risky. There are plenty of use cases which are for very mundane uses which should have no impact on outcomes (same as with asbestos).
Of concern is that pervasiveness - Asbetos can be controlled through regulation, it isn’t clear to me that there will or can be effective regulation of machine learning (question is what do you regulate).
opengraphobject:[360463459475456 : https://www.statnews.com/2019/06/19/what-if-ai-in-health-care-is-next-asbestos/ : title=“What if AI in health care is the next asbestos? - STAT” : description=““I think of machine learning kind of as asbestos,” one speaker said, explaining that AI in medicine could turn out to be as harmful as it is promising.”]
agree with what @jon_herries said, adding a few examples to that
ML/AI use cases in healthcare/healthtech can be broadly categorised into - operational ED demand forecastingpredicting in-patient length of staypatient flowpatient satisfactionclinical detecting diseases from Xray, ECG, EEG… managing drug dosagesautomated triaging
My view is that, operational efficiencies, brought by augmenting ML/AI models, should be considered as feature enhancements to existing softwares systems, auto-completion suggestions in PAS systems, for example. These improvements have huge potential to free up precious clinical time in turn delivering better care.
On the other hand, clinical decisions augmented by ML/AI model may be regulated as they are for any new drug or what IMDRF suggests Software as a Medical Device (SaMD) , FDA has discussion paper on the same topic https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-and-machine-learning-software-medical-device
opengraphobject:[360463459475456 : https://www.statnews.com/2019/06/19/what-if-ai-in-health-care-is-next-asbestos/ : title=“What if AI in health care is the next asbestos? - STAT” : description=““I think of machine learning kind of as asbestos,” one speaker said, explaining that AI in medicine could turn out to be as harmful as it is promising.”], opengraphobject:[360463881953280 : https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-and-machine-learning-software-medical-device : title=“Artificial Intelligence and Machine Learning in Software as a Medical Device | FDA” : description=“”]