Framework for evaluating AI tools in healthcare

Had the pleasure of watching Dr. David Ouyang, MD's presentation on #AI in #healthcare at Fédération des médecins spécialistes du Québec (FMSQ) 's "Journeés de Formation Interdisciplinaire" last Friday November 15th.

Some key takeaways from this presentation was geared towards the "framework for evaluating AI in healthcare".

Science tells us what we can do.

Clinical Trials tells us what we should do.

Observational studies tells us what we are doing.

Some questions we should be asking when evaluating any AI tool are;

• Is the training data robust?
• What training Models are used for the tools?
• How much data is being used to train these models?
• What are some of the practice patterns across US Canada, Europe...etc?
• Are some these AI Models open for inspection?
• Has there been any Clinical Trial?

When looking at tools such as #AIScribe from the lens of a health practitioner, the rewards of using such tools are endless. Nevertheless, you should be looking at the legal risks of AI and how to mitigate them