Everybody loves a good story on life saving applications of AI. The WSJ article on “How Hospitals Are Using AI to Save Lives” is an example. If hospitals can identify patients that are at a higher risk of a Code Blue emergency before it happens, then lives could be saved.
An AI model that improves sepsis predictions of patients at one hospital will not necessarily do the same at another. As the article explains, when there is a mismatch between the data used to train the model and real world use, the AI fails to give correct predictions.
On a related note, check out Erik Stein’s piece on “Is AI testing like software testing?”. We get a lot of questions from our customers and partners on AI testing. This is a topic that you really need to understand to effectively use AI – even if you are using a pre-trained AI model.
- How Hospitals are using AI to Save Lives. Emergency rooms and ICUs are turning to artificial intelligence to identify and treat patients who are most at risk. (WSJ)
- Is Twitter Underestimating Bot Activity. Elon Musk is likely correct in his claims that Twitter provided inaccurate and insufficient information about the platform’s spam and robot accounts, according to former CIA cyber officer Dan Woods. (sDxCentral)
- Is AI testing like software testing? There’s much similarity between testing applied AI/ML (for brevity, I’ll just refer to these as “AI”) applications and traditional software. (SphereOI)