AI trained by researchers from Pennsylvania healthcare provider Geisinger is capable of predicting patients who are more likely to die within a year based on the heart test results. The AI manages to accurately predict the death of patients even if the figures look quite normal to doctors.
The researchers trained the AI with 1.77 million electrocardiogram (ECG) results from 400,000 patients. They made two versions of the AI – one with just the ECG data while the other one had ECG data along with the age and gender of the patients.
Notably, in evaluation, the AI scored above 0.85 out of 1 while the current risk scoring models used by doctors land somewhere between 0.65 and 0.8. Impressive, isn’t it?
“No matter what, the voltage-based model was always better than any model you could build out of things that we already measure from an ECG,” says Brandon Fornwalt, lead researcher of the study.
As I mentioned earlier, the AI predicted the risk of patients which seemed normal by cardiologists. The report states that three cardiologists reviewed these results and were not able to get any signs of risk.
“That finding suggests that the model is seeing things that humans probably can’t see, or at least that we just ignore and think are normal. AI can potentially teach us things that we’ve been maybe misinterpreting for decades.”, Fornwalt told New Scientist.
However, there is no clear answer or explanation of how the AI manages to do so, making it unreliable in nature. This is the reason why medical professionals in the field are refraining from using such AI models. I guess, as technology advances, we will be getting valid explanations from researchers for the behavior of AI.
So, what are your thoughts on the use of AI in the medical field? Let us know in the comments.
