Machine learning dataset correlation

Question

I have a ML model that has features 1,2,3 and its working perfectly but now I suddenly get a test case with feature 'a' that for which it doesn't show right results how to handle that case? a detailed example would be heart attack prediction system trained for weight ,cholesterol level , height and age ,its working correctly for them but now I have a patient also has diabetes that increases the chances of heart attack. how to improve accuracy based on this uncorrelated new feature ?

Answer 1

Add the new feature and retrain your model.

If your first model is really perfect, you could add a structure to choose models. So if the person does not have diabetes, use the perfect model. Otherwise use another model.