I read the article, it picked out 586 variables though the only one mentioned is home visits which I grant is one that probably should be taken into account however what else could there be? It's not like the NHS has genetic screening and how would you quantify the relationship between other conditions? Lets say you have 5000 people who had heart attacks and 62% had a wart on their hand at some point in their life would the AI consider that a factor? Then what if you have 500,000 people with warts that didn't have a heart attack. How would this be measured using machine learning as you are only using the heart attack patients to build up your model then applying that to everyone? Do they consider these things?