Raw LLM Responses

Digesting millions of medical records will allow automation to find patterns that are difficult for people to find. I'm guessing one of the main hindrances to having statistical models of all the medical records is that statistics works best if you can control conditions and use THE SAME data. By learning ways to have data models "learned" by Neural Nets, we now have a means to codify data that is apples and oranges --- because, normal software up until now can store the data of apples and oranges, but not really know there is a difference other than "not equal." The data gleaned this way is going to be invaluable. We might actually be able to accurately predict cancer risks and formation by multiple mechanisms. We might actually learn what diets work for people of different genetics and life experience. We can learn things we weren't even asking the questions about.