Can machine learning be used to treat diseases?

[r731]

Say a person is infected with disease X. I suspect that it's possible for computers to learn the behavior of X via machine learning. This would lead to greater understanding of X. The unsupervised learning approach is pertinent to this.

For another approach, say a new medicine Y is being designed. It should also be possible that an algorithm trains Y via maximizing the cumulative reward for its treatment of the disease.

 

[Choppy]

There are lots of examples of machine learning applications in medicine. To name a few from my own field...

• Outcome prediction. One of my students in currently using a large prostate cancer radiotherapy outcome data set to both identify any unknown factors that may influence the outcome and build a model that predicts outcome for a given case.
• Another student I've worked with has used machine learning to identifying cases in cancers of the head and neck where replanning the radiotherapy treatment part way through a given course of treatment would be beneficial.
• Aiding radiologists in the detection of diseases. There's been a lot of work recently on using various algorithms to detect COVID-19 in chest x-rays and other imaging modalities. Detection of cancer in a host of imaging modalities is another example.
• Last week I attended a vendor presentation of new ML software that used to automatically delineate organs in CT images.