For medicine there are applications of machine learning mainly for diagnosing illnesses.
The following describes two example applications.

1) Analysis of x-ray images

Physicians are specialists in recognizing little patterns and abnormalities on x-ray images.
This requires a lot of experience.
Deep neural networks are algorithms which achieve super-human performance in image recognition tasks,
just like computers have been super-human in adding and multiplying numbers for a long time.

The advantage of computers in image recognition tasks is that they can learn from examples equivalent to the experience of many thousand physicians, whereas the performance of a human can only be based on the number of examples that a human sees during one life.

2) Diagnosing illnesses on the basis of personal medical data

A general practitioner diagnoses illnesses on the basis of data like the patient's blood pressure, age, hight, sex, blood key numbers,...
These are classical classification tasks in machine learning.
Like in the first example, machine learning algorithms can learn from many more examples than a human can. The human experience can never be more than the example cases that he sees during one life span.

Machine learning should not be seen as a competitor to human physicians, but rather as a helping tool in order for physicians to become more time efficient in diagnosing and to spend the saved time on other tasks.