Breaking down healthcare analytics

So you've decided to enter the world of analytics, and you know you want to focus on the healthcare industry. However, that barely narrows down the problem space, as there are hundreds of open problems in healthcare that are being addressed with machine learning and other analytical tools. If you have ever typed the words "machine learning in healthcare" into Google or PubMed, you have probably discovered how vast the ocean of machine learning use cases in healthcare is. In academia, publications focus on problems ranging from predicting dementia onset in the elderly to predicting the occurrence of a heart attack within six months to predicting which antidepressants patients will best respond to. How do you pick the problem on which to focus? This section is all about answering that question. Choosing the appropriate problem to solve is the first essential step in healthcare analytics.

In healthcare, the problems to solve can be broken down into four categories:

• Population
• Medical task
• Data format
• Disease

We review each of these components in this section.