unsupervised learning in medical systems and intelligent systems?

I have a dataset which belongs to a hospital. It contains data about patients and healthy people. The problem is separating healthy ones from patients. I add some new features to dataset to solve this problem. When I reduce the dimensions of data including the new features and visualize the data, the patient and healthy individuals are distinguishable(visually separable). Now if one asks what is the relation between the used approach (feature extraction, visualization, using the human ability, unsupervised methods) and expert system, what can I say? How can I use this method to build an expert system?

Thank you in advance.

Topic unsupervised-learning knowledge-base visualization feature-extraction

Category Data Science

1
answered at

Expert systems rely on knowledge "extracted" from human subject matter experts (e.g. a team of physicians). They then use this knowledge to make prediction about the data (e.g. if blood pressure > x then patient).

Your approach used properties of the dataset you worked on. The knowledge was in fact extracted from the data themselves.

The second approach may or may not be compatible with expert advice. If your algorithm found that blood pressure was a good feature to split healthy people and patient, you will probably find experts agreeing. However, if your main feature comes from a combination of many variables (e.g. PCA) you may have trouble finding an expert validating your approach.

So the way you could use your approach would be to team up with subject matter experts and make sure that you use only features that have a meaning for them (i.e. don't over-engineer data).

About

Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.