The easiest way is data mining by using K-Nearest Neighbour method, Expectation Maximization or kernel methods. See e.g., http://ijesc.org/upload/de2b0344e4610e19b902b9742382756b.Analysis%20of%20Diabetic%20Patients%20Data%20Using%20Data%20Mining%20Clustering%20Techniques.pdf
Article Performance analysis of clustering algorithms for diabetes data