This is a part of collaborative project in which supervised machine learning approach with wrapper method was used for classification and feature extraction. A support vector machine based recursive feature elimination (SVM_RFE) was used to find a small subsets of significant features that together have good predictive power to distinguish injection heroin users from healthy controls.
Citation: Piepenbrink, M.S., Samuel, M., Zheng, B., Carter, B., Fucile, C., Bunce, C., Kiebala, M., Khan, A.A., Thakar, J., Maggirwar, S.B. and Morse, D., 2016. Humoral dysregulation associated with increased systemic inflammation among injection heroin users. PLoS One, 11(7), p.e0158641. (Open)