Neuro-Fuzzy Classifier for Longitudinal Behavioral Intervention Data.

Fuzzy-logic based algorithms have been applied in learning longitudinal behavioral intervention data. This paper proposes a modified generalized network-based neuro-fuzzy (mGNNF) classifier for longitudinal randomized controlled trial (RCT) data with missing values. Specifically, using all available attributes, demographic, nicotine dependence, and intervention attributes, this proposed classifier is used to predict the prolonged smoking abstinence after a longitudinal smoking cessation RCT [1]. Our model comparison study shows that with the same longitudinal RCT data, mGNNF shows a higher accuracy compared to three similar fuzzy-logic based classifiers, although its computational time is slower than two of these comparators.