Satisfiability based reverse analysis method in diabetes detection

Accurate diabetes diagnosis is vital to ensure patient with certain symptom can be treated appropriately. Medical diagnosis is often challenging because not all patient’s symptoms will arrive to the same diagnosis. Medical practitioner or specialist is not able to diagnose the disease accurately when the number of attribute that contribute to the disease increase. Logic mining is a platform for data scientist to induce the logical rule from large data set. In this study, 2 Satisfiability based reverse analysis method (2SATRA) will be proposed to extract the logical rule from the data set. The extracted logical rule from the data set will be use to classify the outcome of the diagnosis. 2SATRA is a method that utilized the beneficial feature of Hopfield neural network and 2 Satisfiability problems or HNN-2SAT. The simulation will be examined by using Microsoft Visual 2015 C++ Express software. The robustness of 2SATRA in extracting logical rule in medical data set will be evaluated based on root mean square error (RMSE), mean absolute error (MAE) and CPU time. The result obtained from the computer simulation demonstrates the effectiveness of 2SATRA in doing diabetis diagnosis.