Design of a mobile headache detection application with Naïve Bayes classifier method

Recently, many people still ignore the dangers of headaches and have not received yet the effective health care. This condition happens because the communities' awareness are still low and lack of knowledge about the type of headache experienced. This study aims to detect type of headache early with the Naive Bayes Classifier on Android. The Naive Bayes Classifier method includes probabilities' calculations in each class of all data (prior), probabilities' features calculations (likelihood) and multiplying of those two probabilities. The highest multiplications values would become the result of detection. The features which were used in headache detection were classified into two, namely red flags and primary headache. The red flags feature would be detected in the first detection, and the primary headache would be detected in the second detection. In the testing process gave accuracy, sensitivity, and specificity at first detection all with 100% values. Whereas the second detection produced 96.67% accuracy, sensitivity of migraine class was 100%, sensitivity of cluster class was 80%, sensitivity of Tension-Type Headache (TTH) class was 100%, specificity of migraine class was 92.86%, specificity of cluster class was 100% and specificity of TTH class was 100%. The results of accuracy, sensitivity, and specificity in this study were proven that the application had a good performance.