PENERAPAN ALGORITMA BACKPROPAGATION DAN OPTIMASI CONJUGATE GRADIENT UNTUK KLASIFIKASI HASIL TES LABORATORIUM

Abstract

A blood test is generally used to evaluate the condition of the blood and its components, conduct screening, and aid diagnosis. Blood tests in the laboratory are commonly used to deliberate whether a patient needs to be hospitalized or treated as an outpatient. Backpropagation algorithm was selected for its ability to solve complex problems. Conjugate gradient optimization is used because it facilitates faster solution search. An electronic medical record containing the results of patient laboratory examinations was obtained from Mendeley. The data was divided into training and testing with a 95:5 ratio, which was discovered to be the best ratio from the experiments. The best architecture was achieved by a combination of 10 neurons in the input layer, 16 neurons in the first hidden layer, 2 neurons in the second hidden layer, and a neuron in the output layer. Purelin is used as the activation function for both the first hidden and output layers, whereas the binary sigmoid is used for the second hidden layer. The analysis revealed that for 100 bootstraps in training data, the network worked with an average accuracy of 60.17% and a recall of 99.77%, while the accuracy results in testing data were 69.23%.