Artificial Neural Network for Predicting Abscess Disease Based on Patient Data Using the Backpropagation Method

Abstract

The development of information technology, particularly Artificial Intelligence (AI), has had a significant impact on the healthcare sector, including its application in supporting disease diagnosis. One of the common diseases encountered is abscess, a pyogenic bacterial infection characterized by the accumulation of pus in body tissues. At Bidadari General Hospital Binjai, abscess cases have shown fluctuations from 2023 to 2025, highlighting the need for faster, more accurate, and more efficient prediction methods to assist medical personnel in decision-making. This study aims to develop an abscess disease prediction model using an Artificial Neural Network (ANN) with the backpropagation algorithm based on patient data, analyze the model’s accuracy, and provide an alternative decision-support system for early diagnosis. The research method applied is quantitative-experimental, involving several stages: problem identification, collection of patient clinical data, data normalization, ANN architecture design, model training using backpropagation, and evaluation using accuracy metrics, Mean Squared Error (MSE), and Confusion Matrix. The prediction results indicate that the average number of abscess patients per month is projected to increase by 14.33% from historical records to the next 12 months. The historical monthly average was 14.9 patients, while the predicted average for the following year reached 17.0%. The model demonstrated good performance with a Mean Absolute Percentage Error (MAPE) of 244.25%. Therefore, the application of backpropagation-based ANN has the potential to serve as an effective solution in assisting medical personnel to perform early diagnosis of abscess disease in a faster, more accurate, and efficient manner.