Pengembangan Sistem Cerdas Berbasis Data Mining untuk Meningkatkan Akurasi Prediksi Kebutuhan Obat di Puskesmas Parit Rantang

Abstract

Effective inventory management is crucial in providing quality healthcare services. Predicting drug needs in the pharmacy warehouse is vital to ensuring adequate availability for patients. This study developed and implemented a prediction system using Artificial Neural Network (ANN) method to forecast drug requirements. Training data comprised drug usage from January 2020 to July 2023, while testing data covered drug usage from August to December 2023. Through several experiments, the best model identified was 12-6-1, with a Mean Absolute Percentage Error (MAPE) of 6.817 and an accuracy of 93.18%. Predictions for Paracetamol drug usage in August were 4603, whereas the actual usage was 4785. This system is expected to enhance drug inventory management efficiency, reduce costs, and improve drug availability for patients.