Perancangan Sistem Prediksi Deteksi Alzheimer Berbasis Random Forest Menggunakan Metode Scrum

Abstract

Alzheimer's disease is a neurodegenerative characterized by a gradual decline in memory and cognitive function, with a prevalence that continues to increase globally and in Indonesia. Constraints in early detection, such as limited healthcare facilities and the high cost of conventional diagnosis, drive the need for easily accessible technology-based solutions. This research aims to develop a web system named MindCare that integrates the Random Forest algorithm to predict the risk of Alzheimer's based on clinical and lifestyle data. The system development method uses the Agile Scrum approach with four sprint cycles, covering needs analysis, model training, web system integration, as well as testing and refinement. The model was trained using Alzheimer's and mental health datasets from Kaggle, with evaluation results showing perfect accuracy and AUC (100%). The features FamilyHistoryAlzheimers, Age, and PhysicalActivity proved to be the most influential in prediction. The resulting web system provides risk prediction features, result visualization, personalized prevention recommendations, and education about Alzheimer's. Black-box testing showed all functions worked as expected. The conclusion of this research is that the MindCare system is suitable for use as an easily accessible medium for early detection and education on Alzheimer's, with recommendations for further development through database expansion, exploration of other algorithms, and the addition of consultation and monitoring features