Restricted Boltzmann Machine Approach for Diagnosing Respiratory Diseases

• Published in: JOIV : international journal on informatics visualization Online Vol. 9; no. 4; p. 1664
• Main Authors: Haviluddin, -, Nurhalifah, Siti, Trahutomo, Dinnuhoni, Wibawa, Aji Prasetya, Utama, Agung Bella Putra
• Format: Journal Article
• Language: English
• Published: 31.07.2025
• ISSN: 2549-9610; 2549-9904
• DOI: 10.62527/joiv.9.4.3427

Respiratory diseases remain a significant global health challenge, particularly in developing countries where high morbidity and mortality rates persist. This study aims to establish a diagnostic approach for respiratory diseases using the Restricted Boltzmann Machine (RBM) method to support early detection and improve clinical decision-making. The research utilizes 180 medical records from patients at I. A Moeis Samarinda Hospital, East Kalimantan, Indonesia, includes 22 symptom variables associated with six respiratory disease types: sinusitis, pharyngitis, bronchitis, pneumonia, tuberculosis, and asthma. The collected data were preprocessed into binary formats to represent symptomatic and asymptomatic conditions, facilitating practical training in the RBM model. Data splitting was conducted with 70:30, 80:20, and 90:10 ratios for training and testing sets. The RBM architecture was optimized to enhance model performance by tuning key parameters, including the number of epochs, learning rate, and hidden neurons. Experimental results demonstrate that the RBM model achieved high diagnostic accuracy, with an accuracy of 98%, sensitivity of 98%, and specificity of 99% under the configuration of 5000 epochs, a learning rate of 0.1, and 53 hidden neurons. These findings indicate the model’s capability to recognize patterns and accurately classify respiratory diseases based on clinical symptoms. The study highlights the potential of integrating AI-based diagnostic systems like RBM into healthcare services, particularly in resource-limited settings. Future research should explore larger, more diverse datasets and consider environmental and socioeconomic factors to improve the model’s generalizability and practical applicability.