Abstract:
The Respiratory Disease Diagnosis and Management System (RDM) employs advanced machine learning, notably Convolutional Neural Networks (CNNs), to improve diagnostic accuracy for respiratory conditions like Chronic Obstructive Pulmonary Disease (COPD) ,Esthma ,UTRL etc. Utilizing a curated and augmented dataset from the Respiratory Sound Database, RDM's neural network combines CNN, LSTM, and Seq Self Attention layers, achieving a remarkable 99% accuracy in disease diagnosis. The Medicine Prescription Model predicts optimal medications, dosages, and frequencies based on patient-specific factors using Random Forest classification, boasting an average accuracy of 98%. Notably, the training data for this model is synthetic and intended for educational insights rather than clinical advice. The project aims to create a voice-based diagnostic tool, enhance medical automation, educate on respiratory diseases, and ensure global accessibility. Methodologies include data augmentation, feature extraction, model design, and rigorous evaluation. Future efforts will concentrate on expanding the dataset, refining algorithms, integrating into clinical workflows, addressing privacy concerns, and facilitating global deployment, ultimately aiming to transform respiratory disease diagnosis and management through advanced AI technologies.
