Paediatrics Respiratory Diseases Diagnostic and Prediction System Using Deep Learning Model

The study presents a deep learning-based diagnostic and prediction method for paediatric respiratory illnesses that have a significant global impact on children's health, notably upper respiratory tract infections (URTI), chronic obstructive pulmonary disease (COPD), bronchiolitis, and pneumonia.The proposed framework combines Gated Recurrent Units (GRU) to describe sequential patterns in Mel-Frequency Cepstral Coefficients (MFCCs) and Convolutional Neural Networks (CNN) to capture local temporal features through the analysis of respiratory sound recordings.The model, which was trained on a labelled dataset, performed 84% of the time and showed good diagnostic and prediction abilities, particularly for cases of bronchiolitis (92% precision and recall) and healthy (100% precision and recall).The model's potential as an accurate and easily accessible tool for diagnosing and predicting paediatric respiratory diseases is proven by the results, despite a few misclassifications.