Respiratory Disorder Detection and Performance Analysis from PPG Signals using Short-Time Fourier Transform (STFT) Approach

Respiratory disorders pose a significant health concern, and their early detection is crucial for timely intervention and treatment. This paper presents a novel approach for the detection of respiratory disorders through the analysis of Photoplethysmography (PPG) signals. We employ the Short-Time Fourier Transform (STFT) to extract essential spectral features from PPG signals, allowing us to monitor respiratory patterns and identify anomalies. The proposed method offers a non-invasive and cost-effective means of assessing respiratory health, making it suitable for widespread deployment. Furthermore, our work includes a comprehensive performance analysis of the STFT-based approach, demonstrating its effectiveness in detecting respiratory disorders. The findings hold promise for improving healthcare by enabling early diagnosis and intervention in respiratory-related conditions. In the first level, STFT technique is employed to reduce the dimensionality of PPG signals. To evaluate the effectiveness of these techniques, five suitable classifiers as Principle Component Analysis, Logistic Regression, Gaussian Mixture Modeling, K Nearest Neighbor, Bayesian LDC applied. The classifier known as Bayesian LDC, which incorporates the STFT dimensionality reduction technique, achieves the highest level of accuracy as 78.05%, F1 score as 76.92 %, MCC as 0.56, error rate as 21.95%, Jaccard metrics as 62.50% and CSI as 53.95 %.