Multi-task learning (MTL) has demonstrated significant advantages in the field of physiological signal measurement. This approach enhances the model's generalization ability by sharing parameters and features between similar tasks, even in data-scarce environments. However, traditional multi-task physiological signal measurement methods face challenges such as feature conflicts between tasks, task imbalance, and excessive model complexity, which limit their application in complex environments. To address these issues, this paper proposes an enhanced multi-scale spatiotemporal network (EMSTN) based on Eulerian video magnification (EVM), super-resolution reconstruction and convolutional multilayer perceptron. First, EVM is introduced in the input stage of the network to amplify subtle color and motion changes in the video, significantly improving the model's ability to capture pulse and respiratory signals. Additionally, a super-resolution reconstruction module is integrated into the network to enhance the image resolution, thereby improving detail capture and increasing the accuracy of facial action unit (AU) tasks. Then, convolutional multilayer perceptron is employed to replace traditional 2D convolutions, improving feature extraction efficiency and flexibility, which significantly boosts the performance of heart rate and respiratory rate measurements. Finally, comprehensive experiments on the Binghamton-Pittsburgh 4D Spontaneous Facial Expression Database (BP4D+) fully validate the effectiveness and superiority of the proposed method in multi-task physiological signal measurement.
Heart sounds are critical for early detection of cardiovascular diseases, yet existing studies mostly focus on traditional signal segmentation, feature extraction, and shallow classifiers, which often fail to sufficiently capture the dynamic and nonlinear characteristics of heart sounds, limit recognition of complex heart sound patterns, and are sensitive to data imbalance, resulting in poor classification performance. To address these limitations, this study proposes a novel heart sound classification method that integrates improved Mel-frequency cepstral coefficients (MFCC) for feature extraction with a convolutional neural network (CNN) and a deep Transformer model. In the preprocessing stage, a Butterworth filter is applied for denoising, and continuous heart sound signals are directly processed without segmenting the cardiac cycles, allowing the improved MFCC features to better capture dynamic characteristics. These features are then fed into a CNN for feature learning, followed by global average pooling (GAP) to reduce model complexity and mitigate overfitting. Lastly, a deep Transformer module is employed to further extract and fuse features, completing the heart sound classification. To handle data imbalance, the model uses focal loss as the objective function. Experiments on two public datasets demonstrate that the proposed method performs effectively in both binary and multi-class classification tasks. This approach enables efficient classification of continuous heart sound signals, provides a reference methodology for future heart sound research for disease classification, and supports the development of wearable devices and home monitoring systems.