Human pose estimation is a crucial task in computer vision with widespread applications in healthcare, rehabilitation, sports, and remote monitoring. In this paper, we propose a deep learning-based method for automatic human pose estimation and joint angle computation, tailored specifically for physiotherapy and telemedicine scenarios. Beyond pose estimation, the proposed method is able to compute angles between joints, enabling analysis of body alignment and posture. The proposed approach is built upon a customized skeleton with 25 anatomical keypoints and a dataset composed of over 150,000 annotated and augmented images derived from multiple open-source datasets. Experimental results demonstrate the effectiveness of the proposed method, achieving a mAP@50 of 0.58 for keypoint localization and 0.98 for object detection. Moreover, we demonstrate several real-world practical use cases in evaluating exercise correctness and identifying postural deviations by exploiting the proposed method, confirming that the proposed method can represent a promising approach for automated motion analysis, with potential impact on digital health, rehabilitation support, and remote patient care.

Microsatellite instability represents a key biomarker in gastrointestinal cancers with significant diagnostic and therapeutic implications. Traditional molecular assays for microsatellite instability detection, while effective, are costly, time-consuming, and require specialized infrastructure. In this paper we propose an explainable deep learning-based method for microsatellite instability detection starting from the analysis of histopathological images. We consider a set of convolutional neural network architectures i.e., MobileNet, Inception, VGG16, VGG19, and a Vision Transformer model, and we propose a way to provide a kind of clinical explainability behind the model prediction through (three) Class Activation Mapping techniques. With the aim to further strengthen trustworthiness in predictions, we introduce a set of robustness metrics aimed to quantify the consistency of highlighted discriminative regions across different Class Activation Mapping methods. Experimental results on a real-world dataset demonstrate that VGG16 and VGG19 models achieve the best performance in terms of accuracy; in particular, the VGG16 model obtains an accuracy of 0.926, while the VGG19 one reaches an accuracy equal to 0.917. Furthermore, Class Activation Mapping techniques confirmed that the developed models consistently focus on similar tissue regions, while robustness analysis highlighted high agreement between different Class Activation Mapping techniques. These results indicate that the proposed method not only achieves interesting predictive accuracy but also provides explainable predictions, with the aim to boost the integration of deep learning into real-world clinical practice.

In recent years, the widespread availability of wearable devices and smartphones has enabled the large-scale collection of human activity data, fostering new opportunities for automatic workout recognition and personalized fitness monitoring. However, the centralized storage of video recordings raises critical privacy concerns, particularly when raw data contain identifiable individuals. Federated Machine Learning provides a paradigm designed with the aim of reducing privacy risks; here, models are collaboratively trained across distributed clients without sharing their sensitive data. In this paper, we propose an approach for workout video classification with Federated Machine Learning, enhanced by explainability through Gradient-weighted Class-Activation Mapping. The proposed method is evaluated on a real-world multi-class exercise video dataset, organized into eight biomechanically coherent macro-classes. In the experimental analysis, we consider several federated configurations in terms of the number of clients, the chosen aggregation strategy, and global communication rounds. The obtained results demonstrate that different aggregation strategies achieve comparable overall accuracy, while explainability effectively highlights the discriminative regions associated with exercise execution, revealing meaningful differences in model behavior between aggregation strategies and uncovering misclassifications driven by contextual biases, demonstrating the trustworthiness of the proposed approach for explainable workout video classification.