Stroke represents a global concern that currently affects a significant part of the world’s population. Physical rehabilitation plays a fundamental role for stroke patients to recover mobility and improve quality of life. This process is costly, considering that patients must attend face-to-face rehabilitation sessions in hospitals or rehabilitation centers. Plus, there is a lack of specialized medical staff, who are usually insufficient to properly address the growing number of stroke patients that need physical rehabilitation. This situation has been exacerbated by the COVID-19 pandemic, as some of the human resources have been devoted to fight against the pandemic, and the physical presence of rehabilitation patients in hospitals has been severely limited. This paper proposes PhyRe Up! , a novel remote rehabilitation system that uses mixed reality and gamification techniques. PhyRe Up! has been devised for stroke patients to perform therapeutic exercises at home, with great precision, and with the potential supervision of clinicians. The system aims to increase the patient’s motivation as well as maintaining the quality of performance for the exercises, similar to the obtained levels when attending face-to-face sessions with therapists. The underlying architecture combines declarative, procedural, and conditional knowledge to manage the rehabilitation process, which offers flexibility and scalability to enhance the capabilities of the proposed system. Experimental results highlight how the combination of mixed reality and gamification significantly influences the accuracy of rehabilitation exercises previously defined by therapists. Particularly, the conducted experiments in the first validation phase of PhyRe Up! shows that our proposal drastically reduces the intermediate steps required to complete an exercise thanks to the provided visual feedback. The accuracy with which the patient performs the assessed exercise for the first time is greater than when using traditional rehabilitation techniques.

Prolonged incorrect sitting postures can lead to various health issues, including musculoskeletal disorders and reduced quality of life. Efficient and accurate methods for monitoring and assessing sitting postures are crucial for promoting overall well-being in our increasingly sedentary society. This study compares the performance of Kolmogorov-Arnold Networks (KANs) and Multilayer Perceptrons (MLPs) in sitting posture recognition, utilizing a multi-task classification approach to simultaneously identify upper and lower body positions. A comprehensive dataset of sitting postures was developed using MediaPipe for skeletal extraction, labeled for both body regions, and encompassing a range of common scenarios. The models were implemented and evaluated using Leave-One-Subject-Out (LOSO) cross-validation to assess generalization capability across different individuals. Results indicated that the KAN model achieved higher accuracy (97.03% for upper body and 92.11% for lower body) with fewer parameters (4,320) compared to the MLP (93.87% and 91.47% respectively, with 7,662 parameters). Although the MLP demonstrated faster inference, both models achieved accuracies exceeding 91% in LOSO validation, suggesting their potential for real-time applications. This study significantly contributes to the field of automatic posture recognition, offering insights into the applicability of KANs and MediaPipe-based pose estimation in ergonomic recognition systems and demonstrating the effectiveness of a multi-task approach for comprehensive sitting posture analysis.

In the rapidly evolving domain of conversational agents, the integration of Large Language Models (LLMs) into Chatbot Development Platforms (CDPs) is a significant innovation. This study compares the efficacy of employing generic and fine-tuned GPT-3.5-turbo models for designing dialog flows, focusing on the intent and entity recognition crucial for dynamic conversational interactions. Two distinct approaches are introduced: a generic GPT-based system (G-GPT) leveraging the pre-trained model with complex prompts for intent and entity detection, and a fine-tuned GPT-based system (FT-GPT) employing customized models for enhanced specificity and efficiency. The evaluation encompassed the systems’ ability to accurately classify intents and recognize named entities, contrasting their adaptability, operational efficiency, and customization capabilities. The results revealed that, while the G-GPT system offers ease of deployment and versatility across various contexts, the FT-GPT system demonstrates superior precision, efficiency, and customization, although it requires initial training and dataset preparation. This research highlights the versatility of LLMs in enriching conversational features for talking assistants, from social robots to interactive chatbots. By tailoring these advanced models, the fluidity and responsiveness of conversational agents can be enhanced, making them more adaptable and effective in a variety of settings, from customer service to interactive learning environments.

Gait analysis has proven to be a key process in the functional assessment of people involving many fields, such as diagnosis of diseases or rehabilitation, and has increased in relevance lately. Gait analysis often requires gathering data, although this can be very expensive and time consuming. One of the main solutions applied in fields when data acquisition is a problem is augmentation of datasets with artificial data. There are two main approaches for doing that: simulation and synthetic data generation. In this article, we propose a parametrizable generative system of synthetic walking simplified human skeletons. For achieving that, a data gathering experiment with up to 26 individuals was conducted. The system consists of two artificial neural networks: a recurrent neural network for the generation of the movement and a multilayer perceptron for determining the size of the segments of the skeletons. The system has been evaluated through four processes: (i) an observational appraisal by researchers in gait analysis, (ii) a visual representation of the distribution of the generated data, (iii) a numerical analysis using the normalized cross-correlation coefficient, and (iv) an angular evaluation to check the kinematic validity of the data. The evaluation concluded that the system is able to generate realistic and accurate gait data. These results reveal a promising path for this research field, which can be further improved through increasing the variety of movements and the user sample.