2 results about "Stress indicator" patented technology

The present application relates to the technical field of affective computing, in particular to a large model emotion embodiment computing method, system, device and storage medium. The large model emotion embodiment computing method reversely deduces implicit physiological feedback from input data; extracts multi-dimensional stress indicators from multi-modal data to construct a dynamic context semantic field; inputs the implicit physiological feedback and the context semantic field into an emotion dynamics system to generate a continuous emotion state evolution trajectory; fuses the emotion state evolution trajectory and the semantic information of the multi-modal data to input into a large language model to generate natural language output with emotional self-awareness; finally, integrates into a differentiable end-to-end framework and trains to improve the generalization ability and emotional understanding depth of the model. The large model emotion embodiment computing method, system, device and storage medium enable the large model to achieve deep understanding, dynamic tracking and interpretable generation of human emotions without the need for wearable devices.

A biometric-authenticated wearable system for remote monitoring, chronic disease management, elderly care, occupational therapy, and mental health rehabilitation is disclosed. The system uses fingerprint, facial, and voice recognition to securely associate physiological data with individual users. It monitors vital signs including heart rate, oxygen saturation, movement, and stress indicators, and transmits data to EMR / EHR systems in real time with privacy compliance. Machine learning modules assess disease risk, detect mobility decline, and perform sentiment-based analysis of speech and behavior. The system supports therapy compliance tracking across wellness programs and daily living tasks. The system may optionally employ large language models (LLMs) to enhance contextual understanding and sentiment interpretation from unstructured speech or text inputs. Real-time alerts are generated for health anomalies or non-compliance, enhancing clinical interventions. This integrated platform advances personalized and secure care through biometric authentication, predictive analytics, and seamless EMR / EHR integration.