4 results about "Wellness state" patented technology

The invention discloses an employee health dynamic intervention strategy method based on deep reinforcement learning, and the method comprises the steps: building a health state classification standard through collecting the historical health data and intervention records of an employee, and predicting the future health state of the employee and the implementation benefit of an intervention strategy through a multi-layer perceptron (MLP) model. And then, a deep Q network (DQN) and a heuristic exploration strategy are introduced to construct an employee health dynamic intervention strategy model, and the target of dynamically selecting an optimal intervention strategy according to the current health state of the employee is achieved. After model training is completed, the trained model is used for selecting an optimal intervention strategy for each employee, and the effectiveness and feasibility of the optimal intervention strategy are verified through a simulation experiment or practical application. The system has the advantages that personalized, automatic and dynamic employee health management can be realized, and a comprehensive and scientific health management scheme is provided for enterprises.

Provided herein are methods and systems for generating a recommendation, comprising receiving a first data of a first wellness type comprising a physical type, a social type, or a mental type; receiving a second data of the first wellness type; receiving a wellness state of a second wellness type comprising the physical type, the social type, or the mental type, wherein the second wellness type is different than the first wellness type; determining a change in the wellness state of the second wellness type based on a change between the first data of the first wellness type and the second data of the first wellness type; and generating, based on the determining, the recommendation.

This invention discloses a personalized intelligent nutritional diet analysis and recommendation system and method, relating to the field of data processing technology. The system includes: a multi-source data acquisition module, a nutrition modeling module, a diet plan generation module, a feedback acquisition module, and an iterative optimization module. Through the multi-source data acquisition module, this invention captures comprehensive user information, including not only static data and dynamic physiological indicators, but also daily dietary records, scenario-based needs, dietary preferences, and dietary budgets. Based on this comprehensive data, the nutrition modeling module constructs a user-specific nutritional needs model and identifies deficiencies in various nutrients through a precise nutritional gap calculation formula. Combining user dietary preferences and cooking scenario conditions, a dynamic diet plan adaptation formula generates a daily personalized diet list and alternative solution library, meeting the diverse needs of users under different health states and environmental conditions.

The invention discloses a diet intervention recommendation method based on reinforcement learning, and the method comprises the following steps: collecting diet behavior data and health state data of a user, and constructing a behavior track matrix and a health feature set; constructing a compliance fatigue function based on the behavior track matrix, and generating a compliance fatigue score; constructing a shielding mapping vector, and determining a diet intervention action set based on a preset threshold rule; inputting the health feature set into an improved Dueling DQN model, and generating a state value and an action advantage value; performing weight correction on each action in the diet intervention action set; fusing the state values with the corrected action advantage values, constructing a state action value function Q value set of the diet intervention actions, and generating recommendation score distribution of the diet intervention actions; and selecting the diet intervention action with the highest score as a recommendation result. According to the invention, the behavior fatigue and health state change of the user can be dynamically perceived, and the individual adaptability and behavior compliance of diet recommendation are improved.