5 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.

According to the intelligent generation method for the healthy diet suggestions of the user, the category and the content of the nutrition ingested by the user are calculated, and if the change amplitude between the historical health state index and the current-day health state index of the user is larger than the preset amplitude threshold value, the healthy diet suggestions of the user are generated. If yes, calculating a difference proportion of corresponding health state data in the historical health state indexes and the health state indexes of the day, selecting target health state data, and adjusting diet data of the user; the category and content of food ingested by a user are obtained through automatic analysis, the category and content of nutrition ingested are calculated, and whether the health state data of the user is abnormal or not is judged according to the change amplitude between the historical health state index of the user and the health state index of the day. If yes, the diet data of the user is adjusted according to the abnormal data to obtain healthy diet data, so that the health state data of the user is tracked in real time, and healthy diet is accurately recommended to improve the diet health of the user.