Research On Diabetes Diet Recommendation System Based On Heterogeneous Information Network

The Diabetes Diet Recommendation System is a product of the development of artificial intelligence.It can effectively recommend diet for diabetic patients based on the recommendation of doctors and dietitians,combined with the dietary preferences of patients.The traditional diabetes diet recommendation method is based on the patient's recommendation for the food score record,but the actual distortion of the recommendation result due to the lack of score data sometimes occurs.Heterogeneous network is a multi-dimensional data network that can analyze multi-angle and deeplevel relationships between data.In theory,the use of heterogeneous networks in the diabetes diet recommendation system can make up for the shortcomings of traditional methods.Based on this,this paper studies the feasibility of heterogeneous network technology to achieve diabetes diet recommendation.By analyzing the process of doctors recommending diet for diabetic patients,define patient type P,patient symptom A,recommendation principle T,and recommendation scheme F as four data types of heterogeneous networks,and explain the selection of attributes included in the data type.At the same time,the potential relationship between data types was analyzed to construct a patient-centered diabetes diet heterogeneous network.The construction of the network serves as the theoretical basis for the implementation of dietary recommendations in this paper.Following establish a model of the recommendation system is based on the heterogeneous network of diabetes diet.In this paper,the model recommendation method is improved,and the idea of using heterogeneous network to recommend diet for diabetic patients is expounded.The model framework of diabetes diet recommendation system is constructed,and the recommended system model flow is explained through actual cases.The model uses traditional clustering and heterogeneous network sorting to achieve patient clustering,and then get a diet recommendation.This paper sets up cluster analysis experiments and comparative experiments.In cluster analysis,the effects of clustering are evaluated by relative entropy,compactness,multidimensional spatial verification,and clustering accuracy.The results show that when the iteration is carried out to the 14 th time,the clustering effect tends to be stable,and patient clustering is realized.In the comparative experiment,the recommended model is higher than the traditional recommendation method in the recommended program popularity,the program medical accuracy and the program diversity index.In summary,this model makes up for the shortcomings of the traditional recommendation method while improving the clustering accuracy.Using the LabVIEW cross-platform development function,a diabetes diet recommendation system including a patient module,a recommendation module,a database management module,and a medical terminal module was built,and the system implemented the functions of dietary recommendation and diet recording,thereby verifying the recommendation model proposed in this paper.Feasibility.