Research On The Relationship Between Chronic Diseases And Food Based On SSNE

In today's society,with the improvement of living standards,chronic diseases have begun to appear in people's lives.Studies have shown that choosing the right food can slow or even control the deterioration of the disease.The traditional method of determining the relationship between disease and food requires experts to formulate nutrition rules for each disease,and then based on nutrition rules to reason about the relationship between disease and food.Due to the complexity of the types of diseases,such methods are labor-intensive.To solve this problem,the paper provides a new research idea,which is to construct a simple relationship graph between chronic diseases and food,and then use the network vectorization model to mine potential relationship,and finally use the generated vectors to determine the relationship.Based on this idea,this paper did the following work:(1)First,chronic disease,food,and the relationship data between them are captured from web pages,then data cleaning and data alignment are performed,and a chronic disease-food relationship diagram is created.The diagram has 32 chronic diseases and 334 foods.(2)For the high-order nonlinear,sparse,and sign diversity of chronic disease-food relationship diagram,this paper proposes semi-supervised signed network embedding(SSNE)based on an auto-encoder using second-order proximity and signed proximity.Then the paper uses the node vectors generated by this model to conduct relationship mining and relationship determination,and based on this,a simple food recommendation system is implemented.(3)According to the edge segmentation,the chronic disease-food relationship diagram was divided into training set and test set.The training set was originally input to the LINE model,Deep Walk model,SDNE model and the SSNE model proposed in this paper.The network recovery ability was tested with the training set.Use test sets to test network link prediction ability.By comparing the experimental results,it is found that the SSNE model is 10% higher than the best result of the unsigned model on the premise of guaranteeing the network recovery ability.The results show that the SSNE model proposed in this paper has better network generalization.Ability,which is more suitable for relationship mining.In summary,this paper constructed a simple chronic diseases-food relationship diagram put forward SSNE model,and based on this mine potential relationship.Finally,the validity of the SSNE model was verified on the data set of chronic disease and food relationships.