Study On The Relationship Between Dietary Inflammation Index And Muscular Attenuation Disease And Application Of Risk Prediction Model

Objectives: The Dietary Inflammation Index(DII)of the elderly in Urumqi,Xinjiang was calculated,the relationship between dietary inflammation index and muscular attenuation disease was analyzed,and the risk prediction model was established to investigate the factors affecting muscular attenuation disease,in order to provide basis and guidance for the prevention of muscular attenuation disease.Methods: A one-to-one survey was conducted on the elderly who underwent physical examination in Urumqi Cihui Health Management Center,and the data were collected according to the sodium standard.A total of 1391 elderly patients over 60 years old in Urumqi from January 2018 to January 2020 were selected as the research objects.General information,sleep information,and daily physical activity information of the respondents were collected.Semi-quantitative Food Frequency questionnaire(SFFQ)was used to collect dietary intake of the respondents.Various foods were converted into nutrients,and DII scores were calculated.The DII scores were divided into three groups according to the tripartite number.T1 group represented the most anti-inflammatory group,T2 group represented the intermediate group,and T3 group represented the most pro-inflammatory group.Logistic regression was used to analyze the relationship between different DII groups and sarcopenia,and restricted cubic splines were used to visualize the risk of DII and sarcopenia.Lasso regression was used to screen variables of influencing factors of sarcopenia,and the influencing factors of screening were incorporated into random forest,XG Boost,Ada Boost,Gaussian Naive Bayes and neural network to construct the risk prediction model.By analyzing the AUC value,Yoden index and accuracy of the test set,the model effect was evaluated comprehensively.Results: 1.A total of 1391 subjects were included in this study,including 194(13.9%)cases of muscular attenuation.In general,there were statistical differences in age,education background,monthly income,marital status,living style,working status and alcohol consumption between muscular attenuated disease and non-muscular attenuated disease(P<0.05).In terms of sleep status,there were statistical differences in subjective sleep quality,sleep time,sleep time,sleep efficiency and daytime dysfunction(P<0.05).In terms of physical activity,there were statistical differences among daily work physical activity,daily traffic physical activity,daily household physical activity,daily sports physical activity and total physical activity amount(P<0.05).In terms of clinical indexes,urea nitrogen,creatinine,uric acid,hemoglobin,diastolic blood pressure had statistical differences(P<0.05).The dietary inflammatory index of non-muscular attenuated patients was(-0.14±1.58),indicating an anti-inflammatory diet tendency;the dietary inflammatory index of those with muscular attenuated patients was(0.84±1.61),indicating a pro-inflammatory diet tendency.2.Differences in grip strength,walking speed,Muscle mass at each site and Skeletal Muscle mass Index(SMI)are statistically significant between different DII groups(P<0.05);The prevalence rates in T1,T2 and T3 groups were 7.6%,12.5% and 21.7%,respectively,and the differences were statistically significant(P<0.05).3.Logistic regression showed that high DII was a risk factor for muscular attenuation(OR=1.909,95%CI=1.201～3.033).Restricted cubic spline results showed that DII and myoattenia showed an approximate "J"-shaped curve relationship,indicating that the risk of myoattenia increased with the increase of dietary inflammatory index score.4.Lasso regression was used to screen variables for myoattenia,and the factors influencing the screening were included in the risk prediction model.After observing the AUC area of each model,it was found that the random forest model had the highest AUC area,and there was a statistical difference between the random forest model and the other four models after comparing the AUC area(P<0.05).By observing other indexes of the model,it can be seen that the XG Boost model and Ada Boost model have the highest values in terms of specificity,the Ada Boost model has the highest value in the negative predictive value,the Ada Boost model has the highest value in the F1 score,and the random forest model has the highest value in the other indexes.According to AUC and other indicators,the prediction model of muscular decay disease risk built by random forest algorithm is better than other models.Conclusions: High DII was associated with an increased risk of developing sarcopenia,and the risk of developing sarcopenia increased as DII scores increased.Through the risk prediction model,it is found that these influencing factors,except dietary inflammatory factors,tend to be muscle strength or activity level,indicating that exercise in addition to diet is also essential for the elderly,and both of these can be relatively easy to intervene in real life.In order to better prevent and treat sarcopenia,appropriate exercise programs accompanied by anti-inflammatory diet intervention can greatly improve the quality of life of elderly people with sarcopenia in later life.