| Objective:This study aims to explore the changing rules and inflection points of aging in middle and old aged female by measuring their aging degree,determine whether state age are "significantly older" or not,trace the multiple factors of aging and anti-aging— dietary type and structure,lifestyle,health care products,beauty products,etc.Then construct multiple anti/postpone aging programs,and achieve optimal result by comparing and verifying multiple machine learning models and experimental evidence.Methods:Multi stage cluster stratified sampling epidemiological survey was used to collect the PPSHAS aging measurement results and the information on diet type structure,behavior lifestyle,as well as the use of drugs,health products and beauty products of middle and old aged female in China;Relevant statistical methods was used to explore the characteristics and laws of aging distribution,LOESS smoothing function method and ROC curve were used to explore the age of aging inflection point,and binary LR model was fitted for verification.Through single factor analysis initially screened the multidimensional factors affecting state age,further filter variables with filters and embedded methods,fitted LR,DT and RF model respectively,and verified model performance with accuracy,recall,specificity,precision,F1 score,AUC and other indicators;The optimal anti/postpone aging program was constructed and validated based on the selected model.Aging markers such as content of IGF-1,content and activity of β-GAL and SOD in serum were detected to support the research.Results:(1)A total of 1720 valid questionnaires were obtained,with an effective rate of96.36%.The aging measurement indicates that the aging standardized score of middle and old aged female in China increases with the increase in chronological age.The trend of changes in physical,psychological,social,and overall dimensions is more obvious after transforming into status age among populations with different chronological ages;Female of different age groups both have aging inflection points:middle aged in 48.0～53.2 years old,young elderly in 66.3～70.0 years old,elderly in78.4～81.2 years old;Aging measurements shows that 14.24% of subjects are significantly younger than chronological age,45.58% are consistent,and 40.17% are significantly older.(2)The results of single factor analysis of anti-aging factors indicated that there were 55 significant variables,including region,26 items of dietary type and structure,15 items of lifestyle,5 items of drugs and health products,and 8 items of beauty products(P<0.01).After filtering feature selection,13 variables with a correlation degree of more than 0.7 were excluded.LASSO regularization screened 24 variables including regions.The diet types and structures include 7 items,such as leafy vegetables,eggplants,water consumption;the life behaviors include 9 items,such as eating out,wearing attention,hobby duration,family emotion,and static behavior duration;the drugs,health care products and beauty products category include 7 items,such as endocrine regulation,the use of skin care products and lotion usage.(3)Through model performance comparison and verification,the RF model is optimal,with accuracy rates of 82.64% and 80.62%,AUC of 0.894 and 0.892,accuracy rates of 82.80% and 89.78%,recall rates of 89.78% and 88.85%,F1 scores of 0.768 and 0.744,and Kappa index higher than the other two models in the test set and validation set.Laboratory testing of biological markers related to aging showed that SOD content and activity,IGF-1 content decrease with aging,while β-GAL activity increased(P<0.05),which to support the research results in a certain extent.(4)In the best anti/postpone aging program,the priority of variables from high to low is skin care product use,hobby duration,sleep time,lotion,eye cream,leafy vegetables,eggplant fruits,carbohydrates,water consumption,spices(importance23.05～66.68);There are regional differences in the factors related to anti/postpone aging among the study subjects between the south and north China.Conclusion:The aging scores of middle and old aged females show certain regular changes in chronological age and status age,and there are aging inflection points between different age groups.Among all the subjects,14.24% of them are significantly younger than chronological age,while 40.17% are significantly older;The optimized RF model constructs a three-dimensional anti/postpone program based on dietary structure types,lifestyle,and beauty products.LR and DT fitting results also provide valuable information on anti/postpone aging. |
PDF Full Text Request