Mining And Regulation Of Biomarkers Of Gut Microbiome In Obese People

In recent years,the number of obese people worldwide has increased rapidly.Obesity will increase the risk of diabetes,cardiovascular disease,cancer and other diseases.Studies have shown that intestinal microflora is one of the important factors affecting obesity.However,due to factors such as geography,diet,ethnicity,and analytical methods,studies in different regions have obtained inconsistent biomarkers of intestinal microflora in obese people,and may even present conflicting results.This topic first analyzed the host factors that affect intestinal microbiota and the impact of obesity on the species diversity of intestinal microbiota based on the big data of intestinal microbiota.Then,based on the idea of integration,mining comprehensive and accurate geographical shared biomarkers.In addition,in order to obtain further regulatory schemes for biomarkers,this topic conducts counterfactual reasoning based on deep reinforcement learning,which can provide personalized and accurate treatment guidance schemes for obese people.Finally,because the excavated biomarkers of intestinal microflora cannot be directly consumed,this topic has established a comprehensive dietary factor evaluation system based on machine learning,which can screen dietary factors that can regulate these biomarkers to help obese people lose weight.The main conclusions are as follows:(1)Based on the big data of gut microbiota,explore the relationship between obesity and intestinal microbiota.Research has shown that obesity factors can significantly change the composition of intestinal microbiota.In addition,geographical factors are the most important factors affecting intestinal microbiota.Therefore,when we explore the biomarkers of intestinal microflora in obese people,we first need to exclude the impact of geographical factors.In order to obtain comprehensive and accurate biomarkers,this topic adopts the idea of integration.To put it simply,first,obtain comprehensive and accurate biomarkers for each country by integrating filtered,wrapped,and embedded feature selection methods to avoid the bias introduced by a single feature selection method.Secondly,by adopting the strategy of obtaining intersections between different countries and eliminating the interference of regional factors,42 regional shared biomarkers were finally determined.The results showed that the obtained biomarkers performed better than a single method in distinguishing between obese and healthy people(AUC = 0.85).(2)Conduct counterfactual reasoning based on deep reinforcement learning to explore the regulatory direction of biomarkers.In order to obtain a further regulatory scheme for biomarkers,this topic conducts counterfactual reasoning based on deep reinforcement learning,enabling the original obesity sample to transform the obesity state determined by the model into a healthy state with minimal changes in intestinal microflora biomarkers,thereby providing personalized and accurate treatment guidance schemes for obese people.In addition,by comparing all obese people and corresponding counterfactual reasoning examples,we found that Akkermansia muciniphila,Faecalibacterium praussnitzii,Prevotella copri,Bacteroides dorei,Bacteroides eggerthia,Alistipes finegoldii,Alistipes shahii,Eubacterium sp.＿CAG＿180and Roseburia hominis may be potentially broad-spectrum targets with modulation consistency in the multi-regional obese population.In summary,this study provides a new direction for using intestinal microbiota to treat obesity and provides a new idea for using intestinal microbiota to intervene and treat other diseases.(3)Establish an evaluation system for dietary factors,and implement a method based on in vitro fermentation to evaluate the impact of dietary factors on the intestinal microbiota of obese people.The evaluation system is based on the idea of integration and the accuracy of machine learning model discrimination,which not only avoids the bias caused by a single method but also takes into account the advantages and disadvantages of the model itself.The results show that most dietary factors fermented in vitro can make the intestinal microbiota of obese people more favorable to healthy people.Among them,ginsenoside,carnosol,artemisinin,paclitaxel,and oligofructose have better effects.