| Background:Caries is one of the most common chronic infectious diseases in children,which is dominated by bacteria.It causes irreversible damage to the hard tissues of the teeth.If no intervention is taken,it can cause pain and even tooth loss,thus affecting these children’s physical and mental health.Studies have shown that changes in the microbial community in dental plaque are closely related to the development of caries.Therefore,this study investigated the differences in the structure of the bacterial community at different caries stages through a metagenomic method and constructed a caries diagnostic model,which provides the etiological basis and methodological support for the diagnosis and treatment of caries.Objective:We analyzed the differences in microbial structure and species diversity at different caries stages,investigated bacteria composition and the trend of dominant plaque microbiome,and then constructed a diagnosis model by machine learning methods,which promotes rapid chair-side diagnosis and monitoring.Methods:Fifteen healthy primary molars and fifteen primary molars with deep caries were included in the study.The healthy plaque was only collected from healthy primary molars as healthy samples(Confident Health,CH).Healthy buccal surfaces(Relative Health,RH),enamel caries(EC),and dentin caries(DC)were collected from deep caries.A total of 60 samples were collected in this study.Firstly,the change of species diversity and structure was investigated by 2b RAD sequencing for Microbiome(2b RAD-M),and the influence of dmfs index(the number of decayed,missing,and filled teeth-surfaces in deciduous dentition),gender,age and tooth position on species diversity was also investigated.Secondly,the composition and trend of bacterial communities were studied at the genus and species levels.Finally,based on the above results,a caries diagnosis model was constructed by machine learning,which diagnoses the different stages of caries.Results:(1)The Chao1 index was greater in the CH than in the other groups(P < 0.05).(2)The effect sizes of age,sex,and tooth position on species diversity were smaller than that of the dmfs index on species diversity.The larger the dmfs index,the lower the species diversity(P < 0.05).(3)In the comparison of the intra-group distance of the four samples,the intra-group distance of CH was significantly higher than that of the others(RH,EC,and DC).(4)The distance between CH and DC is the largest in the between-group distance comparison.In the comparison of the distance between the caries groups(RH,EC,and DC),the distance between RH and EC was the smallest,whereas the distances between RH/EC and DC were large.(5)With the progression of caries,the relative abundance of Propionibacterium_acidifaciens,Streptococcus_mutans,Lactobacillus_paragasseri,and Scardovia_wiggsiae gradually increased.With the progression of caries,the relative abundance of Corynebacterium_matruchotii,Peptidiphage_sp000466165,Neisseria_mucosa,Rothia_aeria,Archnia_propionica,Corynebacterium_durum,Streptococcus_sanguinis,and Neisseria_sicca_B gradually decreased.(6)In this study,we innovatively used the joint multi-factor method to construct the caries diagnostic model.Its accuracy rate was 92%,and AUC is more than 96%,indicating good diagnostic performance.Conclusions:(1)The highest microbial species diversity was found in healthy primary molars.(2)The dmfs index was a vital host factor influencing plaque microbial richness.(3)The community structure of plaque of a healthy primary tooth is stable and diversified.(4)Once caries occurs,the microbial structure will change.Compared to the microbial structure of a healthy state,the microbial structure of relative health is more similar to that of enamel caries.When caries progresses to the dentin layer,the structure of the microbial community changes dramatically.Dentin caries is different from confident/relative health or enamel caries.(5)Some "beneficial bacteria" can maintain the stability and dynamic balance of the flora structure.In contrast,the presence of some "caries bacteria" can lead to the rapid progression of caries.(6)The caries diagnostic model constructed by combining multiple factors has higher accuracy and better diagnostic performance than previous models that relied on "biomarkers". |