关键词: 龋齿, 微生物菌群, 预测模型, 预防, 个性化

Abstract: Objective To construct a microbial flora-based prediction model for childhood dental caries and subsequently establish a personalized caries prevention program based on this model. Methods Utilizing a multi-stage stratified random sampling method, 420 preschool children aged 3 to 6 years were recruited for this study. Oral examinations were performed by dental professionals to document the number of decayed, missing, and filled teeth, from which the DMFT (Decayed, Missing, and Filled Teeth) index was calculated. Concurrently, pertinent behavioral data were gathered via questionnaires. Saliva specimens were collected, and following anaerobic culture, principal microbial counts were quantified. The cohort was randomly allocated into a training set and a validation set at a 7∶3 ratio. Based on the DMFT index, the training set was bifurcated into a caries group (DMFT>0) and a caries-free group (DMFT=0). Demographic data and microbial profiles were compared between these two groups. Significant variables were identified and incorporated into a logistic regression model to ascertain the risk factors associated with childhood dental caries. The predictive accuracy and clinical utility of the model were assessed using Receiver Operating Characteristic (ROC) and decision curve analyses. Results A total of 402 preschool children were incorporated into the final analysis, revealing a caries prevalence of 53.06% (156/294). The multivariate logistic regression analysis indicated that the use of fluoride toothpaste (OR=0.348) and pre-sleep toothbrushing (OR=0.337) served as protective factors against the incidence of caries in this demographic. Conversely, frequent consumption of sweet foods (≥1 time/day; OR=2.260, 3.936), nocturnal eating habits (OR=3.016), and the presence of Streptococcus mutans (OR=2.118), Lactobacillus (OR=1.606), Bifidobacterium (OR=1.222), Scardovia wiggsiae (OR=5.666), and Candida albicans (OR=1.602) were identified as significant risk factors for the development of caries in preschool children (all P<0.05). The C-index for the nomogram model predicting caries in the training set was 0.909 (95% CI: 0.876-0.942), while the validation set yielded a C-index of 0.903 (95% CI: 0.853-0.954). The ROC curve analysis demonstrated that the area under the curve (AUC) for the predictive model, which integrated microbial data with other principal indicators, was 0.909. Furthermore, the decision curve analysis revealed that the predictive model offered a superior net benefit in forecasting the risk of caries compared to single-indicator models. Within a threshold probability range of 0.00 to 1.00, the net benefit remained positive, reaching a maximum of 0.531. Conclusions The predictive model for dental caries in preschool children, which integrates microbial flora with other key indicators, exhibits high predictive efficacy and substantial practical applicability. This model can serve as a valuable reference for the formulation of personalized preventive strategies.

Key words: Dental caries, Microbial flora, Predictive model, Prevention, Personalized