Can oral microbiome predict low birth weight infant delivery?

Area of Science:

• Microbiology
• Genomics
• Machine Learning

Background:

• The oral microbiome plays a role in pregnancy outcomes.
• Identifying specific oral microbiota associated with low birth weight (LBW) is crucial for early intervention.

Purpose of the Study:

• To identify oral microbiota factors linked to LBW in Chinese pregnant women.
• To develop a machine learning-based prediction model for LBW using oral microbiome data.

Main Methods:

• A nested case-control study involved 580 pregnant women (23 LBW cases, 23 controls).
• Saliva samples were analyzed using 16S rRNA gene sequencing to profile the oral microbiome.
• Machine learning algorithms were employed to build a predictive model based on identified microbial biomarkers.

Main Results:

• Over-representation of Streptococcus and under-representation of Saccharibacteria_TM7 were observed in the LBW group.
• Ten oral microbial species, including Streptococcus and opportunistic pathogens, were identified as LBW biomarkers.
• The machine learning model achieved 82% accuracy, 91% sensitivity, and 73% specificity in predicting LBW (AUC-ROC 0.89).

Conclusions:

• A machine learning model utilizing oral microbiome data shows significant potential for predicting LBW.
• Specific oral bacteria, even in the absence of overt oral disease, can serve as indicators of adverse pregnancy outcomes.