Predicting oral malodour based on the microbiota in saliva samples using a deep learning approach

Area of Science:

• Microbiology
• Bioinformatics
• Artificial Intelligence

Background:

• Oral malodour is primarily caused by volatile sulfur compounds from bacterial activity.
• Predicting oral malodour based on specific bacterial species is challenging.
• Salivary microbiota analysis offers a potential avenue for malodour prediction.

Purpose of the Study:

• To develop and validate a deep learning approach for predicting oral malodour.
• To analyze the relationship between salivary microbiota composition and oral malodour.
• To assess the efficacy of deep learning in classifying oral malodour from microbial data.

Main Methods:

• 16S rRNA gene sequencing was performed on saliva samples from 90 subjects.
• Operational taxonomic units (OTUs) and their relative abundances were profiled.
• A deep learning classifier was trained to distinguish between oral malodour and healthy breath.

Main Results:

• The deep learning model achieved a 97% predictive accuracy for oral malodour.
• This significantly outperformed a support vector machine model, which achieved 79% accuracy.
• The model effectively classified subjects based on salivary microbial profiles.

Conclusions:

• Deep learning provides a highly effective method for predicting oral malodour from salivary microbiota.
• This approach can be utilized for pre-clinic screening of oral malodour.
• The findings highlight the potential of AI in diagnosing halitosis.