Exploring AI-Driven Machine Learning Approaches for Optimal Classification of Peri-Implantitis Based on Oral...

Area of Science:

• Computational biology
• Microbiome research
• Artificial intelligence in healthcare

Background:

• Machine learning (ML) shows promise for diagnosing microbiome-related diseases.
• Peri-implantitis (PI) diagnosis can be improved using ML on metagenomic data.

Purpose of the Study:

• To benchmark ML algorithms for predicting peri-implantitis (PI) using oral microbiome data.
• To compare the performance of ML models based on saliva and biofilm microbiomes.

Main Methods:

• Applied two AI-driven auto-ML approaches to 100 metagenomes from saliva and subgingival biofilm.
• Generated 100 ML models for PI prediction and compared them to single-microorganism models.

Main Results:

• ML algorithms outperformed single-microorganism models in predicting PI.
• Auto-ML models achieved high performance (80-100% AUC, sensitivity, specificity).
• Saliva microbiome models demonstrated higher predictive accuracy than biofilm models.

Conclusions:

• ML, particularly using combinations of microorganisms, offers a viable approach for PI diagnosis.
• This study validates ML for developing practical, cost-effective dental diagnostic platforms.