Second Opinion for Non-Surgical Root Canal Treatment Prognosis Using Machine Learning Models

Area of Science:

• Dentistry
• Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Non-surgical root canal treatment (NSRCT) failure prediction relies heavily on clinical experience, which can be prone to errors.
• Current methods for predicting NSRCT outcomes are underdeveloped, necessitating more objective approaches.
• Understanding risk factors associated with NSRCT failure is crucial for improving treatment success rates.

Purpose of the Study:

• To investigate the potential of machine learning (ML) models as a decision support tool for predicting non-surgical root canal treatment (NSRCT) outcomes.
• To evaluate whether ML models can serve as a reliable second opinion to aid dentists in treatment prognosis.
• To assess the accuracy and sensitivity of ML models in predicting the success or failure of NSRCT.

Main Methods:

• A retrospective study involving 119 cases of untreated Apical Periodontitis undergoing NSRCT by a single specialist.
• Data collection using a novel template, defining treatment success (lesion clearance) or failure as the binary outcome.
• Four ML algorithms—Logistic Regression (LR), Random Forest (RF), Naive-Bayes (NB), and K Nearest Neighbors (KNN)—were trained and tested using selected variables.

Main Results:

• The Random Forest (RF) and K Nearest Neighbors (KNN) algorithms demonstrated a statistically significant improvement (p < 0.05) in the sensitivity and accuracy of treatment prognosis.
• Analysis identified key variables associated with NSRCT outcomes, serving as inputs for the ML models.
• The study provides a proof of concept for the utility of ML in enhancing prognostic capabilities for NSRCT.

Conclusions:

• Machine learning models show promise in improving the accuracy and sensitivity of non-surgical root canal treatment prognosis.
• ML models can function as a valuable second opinion tool, supporting clinical decision-making for dentists.
• Further research through randomized clinical trials is warranted to validate the clinical utility of ML in NSRCT prognosis.