A Decision Support Model for Assessment of Endodontic Treatment Outcome

Objective: Optimal dental treatment planning requires an accurate assessment of the outcome of any required endodontic treatment. Endodontic treatment outcome is a multifactorial event. In this work, we present a decision support model that describes the causal relationships among variables that come into play on the endodontic treatment outcome. Here, we describe a novel approach that uses the information from high-level clinical trials to help understand the mutual relationships among multiple variables.

Methods: We applied a data-driven Bayesian Network (BN) to analyze factors influencing endodontic treatment outcomes. Bayesian networks represent domain knowledge qualitatively by the use of graphical diagrams with nodes and arrows that represent variables and the relationships among the variables. Quantitatively, the degree of dependency among variables is expressed in probabilistic terms. Randomized controlled trials of nonsurgical endodontic treatment from January 1966 through August 2007 were chosen to be our data source. The total sample size in the included studies was 8783 cases. To construct the graph of the BN, the dependencies between the variables which lie in the data were learned using the Necessary Path Condition algorithm. The conditional probability distributions of the BN were estimated from the data using the Expectation-Maximization learning algorithm. To evaluate the accuracy of the BN model, stratified ten-fold cross-validation was employed. During each iteration, a BN was trained using the training data set and tested using the validating data set.

Results: Receiver operating characteristic curve analysis showed that the model was highly accurate in predicting the endodontic treatment outcome. The area under the curve (AUC) was 0.902. By implementing the model for predicting the treatment outcome of the patients at a local Hospital, the AUC was 0.883.

Conclusion: A BN decision support model can be constructed from clinical trials to successfully predict the endodontic treatment outcome.

Grant: NT-B-22-MS-14-50-04