Automated artificial intelligence modeling for clinical decision-making of extraction versus non-extraction cases: Unleashing the potential for all

Objectives: To develop and evaluate artificial intelligence–based machine learning models for informed decision-making between extraction and non-extraction orthodontic treatment scenarios, recognizing the irreversible nature of extractions and the potential for adverse treatment outcomes if clinical decisions are made inaccurately. Material and Methods: Eight hundred patients (18–35 years old) with lateral cephalograms and study models were gathered from the department record room. The dataset was split into a training set and a test set to train the model by machine programming to evaluate the accuracy, recall, and precision performances of three distinct models toward decision-making of extraction versus non-extraction cases, respectively. Results: The success rates of the classifier system for the decision tree classifier were 87%, the random forest classifier was 97%, and the XGB classifier was 98%. The XGB classifier had the highest accuracy, recall, and precision scores, exhibiting outstanding reliability. Conclusion: The success rates of the classifier algorithms for the diagnosis of extraction versus non-extraction suggested that AI expert systems with ML algorithms could be a new approach in orthodontic treatment planning and clinical decision-making.