IADR Abstract Archives
Artificial Intelligence Assisted Decision Making in Dentistry: What’s Coming Next
Objectives: Nowadays, practically every single area of knowledge is searching in the application of artificial intelligence (AI) in their processes, and dentistry is not an exception. The use of AI in dentistry has rapidly increased in the last few years, specially with the advent of Machine Learning, assisting in the radiographic detection of caries, diagnosis and prognosis of periodontally compromised teeth, automatic cephalometric analyses, among others. Thus, the objective of this study was to discuss the perspectives of Artificial Intelligence applications in dentistry, and to assess the capabilities of AI process to predict whether tooth extractions will be needed or not during an orthodontic treatment, in order to assist the orthodontist and general practitioner in the decision-making process.
Methods: A sample of 149 orthodontically treated patients were included in this study, feeding 62 different variables from the initial cephalometric analysis and the clinical decision whether tooth extractions were performed or not during treatment. A prediction system was developed using the Machine Learning software Weka 3.8.3, based on the information of 100 of those 149 patients. The data of the remaining 49 was used as a test set.
Results: The predictive system developed achieved a 75.51% of correct predictions and generated a Receiver Operating Characteristics (ROC) curves with an Area Under the Curve (AUC) of 0.843, achieving good accuracy.
Conclusions: This machine learning-based model might assist orthodontists in the decision making of tooth extractions and general practitioners when asked by their patients.
Methods: A sample of 149 orthodontically treated patients were included in this study, feeding 62 different variables from the initial cephalometric analysis and the clinical decision whether tooth extractions were performed or not during treatment. A prediction system was developed using the Machine Learning software Weka 3.8.3, based on the information of 100 of those 149 patients. The data of the remaining 49 was used as a test set.
Results: The predictive system developed achieved a 75.51% of correct predictions and generated a Receiver Operating Characteristics (ROC) curves with an Area Under the Curve (AUC) of 0.843, achieving good accuracy.
Conclusions: This machine learning-based model might assist orthodontists in the decision making of tooth extractions and general practitioners when asked by their patients.