An Artificial Intelligence Tool for the Diagnosis of Facial Pain.

Background and objectives: Differentiation between temporomandibular disorders (TMDs) and trigeminal neuralgia (TN) as causes of orofacial pain is very important because the nature of these disorders and their treatments are vastly different. TMDs are usually treated with a rehabilitative approach, although dental correction or even surgery may be necessary in rare cases where the origin of the pain appears to be related to oral or temporomandibular joint pathology. By contrast, TN is largely treated with anticonvulsant medications, trigeminal nerve surgery, or trigeminal ablative procedures. TMDs are several orders of magnitude more common than TN, which may result in misdiagnosis and mistreatment if the proper diagnosis is not made initially.

Methods: We completed a study of 101 patients with either TMD or TN using supervised machine learning. A predictive model was developed using the 2 inputs of a questionnaire and directed physical examination.

Results: The network was trained to achieve the corresponding correct output, which was based on orofacial physical examination and expert diagnosis of each subject.

Conclusion: The analysis of this network indicated that TMDs and TN can be reliably differentiated using a standardized questionnaire and physical examination with approximately 90% accuracy.