Artificial intelligence in dentistry: from second reading of bite-wing radiographs to dental practice management.

Abstract

Artificial intelligence is moving in dentistry from experimental applications to practical tools for diagnosis, patient communication and practice management. The most mature field is the analysis of imaging data, including bite-wing, periapical and panoramic radiographs and cone-beam computed tomography. Human reading of bite-wing radiographs, particularly for early approximal enamel caries, is associated with substantial under-detection; in the ADEPT study dentists without artificial intelligence support detected 44.3% of enamel-only proximal lesions, compared with 75.8% with such support. The 31.5 percentage-point difference is a more accurate interpretation of the often mentioned "one third" figure than a general statement about a 30% error rate in dental radiography. Safe implementation requires the use of artificial intelligence as a second reader, clinical review of false-positive findings and retention of clinician responsibility. Additional value lies in explaining treatment alternatives, documenting informed consent, providing 24/7 reception services, multilingual communication, automated scheduling, postoperative follow-up, ambient documentation and support for coding and reimbursement.