Research
& Evidence

Why it matters

Torabinejad and colleagues’ 2026 journal pre-proof compared ChatGPT-5.2 and Gemini 3.0 with expert classifications in 645 teeth requiring initial root canal treatment. Individual-endpoint accuracy was 74–88%, but agreement across all three clinical endpoints was 61.1%/59.4%, and across four including case difficulty 47.3%/46.2%. These are distinct agreement measures, not a universal diagnostic error rate or patient-harm estimate. Expert variability and single-pass consumer-interface testing limit interpretation. Unlike simulated board questions, this evaluated retrospective clinical records; neither establishes safe autonomous diagnosis or improved outcomes. This editorial summary and its translations await clinical review.

Why it matters

This retrospective comparison involved 376 teeth and used CBCT radiographic status as the reference. AI tooth-level sensitivity was 67.3% versus clinicians’ 49.3%, but specificity was 82.3% versus 92.5%. Higher sensitivity came with more false alarms. A persistent shadow may still be healing; neither an AI flag nor this imaging reference proves treatment failure. No improved healing or reduction in unnecessary retreatment was demonstrated.

Why it matters

Four chatbots answered questions about four hypothetical trauma scenarios. Specialist-framed questions received better ratings, but prompt wording and clinical content changed together; wording alone cannot be credited. Unsafe advice occurred, especially with patient-style prompts. This was not a trial in patients. Do not let chatbot answers delay urgent dental-trauma assessment.

Why it matters

Sabeti and Torabi’s 2026 journal pre-proof scoping review examined six prognostic models. None had external validation or formal calibration assessment, and none directly compared retreatment with microsurgery for the same clinical scenario. Internal test results are not personal success probabilities. The proposed decision framework is exploratory, not a validated tool for choosing treatment; clinical feasibility and patient preferences remain essential.

Why it matters

This single-center retrospective study tested middle mesial canal detection in 28 held-out CBCT cases: 27 were correctly classified, with one positive case missed. The reference was expert image consensus, not confirmation during treatment. All scans came from one institution and device; there was no external validation. This does not guarantee canal detection, prove better treatment success or justify routine CBCT.

Why it matters

Swanson and colleagues’ JADA perspective (2026;157(9):957–967) discusses bringing emerging technology, including potential AI applications, into care. The question is not simply whether a tool performs well in a study, but whether it improves outcomes that matter to patients. This is not an AI diagnostic-accuracy trial or proof of superior endodontic outcomes. Independent validation and clinical judgment remain important; this summary does not claim our practice uses diagnostic AI.

Why it matters

Turgeon and colleagues (JADA 2026;157(9):993–998) retrospectively studied intraoral radiographs taken in a dental-student setting. Overall retakes were 16.9% with rectangular versus 16.7% with circular collimation (P=.922): no statistically significant difference. Their background cites a 50–60% dose reduction from prior literature; this study did not measure that reduction. These findings concern intraoral X-rays, not CBCT, and do not establish that our practice uses rectangular collimation. Training and appropriate image selection still matter.

Why it matters

An independent international scoping review (India + Brazil) of 22 studies, published in the field's flagship journal in May 2026. Found that while GentleWave outperforms plain syringe irrigation, it shows no significant difference versus ultrasonic or sonic activation — the standard adjuncts used by specialist endodontists — in smear layer removal, biofilm reduction, or removal of obturation materials. The few available clinical studies found no difference in postoperative pain or early healing. The authors' own conclusion: 'the current body of evidence does not support a clear conclusion that GentleWave is superior to other irrigation techniques.' This paper underpins our published position on /root-canal-disinfection-technology and our decision not to charge a premium for GentleWave-branded treatment.

Why it matters

The largest systematic review ever conducted on apicoectomy outcomes — 21 studies, more than 1,600 cases, four decades of literature in five languages. Established the now-standard finding that modern endodontic microsurgery succeeds in 94% of cases (95% CI 0.89–0.98) versus 59% for traditional surgical techniques (95% CI 0.55–0.63). The difference is statistically significant at p < 0.0005. This paper underpins the success rate quoted on our /apicoectomy page.

Why it matters

Companion to Part 1. Isolated the contribution of the surgical operating microscope itself by comparing surgeons who used the same modern instruments and materials but only loupes versus those using a true microscope. Found 88% success without the microscope, 94% with — and the difference was statistically significant for molars specifically (p = 0.011). Cited on every page where we describe the role of magnification.

Why it matters

Critical review comparing the actual evidence base for implants versus endodontic treatment. Documented that 63% of implant outcome studies do not disclose funding, 66% have meaningful bias risk, and that industry-sponsored trials report lower failure rates than independent trials. Also distinguished implant 'survival' (still in the mouth) from 'success' (functional, asymptomatic, no peri-implantitis). 20-year implant data: 89.5% survival but only 75.8% true success. The basis for our recommendation that restorable teeth should generally be saved.

Why it matters

The foundational protocol paper for endodontic microsurgery. Describes the step-by-step technique that endodontic residents have learned worldwide for two decades — the small osteotomy, the shallow root-end resection, the high-magnification inspection, the ultrasonic root-end preparation, and the biocompatible filling. Every surgical case in our practice follows this protocol. The reference for our 'how the procedure actually works' patient education content.