EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Explainable Generative Artificial Intelligence for Clinical Decision Support in Prosthodontics

Pranavanand Satyamurthy, Shruthishree S. H., Ravi Kumar Jalli, Venkata Vaisali Vavilala, A. Swathi, Mutyala Suresh

Computer and Decision Making: An International Journal · 02.11.2026 · 3 (6S) · 3455-3472

DOI: 10.1922/comdem.v3i6s.19087 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 02.11.2026. Verfasst von Pranavanand Satyamurthy und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

While current generative AI solutions can now draft crown geometries, recommend implant placements and answer prosthodontic queries, they leave it largely up to the clinician accepting such recommendations to justify themselves about their rationale for doing so. The current study seeks to uncover the published evidence available and define an explainability standard for generative decision support in prosthodontics. We performed structured critical review of 75 peer-reviewed works published between 2020 and 2026 on prosthodontic and implant AI, generative algorithms and large language models (LLMs), as well as on explainability, decision-support implementation and evidence levels of all included works. Peak output was recorded in 2023 (n = 24). Reviews (33.3%) and offline validation (29.3%) were predominant in the included corpus, whereas only six studies (8.0%) presented clinical, external or randomized evidence. Generative design software sped up design process and aligned better with technician morphology however, only one paper validated its generated crowns through use in patients, and no studies on explainability reached clinical validation level. LLMs answered prosthodontic queries inconsistently while hallucination was the leading error type in implant placement recommendation. Based on our findings we construct TRACE explanation framework which includes traceable evidence, region attribution, alternative counterfactual designs, calibrated uncertainty and expert editing history, with non-learned constraint guard and updating by clinician-in-the-loop bounding it at its boundaries. All components of TRACE are matched to the six prosthodontic decision stages with acceptance criteria specified, while explanation should necessarily encompass not only generated designs and recommendations but also pixel information provided by generative algorithms. Staged testing in light of DECIDE-AI and TRIPOD+AI standards presents a realistic path towards accountability of the chairside practice.

Zitieren

APA

Satyamurthy, P., H., S. S., Jalli, R. K., Vavilala, V. V., Swathi, A., & Suresh, M. (2026). Explainable Generative Artificial Intelligence for Clinical Decision Support in Prosthodontics. Computer and Decision Making: An International Journal, 3(6S), 3455-3472. https://doi.org/10.1922/comdem.v3i6s.19087

Vancouver

Satyamurthy P, H. SS, Jalli RK, Vavilala VV, Swathi A, Suresh M. Explainable Generative Artificial Intelligence for Clinical Decision Support in Prosthodontics. Computer and Decision Making: An International Journal. 2026;3(6S):3455-3472. doi:10.1922/comdem.v3i6s.19087

BibTeX
@article{Satyamurthy2026,
  author = {Satyamurthy, Pranavanand and H., Shruthishree S. and Jalli, Ravi Kumar and Vavilala, Venkata Vaisali and Swathi, A. and Suresh, Mutyala},
  title = {Explainable Generative Artificial Intelligence for Clinical Decision Support in Prosthodontics},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {3455--3472},
  doi = {10.1922/comdem.v3i6s.19087}
}
Kostenlos starten →