Artificial intelligence in predictive, personalized, and regenerative endodontics: A narrative review of emerging paradigms

AI in predictive prognosis

AI has emerged as a transformative tool in predictive prognosis and treatment planning within endodontics and caries management. ML and DL algorithms are capable of analyzing large volumes of clinical and radiographic data to identify subtle patterns that are often imperceptible to human clinicians. In endodontics, AI-assisted diagnostic systems have demonstrated superior performance in detecting periapical lesions and vertical root fractures on CBCT and periapical radiographs.^[10,11] According to a recent systematic review, the accuracy of AI for detecting vertical root fractures ranges from 73.6% to 96.6%, and the accuracy of AI applied over CBCT data is even higher in the case of untreated root canals. However, when it comes to detecting root fractures in instances of obturated root canals, AI applications significantly outperform traditional radiographs.^[12]

Failure to locate canals during endodontic treatment is one of the most common causes of endodontic treatment failure.^[13] It is estimated that 93% of all missed canals are located on the maxillary first molars, and 44% on the maxillary second molars.^[14,15] Convolutional neural networks (CNNs) can classify tooth and root canal morphologies with remarkable accuracy, thereby enhancing preoperative assessment and case selection.^[16] Albitar et al.^[17] Conducted a study to develop a deep learning model to automatically detect and segment unobturated mesial buccal 2 (MB2) canals on endodontically obturated maxillary molars depicted in CBCT studies. The scans were de-identified, and selected CBCT scans were imported into the image segmentation software ITK-SNAP. They concluded that the AI algorithm had the potential to identify both obturated and unobturated canals in endodontically treated teeth; however, it was still somewhat affected by metallic artifacts, variations in canal calcifications, and the applied configuration. In a recent narrative review, Ahmed et al. discussed the AI applications in the study of root and canal anatomy, and also highlighted that a significant challenge is the lack of larger datasets to provide more accurate deep learning models.^[18]

Predictive analytics in endodontics enable the estimation of treatment outcomes by integrating patient-specific data such as lesion size, systemic conditions, and procedural variables. For instance, models based on artificial neural networks (ANNs) have been shown to predict root canal treatment success rates and postoperative pain outcomes with high reliability. ANN is a system that imitates the human brain's structure and function, enabling the analysis of complex relationships between various predictors.^[19] Gao et al.^[20] Conducted a study to evaluate the accuracy of the back propagation (BP) artificial neural network model for predicting postoperative pain following root canal treatment, and it yielded an accuracy of 95.60% for the prediction of postoperative pain following RCT.

In caries management, AI-driven systems analyze intraoral images to quantify lesion activity and caries risk, facilitating minimally invasive and preventive care. AI-based Dental caries risk assessment tools can predict the chances of developing new carious lesions within a specified time frame, as well as the probability of changes in the size or activity of existing lesions.^[21] This approach not only helps prepare specific preventive measures for high-risk individuals but also identifies low-risk patients to avoid unnecessary restorations and surgical intervention. Presently, three AI-based techniques have been proposed for caries detection. These include image-based caries detection using AI, AI-assisted caries-risk assessment, and the Integration of AI with Computer-Aided Diagnosis (CAD) systems.^[22] In a recent systematic review and meta-analysis, Abbott et al.^[23] assessed AI platforms, machine learning methodologies, and associated accuracies used in detecting dental caries from clinical images and dental radiographs. Based on the evidence synthesized, they concluded that AI had superior sensitivity and equal specificity in detecting dental caries from clinical images compared to bitewing radiography. As datasets expand and algorithms evolve, predictive AI models are expected to become indispensable in risk-based endodontic and caries management strategies.

Lee et al.^[24] conducted a study to establish an effective artificial intelligence (AI)-based module for accurate tooth prognosis decisions, based on the Harvard School of Dental Medicine’s comprehensive treatment planning curriculum. This tooth prognosis AI scoring tool integrated parameters such as lesion size, root morphology, and systemic health to predict the likelihood of treatment success. They concluded that an AI-based machine-learning algorithm can be a helpful tool to determine tooth prognosis in consideration of the treatment plan. Similarly, in a study by Kazimierczak et al.^[25] the authors evaluated the diagnostic performance of Diagnocat (Diagnocat Ltd., San Francisco, CA, USA) for assessing endodontic treatment outcomes using panoramic radiographs Such predictive tools promote data-informed decision-making and enhance patient communication.

AI in personalized endodontic care

AI enables a paradigm shift from standardized to personalized endodontic therapy by leveraging patient-specific biological, radiographic, and procedural data. Personalized endodontic care involves tailoring treatment modalities to the unique pulpal and periapical microenvironment of each patient.^[26] Machine learning algorithms integrate variables such as tooth anatomy, systemic comorbidities, microbial diversity, and genetic markers to determine individualized treatment protocols.^[27] Genomics in dentistry is used to assess genetic predisposition to oral diseases. Genetic testing can identify patients who may be at higher risk for dental caries due to genetic factors that influence enamel strength or saliva composition.^[28] By integrating clinical databases with patient genomics and biomarker profiles, AI contributes to the emerging field of precision endodontics, aligning treatment decisions with individual biological responses.^[29] Such approaches enhance therapeutic predictability and patient satisfaction, supporting a more personalized and outcome-oriented model of dental care.

Furthermore, AI-assisted decision support systems enable risk stratification, allowing clinicians to take a patient-need-based approach to individualized patient care and prognoses.^[30] The integration of predictive analytics into electronic dental records could facilitate continuous learning models that refine prognostic accuracy over time. Predictive learning models also aid clinicians in determining the prognosis of retreatment or regenerative procedures for specific cases, thereby minimizing overtreatment or procedural errors. Furthermore, real-time AI-powered chairside systems can interpret intraoperative data—such as working length measurements or canal curvatures—to dynamically adjust treatment strategies in real-time.^[31] The integration of AI into clinical decision support systems (CDSS) is advancing precision treatment planning by suggesting appropriate instrumentation techniques, irrigants, and obturation protocols tailored to each case.^[32]

AI in regenerative endodontics

Regenerative endodontics seeks to restore the natural function and vitality of the pulp-dentin complex through biological regeneration rather than replacement. AI technologies are playing an increasingly crucial role in this domain by optimizing scaffold design, modeling cellular behavior, and selecting growth factors.^[33] Deep learning algorithms simulate stem cell proliferation, differentiation, and signaling responses under various biological conditions, enabling the prediction of the most favorable regenerative outcomes. Additionally, AI models can analyze gene expression data and predict the conditions for differentiating mesenchymal stem cells (MSCs) into odontoblast-like cells.^[34] AI-driven bioinformatics tools analyze gene expression data to identify key molecular pathways involved in the regeneration of dentin and pulp. Computational modeling aids in designing biomimetic scaffolds with optimal porosity, mechanical strength, and degradation profiles, thereby enhancing tissue regeneration.^[35]

Moreover, AI aids in digital multifactorial analysis to assess regenerative success, providing objective and reproducible measures of tissue healing.^[36] It integrates patient demographic data, medical and dental histories, radiographic and microbiological profiles, and treatment protocols. The integration of AI into regenerative endodontics represents a paradigm shift in how clinicians select cases, predict outcomes, and plan treatments. Currently, regenerative endodontic procedures are unpredictable due to the complex interplay of various biological, patient-specific, and procedural variables.^[36] AI has also shown promising results in automatically scoring periapical healing, root maturation, and dentinal wall thickening, key indices of regenerative success. These findings suggest that AI may become a crucial tool in standardizing prognostic assessments and reducing subjectivity within conventional methods. For clinicians, AI-informed predictions can improve case selection and treatment planning efficiency. The integration of AI with nanotechnology and 3D bioprinting further accelerates the development of intelligent biomaterials and patient-specific regenerative therapies.^[35] The convergence of AI with regenerative biology represents the next frontier in endodontic innovation, promising to redefine treatment success from mere symptom resolution to functional biological restoration.

Challenges & limitations

AI-based predictive learning models in endodontics rely heavily on the quality and diversity of training datasets. Most existing datasets are limited in size, region-specific, or lack standardized labeling, which restricts the generalizability of algorithms across diverse patient populations.^[18] Image-based AI tools trained on CBCT or radiographic data may misinterpret artifacts or variations in imaging protocols, leading to false predictions.^[37] Additionally, outcome prediction models often utilize retrospective data with incomplete clinical variables, which limits their ability to account for complex biological and procedural interactions that influence endodontic success.^[38] Integration of AI into chairside practice also requires sophisticated infrastructure, data management systems, and clinician training—factors that are often lacking in clinical settings.^[39]

Personalized endodontics demands the integration of multi-dimensional data—including radiographic, clinical, genomic, and microbiological inputs—into cohesive AI frameworks. However, such data fusion remains challenging due to inconsistent data formats and limited interoperability among dental software systems.^[40] Ethical concerns related to data sharing, patient consent, and cybersecurity further complicate the implementation of large-scale initiatives.^[41] Another major limitation is the “black box” nature of deep learning algorithms, which makes it difficult for clinicians to interpret how AI arrives at specific recommendations, thereby reducing clinical trust and accountability.^[42] Moreover, AI models may perpetuate diagnostic bias if trained on non-representative datasets, potentially compromising treatment equity across populations.^[43]

In regenerative endodontics, AI applications are still in a research phase. Challenges arise in translating insilico predictions of scaffold design, stem cell behavior, or molecular signaling into reproducible biological outcomes. Current computational models often simplify complex biological systems, limiting their predictive accuracy in real-world regenerative environments.^[7] Additionally, high costs of computational modeling, lack of standardized validation frameworks, and limited interdisciplinary collaboration between AI scientists and dental biologists hinder progress. Regulatory and ethical oversight for AI-guided tissue engineering remains underdeveloped, posing barriers to clinical translation.

Overall, while AI offers significant promise in enhancing predictive, personalized, and regenerative endodontics, its potential is constrained by issues of data quality, algorithm transparency, ethical governance, and translational validation. Overcoming these limitations will require standardized data-sharing protocols, explainable AI models, regulatory clarity, and multidisciplinary collaboration to ensure that AI augments rather than replaces clinical expertise in endodontic care. Table 2 summarizes the AI tools used in predictive, personalized, and regenerative endodontics.

Conclusion & future scope

AI is progressively redefining endodontic science by integrating data-driven precision with biologically grounded treatment approaches. Through predictive analytics, AI facilitates accurate diagnosis, outcome prediction, and risk assessment in both endodontic and caries management, enabling clinicians to adopt proactive and minimally invasive strategies. Its role in personalized endodontic care enhances decision-making by incorporating patient-specific variables such as pulpal status, systemic factors, and radiographic features, thereby advancing the concept of precision dentistry.^[3] In regenerative endodontics, AI contributes to optimizing biomaterial design, predicting cellular responses, and personalizing regenerative protocols, thereby accelerating progress toward the regeneration of a functional pulp-dentin complex.

Despite these advancements, several challenges remain before AI can achieve full clinical integration. Data heterogeneity, lack of standardized datasets, ethical considerations regarding patient privacy, and the need for regulatory validation hinder large-scale adoption. Additionally, AI algorithms must be trained on diverse populations to ensure clinical applicability and avoid bias. Bridging the gap between computational models and clinical translation will require interdisciplinary collaboration among dental clinicians, bioengineers, and data scientists.

The future of AI in endodontics lies in the development of integrated clinical decision support systems that can perform real-time analytics, image-guided therapy, and outcome prediction. The fusion of AI with digital workflows, augmented reality, and 3D bioprinting is likely to yield intelligent, patient-specific treatment paradigms. Moreover, advances in explainable AI (XAI) will enhance clinician trust and accountability by making algorithmic decisions transparent and interpretable.

In essence, AI-driven predictive, personalized, and regenerative endodontics represent not merely technological augmentation but a paradigm shift toward truly intelligent dental therapeutics. As evidence matures, AI will become an indispensable ally in achieving precision, predictability, and biological restoration in endodontic care.