Artificial Intelligence–Assisted Evaluation of Periodontal Treatment Planning: A Pilot Clinical Investigation
- Dr. Haseena Begum Haleel Rahman , Department of Periodontics, JKKN Dental College and Hospital, Komarapalayam-638183, Tamil Nadu, India
- Dr. Dhivya Rajendran , Department of Periodontics, JKKN Dental College and Hospital, Komarapalayam-638183, Tamil Nadu, India.
- Dr. Sasikumar PK , Department of Periodontics, JKKN Dental College and Hospital, Komarapalayam-638183, Tamil Nadu, India
- Dr. Sakthi Saranya Devi K , Department of Oral medicine and Radiology, JKKN Dental College and Hospital, Komarapalayam-638183, Tamil Nadu, India
- Dr. Hariprasath Gopalakrishnan , Department of Periodontics, JKKN Dental College and Hospital, Komarapalayam-638183, Tamil Nadu, India
- Dr. Gowtham Raj Srinivasan , Department of Periodontics, JKKN Dental College and Hospital, Komarapalayam-638183, Tamil Nadu, India.
Article Information:
Abstract:
Background: Precise assessment of periodontal disease severity is critical for selecting appropriate treatment strategies and determining prognosis. Conventional evaluation relies on clinician- based interpretation of clinical findings and radiographs, which may introduce subjective variability. Artificial intelligence (AI) has the potential to enhance diagnostic consistency by analyzing complex clinical patterns through data-driven algorithms. Objective: To evaluate the feasibility and diagnostic performance of an AI-based deep learning model in assessing periodontal treatment parameters using combined clinical and photographic data. Materials and Methods: A pilot cross-sectional study was conducted among 15 patients aged 25–65 years presenting with varying severities of periodontal disease. Clinical parameters recorded included gingival recession, tooth mobility (Miller’s classification), Russell’s Periodontal Index, and clinical attachment level (CAL). Standardized intraoral photographs were obtained under uniform conditions. A convolutional neural network (CNN) utilizing transfer learning with a MobileNet backbone was developed. The dataset was divided into 80% training and 20% validation subsets. Model performance was assessed using accuracy, sensitivity, specificity, and confusion matrix evaluation. Results: The AI model demonstrated an overall classification accuracy of 90.8%. Sensitivity for identifying moderate to severe periodontal involvement was 98%. Misclassification primarily occurred between adjacent disease severity categories. Data augmentation techniques improved validation stability and reduced overfitting. Conclusion: AI-assisted evaluation showed promising diagnostic reliability in periodontal assessment. Although encouraging, these findings require validation through larger, multi-center studies before routine clinical implementation
Keywords:
Article :
INTRODUCTION:
Accurate evaluation of periodontal disease is essential for determining appropriate treatment strategies and predicting prognosis. Clinicians traditionally rely on clinical measurements such as attachment loss, gingival recession, probing depth, mobility, and radiographic findings. While these methods are well established, they depend largely on examiner expertise and clinical experience. As a result, some degree of inter-examiner variability is inevitable.
With advancements in computational sciences, artificial intelligence has emerged as a supportive tool in various medical and dental fields. Machine learning and deep learning algorithms are increasingly being applied for pattern recognition, image interpretation, and disease classification. In dentistry, AI applications have expanded into radiographic diagnostics, orthodontic planning, and detection of periodontal bone loss.
Considering the multifactorial nature of periodontal disease, integrating AI into treatment planning may help analyze multiple parameters simultaneously and provide consistent evaluations. The present study was undertaken to explore whether an AI-based model could assist in the assessment of periodontal treatment parameters in a clinical setting.
MATERIALS AND METHODS
Study Design and Ethical Approval
This pilot cross-sectional study was conducted after obtaining approval from the Institutional Ethical Committee. Written informed consent was obtained from all participants prior to data collection.
STUDY POPULATION
Fifteen patients between 25 and 65 years of age diagnosed with localized or generalized periodontal disease were included using purposive sampling. All participants underwent comprehensive clinical examination.
INCLUSION CRITERIA
Age 18 years and above Clinical evidence of periodontal disease Availability of clear intraoral photographic records Willingness to participate
EXCLUSION CRITERIA
Patients undergoing orthodontic treatment Individuals with systemic conditions known to significantly influence periodontal health History of periodontal surgery within the last six months More than eight missing teeth (excluding third molars)
CLINICAL ASSESSMENT
A full-mouth periodontal examination was performed for each participant. The following parameters were recorded:
Gingival recession: Measured from the cementoenamel junction to the free gingival margin using a UNC-15 periodontal probe
Tooth mobility: Graded according to Miller’s Mobility Index
Clinical attachment level: Determined from the cementoenamel junction to the base of the periodontal pocket
Russell’s Periodontal Index score
Standardized intraoral photographs were captured under consistent clinical conditions to ensure uniformity for AI analysis.
DEVELOPMENT OF THE AI MODEL
Clinical data and corresponding images were organized based on severity categories. The dataset was divided into training (80%) and validation (20%) groups.
A convolutional neural network (CNN) architecture incorporating transfer learning with a MobileNet backbone was utilized. To enhance the robustness of the model and reduce overfitting, data augmentation techniques such as rotation and scaling were applied.
EVALUATION OF MODEL PERFORMANCE
The predictions generated by the AI system were compared with clinical assessments made by experienced periodontists. Performance indicators included overall accuracy, sensitivity, specificity, and confusion matrix analysis
RESULTS
DEMOGRAPHIC DETAILS
A total of 15 patients were evaluated, with ages ranging from 25 to 65 years. Clinical parameters including gingival recession, mobility, CAL, and Russell’s Index were documented for all participants.
AI MODEL OUTCOMES
The AI model achieved an overall classification accuracy of 90.8%. Sensitivity in detecting moderate to severe periodontal involvement was 98%. Misclassification was limited and occurred mainly between closely related disease stages. The application of data augmentation improved validation performance and reduced variability in predictions
DISCUSSION
This pilot investigation evaluated the feasibility of implementing an AI-based deep learning model in periodontal treatment assessment. The results indicate that the proposed system achieved high diagnostic performance, particularly in identifying moderate to severe periodontal involvement, where timely intervention is critical.
The observed accuracy aligns with previously reported outcomes in AI- assisted periodontal diagnostics. Deep learning architectures have demonstrated substantial capability in analyzing dental images and predicting disease patterns.
By minimizing examiner-dependent
variability, AI-driven tools may provide structured and reproducible support during clinical decision- making. However, several limitations must be acknowledged. The sample size was small, and only internal validation was performed. Therefore, extrapolation of these findings to broader populations should be approached with caution. Additionally, the present model did not incorporate advanced imaging modalities such as CBCT or longitudinal treatment response data, which could further refine predictive performance.
Future research should focus on multi- center datasets, external validation protocols, and integration of radiographic as well as microbiological parameters. Such advancements may enable development of robust clinical decision support systems tailored to periodontal therapy.
CONCLUSION
Within the limitations of this pilot study, AI-assisted evaluation showed encouraging accuracy in assessing periodontal treatment parameters. Although the results are promising, larger-scale investigations are required before such systems can be routinely integrated into clinical practice. AI should be viewed as an adjunctive support tool that complements, rather than replaces, professional clinical judgment.
ACKNOWLEDGMENTS:
I would like to express my sincere gratitude to the staff members of the Department of Periodontics, J.K.K.N Dental College, for their unwavering support, valuable guidance, and encouragement throughout the course of my research. Their expertise and insightful suggestions have been fundamental to the successful completion of this article.
REFERENCES
1. Revilla-León M, Özcan M. Artificial intelligence in prosthetic and periodontal diagnosis: A systematic review. The Journal of Prosthetic Dentistry. 2022;128(6):1004-1013.
2. Lee JH, Kim DH, Jeong SN, Choi SH. Diagnosis and prediction of periodontally compromised teeth using a deep learning-based convolutional neural network algorithm. Journal of Periodontal and Implant Science. 2018;48(2):114-123.
3. Gulati M, Anand V, Anand B, et al. AI and periodontal health: Transforming disease prediction and diagnosis. Journal of Indian Society of Periodontology. 2023;27(1):12-19.
4. Bravaccini C, Aiuto R, Di Girolamo G, et al. Application of artificial intelligence in periodontology: A comprehensive review. Frontiers in Oral Health. 2022;3:867192.
5. Krois J, Ekert T, Meinhold L, et al. Deep learning for the radiographic detection of periodontal bone loss. Scientific Reports. 2019;9(1):8495.
6. Chang HJ, Lee SJ, Kim YJ, et al. Application of deep learning to predict periodontal disease based on clinical and radiographic features. International Journal of Environmental Research and Public Health. 2020;17(18):6945.
7. Schwendicke F, Samek W, Krois J. Artificial intelligence in dentistry: Chances and challenges. Journal of Dental Research. 2020;99(7):769-774.
8. Yu KH, Beam AL, Kohane IS. Artificial intelligence in healthcare. Nature Biomedical Engineering. 2018;2(10):719-731.
9. Lalla AM, Srivastava N, Ghosh S, et al. Artificial intelligence in periodontics: Current applications and future perspectives. Journal of Clinical Periodontology. 2023;50(9):1127-1142.
10. Parvez K, Ahmed T, Hossain MS, et al. Explainable AI frameworks for periodontal diagnosis and treatment planning: An evidence-based review. International Journal of Dental Sciences. 2024;22(1):45-58.
11. Goh V, Corbet EF, Leung WK. Machine learning in periodontal disease assessment: A systematic review. Journal of Periodontal Research. 2022;57(3):423- 437.
12. Schwarz F, Derks J, Saxer VP, et al. Clinical performance of AI-assisted clinical decision support systems in periodontal therapy: A randomized controlled trial. Journal of Dental Research. 2023;102(11):1198-1207.
13. Nguyen TT, Le H, Pham L, et al. Patient-specific response prediction to periodontal therapy using ensemble learning with clinical and microbiological parameters. Nature Digital Medicine. 2023;6(1):127