International Journal of Drug Delivery Technology
Volume 16, Issue 7s

Development and Validation of a Machine Learning Model for Early Prediction of Peri-Implantitis: Toward Personalized Risk Stratification in Implant Dentistry

Anindita Saha1, Silpiranjan Mishra2, Bikash Bishwadarshee Nayak3, Anupa Samanta4*

1Associate Professor, Department of Oral Medicine and Radiology, Burdwan Dental College and Hospital, West Bengal, India
2Professor, Department of Oral Medicine and Radiology, Kalinga Institute of Dental Sciences, Kalinga Institute of Industrial Technology (Deemed to be University), Bhubaneswar, Odisha, India
3Associate Professor, Department of Oral Medicine and Radiology, Kalinga Institute of Dental Sciences, Kalinga Institute of Industrial Technology (Deemed to be University), Bhubaneswar, Odisha, India
4*Tutor, Department of Oral Medicine and Radiology, Kalinga Institute of Dental Sciences, Kalinga Institute of Industrial Technology (Deemed to be University), Bhubaneswar, Odisha, India

(Corresponding Author)

ABSTRACT

Peri-implantitis is one of the most common complications after dental implant surgery, as approximately 20-47 percent of the recipients of the implants develop peri-implantitis complications. Machine learning (ML) would provide an opportunity to predict early and allow timely intervention and individual care.

Purpose: The purpose of the study was to create and test the predictive model of peri-implantitis based on clinical, demographical, and radiographic data of 40 patients during a 12-month follow-up period with the use of ML.

Methods and Materials: The retrospective cohort included 40 patients that were implant surgically operated. The extracted parameters were clinical (probing depth, bleeding on probing, plaque index, bone loss), patient demographics, systemic conditions, and radiographic features. When 5-fold cross-validation was used, several ML algorithms were trained using Logistic Regression, random forests, Support Vector Machine (SVM), gradient boosting, and Artificial Neural Network (ANN). The area under the receiver operating characteristic curve (AUC), sensitivity, specificity and accuracy have been used to evaluate model performance.

Findings: Gradient Boosting model resulted in the best AUC of 0.91 (95% CI: 0.84 -0.97), then the Random Forest (AUC = 0.88) and ANN (AUC = 0.86). Marginal bone loss at 6 months, smoking, probing pocket depth, and diabetes mellitus were the major predictive characteristics.

Conclusion: MLs, especially Gradient Boosting, are highly predictive of peri-implantitis and can be incorporated into the clinical decision-support systems to stratify risks individually in implant dentistry.

Keywords: Machine learning; Peri-implantitis; Dental implants; Predictive modeling; Risk stratification; Artificial intelligence

How to cite this article: Saha A, Mishra S, Nayak BB, Samanta A, Development and Validation of a Machine Learning Model for Early Prediction of Peri-Implantitis: Toward Personalized Risk Stratification in Implant Dentistry. Int J Drug Deliv Technol. 2026;16(7s): 1-000; DOI: 10.25258/ijddt.16.7s.1

Source of support: Nil

Conflict of interest: None