The Effect of Hypertension on Periapical Disease - a Retrospective Study Utilizing Machine Learning to Assess a Dental School Population - Conclusion

Aug 4, 2026

This retrospective machine-learning study evaluates the association between hypertension and periapical disease using electronic dental record data from the Temple Kornberg School of Dentistry. Periapical disease was defined by multiple endodontic diagnoses, and data imbalance was addressed using SMOTE. Separate predictive models were trained in R with XGB, random forest, Lasso, and SVM, then compared by brier score, accuracy, and AUC ROC. XGB achieved the strongest overall performance and highlighted key predictors including age, sex, high blood pressure, and Medicaid insurance.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages29

Reading Time1 h 13 min

AccessFree to download

Abstract

Introduction

Materials and Methods

Data source and study design

Case definition and imbalance handling

Model training

Results

Performance comparison

Key predictive variables

Discussion

Conclusions

The Effect of Hypertension on Periapical Disease: a Retrospective Study Utilizing Machine Learning to Assess a Dental School Population | A Dissertation
Submitted to the Temple University Graduate Board | | --- | | In Partial Fulfillment of the Requirements for the Degree MASTERS OF SCIENCE | by Jeffrey K. Asano August 2024 Thesis Approvals: Maobin Yang, Department of Endodontology Chukwuebuka Ogwo, Department of Oral Health Sciences Jong Kim, Department of Endodontology ABSTRACT Introduction: Hypertension is a chronic medical condition in which the blood pressure is elevated. Apical periodontitis and hypertension are both chronic conditions triggered by inflammatory processes that share similar molecular players. Previous studies have shown a relationship between cardiovascular disease and apical periodontitis, but few have used machine learning algorithms to process the data. Machine learning algorithms use artificial neural networks to form pattern recognition pathways and are highly customizable to any data-driven task. The purpose of this study was to predict the association between hypertension and periapical disease from the Temple Kornberg School of Dentistry electronic dental record using machine learning algorithms. Materials and Methods: The integrated Axium data at the Temple Kornberg School of Dentistry was examined retrospectively after approval from the Temple University Institutional Review Board. The complete Health history data of patients who required primary endodontic therapy was collected, and periapical disease was defined as having symptomatic apical periodontitis, asymptomatic apical periodontitis, chronic apical abscess, or acute apical abscess. We identified large imbalances within the data, and so synthetic minority oversampling technique (SMOTE) was used for statistical analysis. After the application of SMOTE, XGB, random forest, Lasso, and SVM algorithms built separate models designed within R to predict periapical disease from the patients’relevant information. Results: The complete health history reports of 3888 patients who required primary endodontic therapy from January 2018 to December 2022 were collected. 1511 patients were diagnosed with some form of periapical disease. 610 Patients were diagnosed with hypertension. Among the four machine learning algorithms, XGB had the lowest brierscore, highest accuracy and highest AUC ROC values. The mean brier score, accuracy, and AUC ROC was 0.137, 83.73%, and 88.59% respectively. XGB found “age”,“sex”,“high blood pressure”, and “Insurance – Medicaid” to be the most significant variables able to predict periapical disease. Conclusion: While not statistically significant, hypertension was found to be one of the strongest health-related predictors of periapical disease. Surprisingly, we found nonhealth related variables such as insurance type, tobacco cessation, and insurance type to have a strong correlation with predicting periapical disease. With more patient data from Temple, it is possible to fine tune the algorithm to more accurately predict periapical disease based on each patient’s relevant health information. TABLE OF CONTENTS Page ABSTRACT……………………..………………………………………………...……...ii LISTS OF TABLES……………………………………………………………….………v LISTS OF FIGURES………………………………….………………………………….vi CHAPTER 1. INTRODUCTION……………………………………………...……………..1 2. METHODS…………………...………………………...……………………..4 3. RESULTS…………………………………………………………………......8 4. DISCUSSION………………………………………………………………..16 5. CONCLUSIONS…………………………………………………………….21 REFERENCES CITED………………………………………………………………….22 LIST OF TABLES Table Page 1. List of prognostic variables examined …………………………………..…….....5 2. List of tooth types examined……………………………………………………….....6 3. List of insurance types examined……………………………………………………..6 4. Assessment of variable importance………………………………………….13 LIST OF FIGURES Figure Page 1. The adaptive and innate immune response in apical periodontitis .

FAQs

How is periapical disease defined in the study?

Periapical disease was defined using endodontic diagnostic categories, including symptomatic and asymptomatic apical periodontitis, chronic apical abscess, and acute apical abscess.

Which machine learning model performed best?

Among the four algorithms, XGB showed the lowest brier score and the highest accuracy and AUC ROC values.

What variables were most important for predicting periapical disease?

The study reports that age, sex, high blood pressure, and Insurance – Medicaid were the most significant variables for prediction in the XGB model.

Aug 4, 2026

Upload by Evangeline