Developing and testing a prediction model for periodontal disease using machine learning and big electronic dental record data

Application of machine learning in dentistry: insights, prospects and challenges

BACKGROUND

Machine learning (ML) is transforming dentistry by setting new standards for precision and efficiency in clinical practice, while driving improvements in care delivery and quality.

OBJECTIVES

This review: (1) states the necessity to develop ML in dentistry for the purpose of breaking the limitations of traditional dental technologies; (2) discusses the principles of ML-based models utilised in dental clinical practice and care; (3) outlines the application respects of ML in dentistry; and (4) highlights the prospects and challenges to be addressed.

DATA AND SOURCES

In this narrative review, a comprehensive search was conducted in PubMed/MEDLINE, Web of Science, ScienceDirect, and Institute of Electrical and Electronics Engineers (IEEE) Xplore databases.  Conclusions: Machine Learning has demonstrated significant potential in dentistry with its intelligently assistive function, promoting diagnostic efficiency, personalised treatment plans and related streamline workflows. However, challenges related to data privacy, security, interpretability, and ethical considerations were highly urgent to be addressed in the next review, with the objective of creating a backdrop for future research in this rapidly expanding arena.  Clinical significance: Development of ML brought transformative impact in the fields of dentistry, from diagnostic, personalised treatment plan to dental care workflows. Particularly, integrating ML-based models with diagnostic tools will significantly enhance the diagnostic efficiency and precision in dental surgeries and treatments.

Estimating Periodontal Stability Using Computer Vision

Periodontitis is a severe infection affecting oral and systemic health and is traditionally diagnosed through clinical probing-a process that is time-consuming, uncomfortable for patients, and subject to variability based on the operator's skill. We hypothesized that computer vision can be used to estimate periodontal stability from radiographs alone. At the tooth level, we used intraoral radiographs to detect and categorize individual teeth according to their periodontal stability and corresponding treatment needs: healthy (prevention), stable (maintenance), and unstable (active treatment). At the patient level, we assessed full-mouth series and classified patients as stable or unstable by the presence of at least 1 unstable tooth. Our 3-way tooth classification model achieved an area under the receiver operating characteristic curve of 0.71 for healthy teeth, 0.56 for stable, and 0.67 for unstable. The model achieved an F1 score of 0.45 for healthy teeth, 0.57 for stable, and 0.54 for unstable (recall, 0.70). Saliency maps generated by gradient-weighted class activation mapping primarily showed highly activated areas corresponding to clinically probed regions around teeth. Our binary patient classifier achieved an area under the receiver operating characteristic curve of 0.68 and an F1 score of 0.74 (recall, 0.70). Taken together, our results suggest that it is feasible to estimate periodontal stability, which traditionally requires clinical and radiographic examination, from radiographic signal alone using computer vision. Variations in model performance across different classes at the tooth level indicate the necessity of further refinement.

Classification of periodontitis stage and grade using natural language processing techniques

Periodontitis is a complex and microbiome-related inflammatory condition impacting dental supporting tissues. Emphasizing the potential of Clinical Decision Support Systems (CDSS), this study aims to facilitate early diagnosis of periodontitis by extracting patients' information collected as dental charts and notes. We developed a CDSS to predict the stage and grade of periodontitis using natural language processing (NLP) techniques including bidirectional encoder representation for transformers (BERT). We compared the performance of BERT with that of a baseline feature-engineered model. A secondary data analysis was conducted using 309 anonymized patient periodontal charts and corresponding clinician's notes obtained from the university periodontal clinic. After data preprocessing, we added a classification layer on top of the pre-trained BERT model to classify the clinical notes into their corresponding stage and grades. Then, we fine-tuned the pre-trained BERT model on 70% of our data. The performance of the model was evaluated on 32 unseen new patients' clinical notes. The results were compared with the output of a baseline feature-engineered algorithm coupled with MLP techniques to classify the stage and grade of periodontitis. Our proposed BERT model predicted the patients' stage and grade with 77% and 75% accuracy, respectively. MLP model showed that the accuracy of correct classification of stage and grade of the periodontitis on a set of 32 new unseen data was 59.4% and 62.5%, respectively. The BERT model could predict the periodontitis stage and grade on the same new dataset with higher accuracy (66% and 72%, respectively). The utilization of BERT in this context represents a groundbreaking application in dentistry, particularly in CDSS. Our BERT model outperformed baseline models, even with reduced information, promising efficient review of patient notes. This integration of advanced NLP techniques with CDSS frameworks holds potential for timely interventions, preventing complications and reducing healthcare costs.

Accuracy of Artificial Intelligence Models in the Prediction of Periodontitis: A Systematic Review

INTRODUCTION

Periodontitis is the main cause of tooth loss and is related to many systemic diseases. Artificial intelligence (AI) in periodontics has the potential to improve the accuracy of risk assessment and provide personalized treatment planning for patients with periodontitis. This systematic review aims to examine the actual evidence on the accuracy of various AI models in predicting periodontitis.

METHODS

Using a mix of MeSH keywords and free text words pooled by Boolean operators ('AND', 'OR'), a search strategy without a time frame setting was conducted on the following databases: Web of Science, ProQuest, PubMed, Scopus, and IEEE Explore. The QUADAS-2 risk of bias assessment was then performed.

RESULTS

From a total of 961 identified records screened, 8 articles were included for qualitative analysis: 4 studies showed an overall low risk of bias, 2 studies an unclear risk, and the remaining 2 studies a high risk. The most employed algorithms for periodontitis prediction were artificial neural networks, followed by support vector machines, decision trees, logistic regression, and random forest. The models showed good predictive performance for periodontitis according to different evaluation metrics, but the presented methods were heterogeneous.

CONCLUSIONS

AI algorithms may improve in the future the accuracy and reliability of periodontitis prediction. However, to date, most of the studies had a retrospective design and did not consider the most modern deep learning networks. Although the available evidence is limited by a lack of standardized data collection and protocols, the potential benefits of using AI in periodontics are significant and warrant further research and development in this area.

KNOWLEDGE TRANSFER STATEMENT

The use of AI in periodontics can lead to more accurate diagnosis and treatment planning, as well as improved patient education and engagement. Despite the current challenges and limitations of the available evidence, particularly the lack of standardized data collection and analysis protocols, the potential benefits of using AI in periodontics are significant and warrant further research and development in this area.

Applied artificial intelligence in dentistry: emerging data modalities and modeling approaches

Artificial intelligence (AI) is increasingly applied across all disciplines of medicine, including dentistry. Oral health research is experiencing a rapidly increasing use of machine learning (ML), the branch of AI that identifies inherent patterns in data similarly to how humans learn. In contemporary clinical dentistry, ML supports computer-aided diagnostics, risk stratification, individual risk prediction, and decision support to ultimately improve clinical oral health care efficiency, outcomes, and reduce disparities. Further, ML is progressively used in dental and oral health research, from basic and translational science to clinical investigations. With an ML perspective, this review provides a comprehensive overview of how dental medicine leverages AI for diagnostic, prognostic, and generative tasks. The spectrum of available data modalities in dentistry and their compatibility with various methods of applied AI are presented. Finally, current challenges and limitations as well as future possibilities and considerations for AI application in dental medicine are summarized.

Data-driven prediction model for periodontal disease based on correlational feature analysis and clinical validation

Objectives

This study aimed to investigate the performance and reliability of data-driven models employing correlational feature analysis and clinical validation for predicting periodontal disease.

Methods

The 7th Korea National Health and Nutrition Examination Survey (n = 10,654) was used for correlation analysis to identify significant risk factors for periodontitis. Periodontal prediction models were developed with the selected factors and database, followed by internal validation with 5-fold cross-validation and 1000 bootstrap resampling. External validation was conducted with clinical data (n = 120) collected through self-reported questionnaires, clinical periodontal parameters, and radiographic image analysis. Predictive performance was assessed for logistics regression, support vector machine, random forest, XGBoost, and neural network algorithms using the area under the receiver operating characteristic curves (AUC) and other performance metrics.

Results

Correlation analysis identified 16 features from over 1000 potential risk factors for periodontitis. The best data-driven model (XGBoost) showed AUC values of 0.823 and 0.796 for internal and external validations, respectively. Modeling with clinical data revealed those same measures to be 0.836 and 0.649, respectively. In addition, the data-driven model could predict other clinical periodontal parameters including severe bone loss (AUC = 0.813), gingival bleeding (AUC = 0.694), and tooth loss (AUC = 0.734). A patient case study about prognostic predictions revealed that the probability of periodontitis can be reduced by 6.0 % (stop smoking) and 0.6 % (stop drinking) on average.

Conclusions

Data-driven models for predicting periodontitis and other periodontal parameters were developed from 16 risk factors, demonstrating enhanced prediction performance and reproducibility in internal-external validations.

Multimodal feature fusion in deep learning for comprehensive dental condition classification

BACKGROUND

Dental health issues are on the rise, necessitating prompt and precise diagnosis. Automated dental condition classification can support this need.

OBJECTIVE

The study aims to evaluate the effectiveness of deep learning methods and multimodal feature fusion techniques in advancing the field of automated dental condition classification.

METHODS AND MATERIALS

A dataset of 11,653 clinically sourced images representing six prevalent dental conditions-caries, calculus, gingivitis, tooth discoloration, ulcers, and hypodontia-was utilized. Features were extracted using five Convolutional Neural Network (CNN) models, then fused into a matrix. Classification models were constructed using Support Vector Machines (SVM) and Naive Bayes classifiers. Evaluation metrics included accuracy, recall rate, precision, and Kappa index.

RESULTS

The SVM classifier integrated with feature fusion demonstrated superior performance with a Kappa index of 0.909 and accuracy of 0.925. This significantly surpassed individual CNN models such as EfficientNetB0, which achieved a Kappa of 0.814 and accuracy of 0.847.

CONCLUSIONS

The amalgamation of feature fusion with advanced machine learning algorithms can significantly bolster the precision and robustness of dental condition classification systems. Such a method presents a valuable tool for dental professionals, facilitating enhanced diagnostic accuracy and subsequently improved patient outcomes.

Automated Machine Learning to Develop Predictive Models of Metabolic Syndrome in Patients with Periodontal Disease

Metabolic syndrome is experiencing a concerning and escalating rise in prevalence today. The link between metabolic syndrome and periodontal disease is a highly relevant area of research. Some studies have suggested a bidirectional relationship between metabolic syndrome and periodontal disease, where one condition may exacerbate the other. Furthermore, the existence of periodontal disease among these individuals significantly impacts overall health management. This research focuses on the relationship between periodontal disease and metabolic syndrome, while also incorporating data on general health status and overall well-being. We aimed to develop advanced machine learning models that efficiently identify key predictors of metabolic syndrome, a significant emphasis being placed on thoroughly explaining the predictions generated by the models. We studied a group of 296 patients, hospitalized in SCJU Sibiu, aged between 45-79 years, of which 57% had metabolic syndrome. The patients underwent dental consultations and subsequently responded to a dedicated questionnaire, along with a standard EuroQol 5-Dimensions 5-Levels (EQ-5D-5L) questionnaire. The following data were recorded: DMFT (Decayed, Missing due to caries, and Filled Teeth), CPI (Community Periodontal Index), periodontal pockets depth, loss of epithelial insertion, bleeding after probing, frequency of tooth brushing, regular dental control, cardiovascular risk, carotid atherosclerosis, and EQ-5D-5L score. We used Automated Machine Learning (AutoML) frameworks to build predictive models in order to determine which of these risk factors exhibits the most robust association with metabolic syndrome. To gain confidence in the results provided by the machine learning models provided by the AutoML pipelines, we used SHapley Additive exPlanations (SHAP) values for the interpretability of these models, from a global and local perspective. The obtained results confirm that the severity of periodontal disease, high cardiovascular risk, and low EQ-5D-5L score have the greatest impact in the occurrence of metabolic syndrome.

Periodontitis and Metabolic Syndrome: Statistical and Machine Learning Analytics of a Nationwide Study

This study aimed to analyze the associations between periodontitis and metabolic syndrome (MetS) components and related conditions while controlling for sociodemographics, health behaviors, and caries levels among young and middle-aged adults. We analyzed data from the Dental, Oral, and Medical Epidemiological (DOME) record-based cross-sectional study that combines comprehensive sociodemographic, medical, and dental databases of a nationally representative sample of military personnel. The research consisted of 57,496 records of patients, and the prevalence of periodontitis was 9.79% (5630/57,496). The following parameters retained a significant positive association with subsequent periodontitis multivariate analysis (from the highest to the lowest OR (odds ratio)): brushing teeth (OR = 2.985 (2.739-3.257)), obstructive sleep apnea (OSA) (OR = 2.188 (1.545-3.105)), cariogenic diet consumption (OR = 1.652 (1.536-1.776)), non-alcoholic fatty liver disease (NAFLD) (OR = 1.483 (1.171-1.879)), smoking (OR = 1.176 (1.047-1.322)), and age (OR = 1.040 (1.035-1.046)). The following parameters retained a significant negative association (protective effect) with periodontitis in the multivariate analysis (from the highest to the lowest OR): the mean number of decayed teeth (OR = 0.980 (0.970-0.991)); North America as the birth country compared to native Israelis (OR = 0.775 (0.608-0.988)); urban non-Jewish (OR = 0.442 (0.280-0.698)); and urban Jewish (OR = 0.395 (0.251-0.620)) compared to the rural locality of residence. Feature importance analysis using the eXtreme Gradient Boosting (XGBoost) machine learning algorithm with periodontitis as the target variable ranked obesity, OSA, and NAFLD as the most important systemic conditions in the model. We identified a profile of the "patient vulnerable to periodontitis" characterized by older age, rural residency, smoking, brushing teeth, cariogenic diet, comorbidities of obesity, OSA and NAFLD, and fewer untreated decayed teeth. North American-born individuals had a lower prevalence of periodontitis than native Israelis. This study emphasizes the holistic view of the MetS cluster and explores less-investigated MetS-related conditions in the context of periodontitis. A comprehensive assessment of disease risk factors is crucial to target high-risk populations for periodontitis and MetS.

Unique miRomics Expression Profiles in Tannerella forsythia-Infected Mandibles during Periodontitis Using Machine Learning

T. forsythia is a subgingival periodontal bacterium constituting the subgingival pathogenic polymicrobial milieu during periodontitis (PD). miRNAs play a pivotal role in maintaining periodontal tissue homeostasis at the transcriptional, post-transcriptional, and epigenetic levels. The aim of this study was to characterize the global microRNAs (miRNA, miR) expression kinetics in 8- and 16-week-old T. forsythia-infected C57BL/6J mouse mandibles and to identify the miRNA bacterial biomarkers of disease process at specific time points. We examined the differential expression (DE) of miRNAs in mouse mandibles (n = 10) using high-throughput NanoString nCounter® miRNA expression panels, which provided significant advantages over specific candidate miRNA or pathway analyses. All the T. forsythia-infected mice at two specific time points showed bacterial colonization (100%) in the gingival surface, along with a significant increase in alveolar bone resorption (ABR) (p < 0.0001). We performed a NanoString analysis of specific miRNA signatures, miRNA target pathways, and gene network analysis. A total of 115 miRNAs were DE in the mandible tissue during 8 and 16 weeks The T. forsythia infection, compared with sham infection, and the majority (99) of DE miRNAs were downregulated. nCounter miRNA expression kinetics identified 67 downregulated miRNAs (e.g., miR-375, miR-200c, miR-200b, miR-34b-5p, miR-141) during an 8-week infection, whereas 16 upregulated miRNAs (e.g., miR-1902, miR-let-7c, miR-146a) and 32 downregulated miRNAs (e.g., miR-2135, miR-720, miR-376c) were identified during a 16-week infection. Two miRNAs, miR-375 and miR-200c, were highly downregulated with >twofold change during an 8-week infection. Six miRNAs in the 8-week infection (miR-200b, miR-141, miR-205, miR-423-3p, miR-141-3p, miR-34a-5p) and two miRNAs in the 16-week infection (miR-27a-3p, miR-15a-5p) that were downregulated have also been reported in the gingival tissue and saliva of periodontitis patients. This preclinical in vivo study identified T. forsythia-specific miRNAs (miR-let-7c, miR-210, miR-146a, miR-423-5p, miR-24, miR-218, miR-26b, miR-23a-3p) and these miRs have also been reported in the gingival tissues and saliva of periodontitis patients. Further, several DE miRNAs that are significantly upregulated (e.g., miR-101b, miR-218, miR-127, miR-24) are also associated with many systemic diseases such as atherosclerosis, Alzheimer's disease, rheumatoid arthritis, osteoarthritis, diabetes, obesity, and several cancers. In addition to DE analysis, we utilized the XGBoost (eXtreme Gradient boost) and Random Forest machine learning (ML) algorithms to assess the impact that the number of miRNA copies has on predicting whether a mouse is infected. XGBoost found that miR-339-5p was most predictive for mice infection at 16 weeks. miR-592-5p was most predictive for mice infection at 8 weeks and also when the 8-week and 16-week results were grouped together. Random Forest predicted miR-592 as most predictive at 8 weeks as well as the combined 8-week and 16-week results, but miR-423-5p was most predictive at 16 weeks. In conclusion, the expression levels of miR-375 and miR-200c family differed significantly during disease process, and these miRNAs establishes a link between T. forsythia and development of periodontitis genesis, offering new insights regarding the pathobiology of this bacterium.

Identifying predictors of tooth loss using a rule-based machine learning approach: A retrospective study at university-setting clinics

BACKGROUND

This study aimed to identify predictors associated with tooth loss in a large periodontitis patient cohort in the university setting using the machine learning approach.

METHODS

Information on periodontitis patients and 18 factors identified at the initial visit was extracted from electronic health records. A two-step machine learning pipeline was proposed to develop the tooth loss prediction model. The primary outcome is tooth loss count. The prediction model was built on significant factors (single or combination) selected by the RuleFit algorithm, and these factors were further adopted by the count regression model. Model performance was evaluated by root-mean-squared error (RMSE). Associations between predictors and tooth loss were also assessed by a classical statistical approach to validate the performance of the machine learning model.

RESULTS

In total, 7840 patients were included. The machine learning model predicting tooth loss count achieved RMSE of 2.71. Age, smoking, frequency of brushing, frequency of flossing, periodontal diagnosis, bleeding on probing percentage, number of missing teeth at baseline, and tooth mobility were associated with tooth loss in both machine learning and classical statistical models.

CONCLUSION

The two-step machine learning pipeline is feasible to predict tooth loss in periodontitis patients. Compared to classical statistical methods, this rule-based machine learning approach improves model explainability. However, the model's generalizability needs to be further validated by external datasets.