-
Email [email protected]
-
Address 848 N. Rainbow Blvd. #5486 Las Vegas, NV 89107, USA
Centro Escolar University, Faculty of Dentistry, Manila, Philippines.
*Corresponding author: Omid Panahi
Centro Escolar University, Faculty of Dentistry, Manila, Philippines.
Email ID: [email protected]
Received: Apr 12, 2025
Accepted: May 20, 2025
Published Online: May 27, 2025
Journal: Annals of Cardiology
Copyright: Panahi O © All rights are reserved
Citation: Panahi O. The oral-systemic link: Using AI to analyze dental data for cardiovascular risk assessment. Ann Cardiol. 2025; 2(1): 1008.
The established association between oral health and cardiovascular disease highlights the potential of dental data to provide valuable insights into cardiovascular risk. This study investigates the application of Artificial Intelligence (AI) to analyze comprehensive dental records for the prediction of cardiovascular risk. By leveraging machine learning algorithms on a rich dataset encompassing periodontal parameters, caries history, radiographic findings, and treatment records, we aim to identify patterns and features indicative of elevated cardiovascular risk. The performance of various AI models will be evaluated against traditional cardiovascular risk scores. This research seeks to demonstrate the utility of readily available dental data, when analyzed through AI, as a non-invasive and potentially early indicator of cardiovascular risk, thereby facilitating timely preventive strategies and improving patient outcomes.
Keywords: Artificial intelligence; Machine learning; Dentistry; Cardiovascular risk assessment; Oral-systemic link; Dental data; Risk prediction; Preventive cardiology; Early detection.
The intricate and increasingly recognized bidirectional relationship between oral health and systemic diseases has spurred significant interest in leveraging information from the oral cavity to gain insights into overall health. Among these systemic connections, the association between oral health, particularly periodontal disease, and Cardiovascular Disease (CVD) stands out as a well documented and actively researched area. Epidemiological studies consistently demonstrate a higher prevalence and incidence of CVD, including conditions like coronary artery disease, stroke, and peripheral artery disease, among individuals with poor oral health, independent of traditional cardiovascular risk factors. This “oral-systemic link” suggests that the oral cavity, with its unique microbial environment and inflammatory processes, can serve as a window into an individual’s systemic health and potentially contribute to or reflect underlying cardiovascular risk.
Traditional cardiovascular risk assessment primarily relies on established risk factors such as age, sex, blood pressure, cholesterol levels, smoking status, and family history. While these models have proven valuable in identifying individuals at risk, they may not fully capture the complex interplay of factors contributing to CVD development. Furthermore, these assessments often require specific medical examinations and laboratory tests, which may not be routinely performed on all individuals. In contrast, dental examinations are a relatively common and often more frequent healthcare encounter for a significant portion of the population. The wealth of data collected during routine dental visits, encompassing periodontal health, caries prevalence, radiographic findings, and treatment history, represents a potentially untapped resource for cardiovascular risk stratification.
The challenge lies in effectively extracting and interpreting the subtle patterns within this complex dental data that may be indicative of underlying cardiovascular risk. Traditional statistical methods may struggle to discern these intricate relationships within the high-dimensional and often non-linear nature of dental datasets. This is where the transformative power of Artificial Intelligence (AI), particularly Machine Learning (ML), emerges as a promising solution. AI algorithms [1-23] possess the capability to analyze vast amounts of data, identify complex patterns, and build predictive models that can uncover previously unrecognized associations between dental features and cardiovascular risk.
By training AI models on comprehensive dental records linked with cardiovascular outcomes or established cardiovascular risk scores, we can potentially develop novel, non-invasive tools for cardiovascular risk assessment. For instance, AI could learn to identify specific patterns in periodontal parameters (e.g., probing depth, attachment loss, bleeding on probing), radiographic features (e.g., alveolar bone loss), or the presence of certain oral conditions that are strongly correlated with an increased likelihood of future cardiovascular events. This approach could offer several advantages over traditional methods. Firstly, it leverages data that is routinely collected during dental visits, potentially allowing for opportunistic risk screening without requiring additional medical tests. Secondly, AI’s ability to analyze complex interactions between multiple dental variables may uncover risk factors that are not captured by conventional risk assessment models. Finally, it could potentially facilitate earlier identification of at-risk individuals, enabling timely implementation of preventive strategies and lifestyle modifications to mitigate the development of CVD.
This investigation into using AI [24-42] to analyze dental data for cardiovascular risk assessment is driven by the hypothesis that valuable information about an individual’s cardiovascular risk profile is embedded within their dental records. By harnessing the pattern recognition capabilities of AI, we aim to demonstrate the feasibility of developing a non-invasive and readily accessible tool for cardiovascular risk stratification, ultimately contributing to improved early detection, personalized prevention, and better overall patient outcomes. The subsequent sections will delve into the specific types of dental data that can be utilized, the AI methodologies that are most relevant for this task, and the potential clinical implications and challenges associated with this innovative approach to cardiovascular risk assessment.
The biological plausibility of the oral-systemic link, particularly the connection between periodontal disease and CVD, provides a strong foundation for exploring the utility of dental data in cardiovascular risk assessment. Periodontal disease is characterized by chronic inflammation and the presence of pathogenic bacteria in the oral cavity. These bacteria and their byproducts can enter the systemic circulation, triggering or exacerbating systemic inflammation, endothelial dysfunction, and the development of atherosclerosis – the underlying pathology of many cardiovascular events. Furthermore, shared risk factors such as smoking, diabetes, and poor dietary habits contribute to both periodontal disease and CVD, highlighting the interconnectedness of these conditions.
Given this biological rationale, dental data offers a rich source of information that may reflect an individual’s systemic inflammatory burden, microbial profile, and exposure to shared risk factors, all of which are relevant to cardiovascular risk. Specific dental parameters that hold potential for cardiovascular risk prediction include:
• Periodontal parameters: Probing depth, clinical attachment loss, bleeding on probing, gingival recession, furcation involvement, and the number of missing teeth due to periodontal disease provide insights into the severity and extent of periodontal inflammation and tissue destruction.
• Caries history: The presence, severity, and history of dental caries may reflect dietary habits and oral hygiene practices, which are also linked to cardiovascular health.
• Radiographic findings: Panoramic and periapical radiographs can reveal alveolar bone loss, a hallmark of periodontal disease, as well as other calcifications or abnormalities that might be associated with systemic conditions.
• Oral microbiome data: Advances in microbial sequencing technologies allow for the identification and quantification of specific oral bacteria, some of which have been implicated in systemic inflammation and CVD. While not routinely collected, future integration of such data holds significant promise.
• Treatment history: The frequency and type of dental treatments received may indirectly reflect the burden of oral disease and patient compliance with oral hygiene recommendations.
• Artificial intelligence: With its ability to handle high-dimensional and complex datasets, offers a powerful tool to analyze these diverse dental parameters and identify intricate patterns that may be predictive of cardiovascular risk. Various machine learning algorithms can be employed, including:
• Supervised learning: Algorithms [43-52] like logistic regression, support vector machines, random forests, and gradient boosting can be trained on dental data labeled with known cardiovascular outcomes or risk scores to predict an individual’s risk level.
• Deep learning: Neural networks, particularly Convolutional Neural Networks (CNNs) for image analysis of dental radiographs and Recurrent Neural Networks (RNNs) for sequential data analysis of treatment history, can automatically learn complex features from the raw data without explicit feature engineering.
• Unsupervised learning: Clustering algorithms can be used to identify distinct subgroups of patients based on their dental profiles, which may correlate with varying levels of cardiovascular risk.
The development of AI models for cardiovascular risk assessment using dental data holds the potential to transform preventive cardiology. By leveraging the routine collection of dental information, we can envision a future where opportunistic risk screening becomes integrated into dental practice. Dentists, equipped with AI-powered tools, could potentially identify patients who may be at elevated cardiovascular risk and refer them for further medical evaluation and preventive interventions. This proactive approach could lead to earlier detection of CVD, improved patient outcomes, and a more integrated model of healthcare delivery that bridges the gap between oral and systemic health. The subsequent sections will explore the methodological approaches for building and evaluating such AI models [53-67], along with the ethical and practical considerations for their clinical implementation.
While the prospect of leveraging AI to analyze dental data for cardiovascular risk assessment holds significant promise, several challenges must be addressed to ensure its successful development, validation, and clinical implementation. These challenges span data-related issues, methodological considerations, and practical and ethical implications.
Data heterogeneity and standardization in dental records
One of the primary hurdles lies in the inherent heterogeneity and lack of standardization in dental records. Data collection practices, terminology, and the level of detail documented can vary significantly across different dental practices, Electronic Health Record (EHR) systems, and even individual practitioners. Periodontal charting, radiographic interpretations, and narrative notes may lack consistent coding and formatting, making it difficult to aggregate and analyze large-scale datasets effectively. Standardizing dental data collection and promoting the adoption of common data models and terminologies are crucial prerequisites for building robust AI models.
Data quality and completeness
The quality and completeness of dental data can also pose challenges. Missing data points, inaccuracies in recording measurements (e.g., probing depths), and inconsistent documentation of oral conditions can introduce noise and bias into the datasets used for AI training. Ensuring data accuracy through standardized training and quality control measures, as well as developing strategies for handling missing data, will be essential for building reliable predictive models.
Linking dental data with cardiovascular outcomes
A critical challenge involves establishing robust linkages between dental data and cardiovascular outcomes or established cardiovascular risk scores. This often requires access to integrated datasets that combine dental EHRs with medical records, which may be subject to privacy regulations and data sharing limitations. Secure and privacy-preserving methods for data linkage are necessary to create the datasets required for training and validating AI models [68-78] for cardiovascular risk prediction.
Feature engineering and selection
Identifying the most relevant dental features that are predictive of cardiovascular risk requires careful feature engineering and selection. While AI algorithms can automatically learn features from raw data, domain expertise from both dentistry and cardiology is crucial in guiding this process and ensuring that the selected features have biological plausibility and clinical relevance. Determining the optimal combination of dental parameters, radiographic features, and other relevant information (e.g., treatment history) to maximize predictive accuracy is an ongoing challenge.
Model development and validation
Developing robust and generalizable AI models requires careful consideration of the choice of algorithms, model architecture, and training methodologies. Overfitting, where the model performs well on the training data but poorly on unseen data, is a significant concern. Rigorous internal and external validation on independent datasets is essential to assess the model’s performance, calibration, and clinical utility across different populations and settings.
Addressing confounding factors and bias
The relationship between oral health and cardiovascular disease is complex and influenced by various confounding factors, such as smoking, diabetes, socioeconomic status, and access to healthcare. AI models must be carefully designed to account for these confounders and avoid introducing or amplifying existing biases present in the training data. Ensuring fairness and equity in risk prediction across different demographic groups is a critical ethical consideration.
Interpretability and clinical trust
Many advanced AI models, particularly deep learning algorithms, operate as “black boxes,” making it difficult to understand the reasoning behind their predictions. This lack of interpretability can hinder clinical trust and adoption. Developing explainable AI (XAI) techniques that can provide insights into the dental features driving the risk predictions is crucial for facilitating clinical acceptance and integration into dental practice.
Integration into dental workflows
Successfully integrating AI-powered cardiovascular risk assessment tools into existing dental workflows presents practical challenges. Dentists need user-friendly interfaces and clear guidance on how to interpret and act upon the AI-generated risk assessments. Seamless integration with dental EHR systems and minimal disruption to routine clinical practice are essential for adoption.
Ethical and legal considerations
The use of AI for cardiovascular risk assessment in the dental setting raises ethical and legal considerations related to the scope of dental practice, the responsibility for communicating risk information to patients, and the potential for anxiety or over-referral. Clear guidelines and protocols need to be established to address these issues and ensure responsible implementation.
Regulatory and approval pathways
The regulatory landscape for AI-powered [79-90] diagnostic tools in dentistry is still evolving. Determining the appropriate regulatory pathways and ensuring compliance with relevant guidelines will be necessary for the widespread clinical use of AI for cardiovascular risk assessment based on dental data.
Building upon the current research and addressing the existing challenges, future works in the field of utilizing AI to analyze dental data for cardiovascular risk assessment can focus on several key areas to further refine its accuracy, clinical utility, and integration into healthcare systems.
Enhanced data integration and harmonization
Future research should prioritize the development of standardized data models and ontologies specifically designed for integrating dental and cardiovascular data. This includes creating common data elements for periodontal parameters, radiographic findings, and other relevant dental information that can be seamlessly linked with cardiovascular risk factors and outcomes from medical records. Exploring the use of Fast Healthcare Interoperability Resources (FHIR) standards for data exchange could facilitate this integration.
Development of more sophisticated and interpretable AI models
Future efforts should focus on developing more advanced AI models, including deep learning architectures, capable of capturing complex, non-linear relationships within the dental data that may be indicative of cardiovascular risk. Simultaneously, a strong emphasis should be placed on enhancing the interpretability of these models using XAI techniques. This could involve visualizing the dental features that most strongly influence risk predictions, providing clinicians with a better understanding of the AI’s reasoning [91-103] and fostering greater trust in its output.
Integration of multi-omics and lifestyle data
Future research can explore the integration of other relevant data layers to enhance the predictive power of AI models. This includes incorporating oral microbiome data (e.g., through metagenomic sequencing), genetic predispositions related to both oral and cardiovascular health, and lifestyle factors (e.g., diet, physical activity, smoking habits) captured through questionnaires or wearable devices. Analyzing these multi-omics and lifestyle data in conjunction with dental information could provide a more holistic view of an individual’s risk profile.
Longitudinal studies and predictive modeling of cardiovascular events
Current research often focuses on predicting existing cardiovascular risk scores. Future studies should aim to leverage longitudinal dental data to directly predict future cardiovascular events (e.g., myocardial infarction, stroke). This requires the development of time-series AI models that can analyze changes in dental parameters over time and identify patterns that precede cardiovascular events.
Development of AI-powered risk stratification tools for dental professionals
Future work should focus on creating user-friendly AI-powered tools that can be seamlessly integrated into dental practice management software and clinical workflows. These tools could analyze routinely collected dental data in real-time and provide dentists with an immediate assessment of a patient’s potential cardiovascular risk, prompting appropriate referrals or discussions about preventive measures.
Personalized risk assessment and intervention strategies
AI models could be further developed to provide personalized cardiovascular risk assessments based on an individual’s unique dental profile and other risk factors. This could lead to tailored preventive strategies, such as specific oral hygiene recommendations or lifestyle modifications, that address both oral and cardiovascular health simultaneously.
Validation in diverse populations and settings
To ensure the generalizability and clinical utility of AI models, future research must prioritize validation in diverse patient populations, considering variations in age, sex, race, ethnicity, socioeconomic status, and geographic location. Studies conducted in different dental care settings (e.g., private practices, community clinics, academic institutions) are also crucial.
Exploration of the impact of dental interventions on cardiovascular risk
Future research could investigate the potential impact of dental interventions (e.g., periodontal therapy, management of oral infections) on reducing cardiovascular risk, using AI to analyze longitudinal data before and after treatment. This could provide valuable evidence for the role of dental care in overall cardiovascular health management.
Addressing ethical and societal implications
Ongoing research is needed to address the ethical and societal implications of using AI for cardiovascular risk assessment in dentistry. This includes developing guidelines for responsible data sharing, ensuring patient privacy and data security, mitigating algorithmic bias, and establishing clear protocols for communicating risk information to patients and coordinating care with medical professionals.
Integration with telehealth and remote monitoring
Future works can explore the integration of AI-powered dental risk assessment with telehealth platforms and remote monitoring devices. This could enable convenient and accessible risk screening for individuals in underserved areas or those with limited access to traditional healthcare settings.
The exploration into leveraging artificial intelligence to analyze dental data for cardiovascular risk assessment reveals a promising avenue for enhancing preventive cardiology. By harnessing the pattern recognition capabilities of AI on routinely collected dental information, we can potentially unlock a wealth of insights into an individual’s cardiovascular risk profile, offering a non-invasive and readily accessible screening tool.
The potential benefits are significant. AI’s ability to discern complex relationships within dental parameters, often subtle and beyond the scope of traditional analysis, can lead to earlier identification of at-risk individuals. This opportunistic screening during dental visits could facilitate timely referrals and the implementation of preventive strategies, potentially mitigating the burden of cardiovascular disease. Furthermore, this approach underscores the importance of the oral-systemic link and the valuable information embedded within dental records that extends beyond oral health alone.
