Development and Validation of an AI-Enabled Composite Oral Score Using Large-Scale Dental Data

Development and Validation of an AI-Enabled Composite Oral Score Using Large-Scale Dental Data | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Development and Validation of an AI-Enabled Composite Oral Score Using Large-Scale Dental Data Sri Kalyan Yarlagadda, Navid Samavati, Mina Ghorbanifarajzadeh, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5375490/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 01 Jul, 2025 Read the published version in Scientific Reports → Version 1 posted 14 You are reading this latest preprint version Abstract This research introduces Oral Score Basic (OS-B), a novel Artificial Intelligence (AI) derived methodology designed to provide a comprehensive, objective assessment of individual teeth and overall oral health. Leveraging data from more than 340,000 patients across 2,558 U.S. dental practices, OS-B combines radiographic findings and periodontal probing depths with a treatment probability-weighted cost function to quantify the severity of dental conditions. The OS-B score aims to address limitations in prior oral health scoring systems by incorporating nuanced clinical data, accounting for disease severity, and providing a scalable, data-driven approach to measuring oral health. This score was developed using Overjet’s FDA-cleared AI platform, which detects dental conditions using bitewing and periapical radiographs, providing a detailed analysis of each tooth. OS-B’s effectiveness was validated by demonstrating a strong correlation between tooth scores and treatment costs, surpassing the predictive power of previous scoring systems. This research presents a foundational framework for AI-enabled oral health scoring, with potential applications in value-based care, population risk analysis, and consumer health management. Future iterations may expand to include additional dimensions of oral health beyond clinical conditions such as risk factors and measures of oral function and esthetics, further enhancing the score’s clinical utility and patient engagement. Health sciences/Health care/Dentistry/Dental conditions Health sciences/Health care/Dentistry/Dental public health Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Introduction Oral health is a critical component of overall health and well-being yet quantifying it comprehensively has remained a challenge. Over the past five decades, numerous oral health scores have been developed to summarize oral health status and to measure the impact of healthcare interventions. Notable examples include the work of Nikkias et al. [ 1 , 2 ], the Index of Oral Health Status by Markus et al. [ 3 ], the Oral Health Index published by Burke and Wilson [ 4 ] that was later modified and developed by Denplan (Winchester, UK) and renamed the Oral Health Score [ 5 ]. More recently, commercial products such as Previser have emerged as an evidence-based risk score for oral diseases [ 6 ]. While these previous efforts have been valuable, they are constrained by limited sample sizes and often rely on binary disease classifications, failing to capture the nuanced complexity of oral health conditions. The advent of artificial intelligence (AI) and advanced computer vision techniques powered by deep learning, now presents an unprecedented opportunity to revolutionize oral health assessment. The development of a more sophisticated oral health score is imperative, driven by several significant healthcare trends. The ongoing transformation from fee-for-service to value-based care models necessitates robust outcome measures. An AI-derived Oral Health Score could precisely quantify changes in oral health related to clinical interventions, enabling more accurate assessment of care effectiveness. Concurrently, the shift in dental practice modality, with an increasing rate of dentists affiliating with dental support organizations and practicing in groups [ 7 ], provides an opportunity to measure and monitor services provided and their impact on health status. Moreover, the growing consumer interest in health monitoring and management calls for accessible tools that empower individuals. A consumer-friendly oral health score could play a crucial role in early detection and prevention, potentially reducing the need for invasive and costly treatments. Gamification of such a score could further engage and motivate consumers to better manage their oral health. Finally, private or public payers of care would benefit from an objective clinical outcome measure that could be utilized in population risk analysis, plan design, and provider network assessment. These factors collectively underscore the need for a comprehensive, AI-driven oral health score that can serve multiple stakeholders in the healthcare ecosystem. This research aims to address these needs by developing and validating a novel, AI-enabled composite oral health score. Our methodology leverages large-scale data extraction from dental practices to derive a treatment probability-weighted tooth health score which is then averaged to create an aggregate oral score. Because this initial version specifically focuses on dental indicators of health, we refer to the score as Oral Health - Basic (OH-B). This foundational research sets the groundwork for future expansions into more holistic measures of oral health to include oral function and aesthetics, for example, in a way that is consistent with the World Health Organization’s broad definition of oral health [ 8 ]. We hypothesize that AI-driven analysis can yield a more consistent, objective, and scalable approach to oral health scoring, based on the most comprehensive and accurate data available. Our study utilizes AI analysis of dental radiographs from an unprecedented sample of 343,297 patients across 2,558 dental practices in the United States. This extensive dataset allows for a more nuanced and comprehensive assessment of oral health than ever before possible. By developing this innovative scoring system, we aim to provide a valuable tool for clinicians, researchers, and policymakers to better understand, monitor, and improve oral health outcomes. Additionally, the OS-B has the potential to empower consumers in managing their oral health, ultimately contributing to improved overall health and reduced healthcare costs. Methods The development of the OS-B included defining the clinical components of the score, developing a test dataset and subsets, and developing a novel treatment probability weighted cost-function to calculate a weighted individual tooth score from each of the patient’s 28 permanent teeth, excluding third molars. The adult human dentition typically includes up to 32 teeth, including four third molars (wisdom teeth). Contemporary dental public health research increasingly adopts 28-tooth frameworks for population-level studies. This approach provides a more consistent and comparable metric across diverse demographic groups, minimizing confounding variables associated with third molar variability. Individual tooth scores were then averaged into a mouth-level summary score called OS-B. Once constructed, we conducted a preliminary validation of OS-B on the test dataset and compared the OS-B to the Marcus et. al Oral Health Status Index (OHSI) [ 3 ]. The OS-B is built using data from 2,558 dental practices across the United States who used the Overjet, Inc. Practice Application [ 9 ] and includes data from 321,530 adult patients who were 21 years of age or older (Fig. 1 ). All patient data were deidentified in accordance with HIPAA guidelines to ensure confidentiality. The clinical condition of the 28 permanent teeth was assessed using findings from the Overjet AI platform and its proprietary, FDA-cleared Machine Learning Algorithms (MLA) along with periodontal probing depth data from patient electronic records. Overjet’s algorithms detect and segment clinical conditions on bitewing and periapical radiographs, and Figs. 2 and 3 provide examples of how these clinical findings are noted on dental radiographs. The dental conditions analyzed by the Overjet AI platform include: Tooth status as either present, missing, or a root tip, which is defined as a tooth with more than 95 percent of the anatomical crown either missing or decayed. Radiolucencies on the tooth structure indicative of demineralization and/or dental caries. The type and extent of dental restorations on an individual tooth including radiographic evidence of full and partial coverage crowns, fillings, root canals, and/or the presence of a dental implant in place of the tooth. The percentage of the tooth’s coronal tooth structure that is decayed, missing and/or filled, calculated by the Overjet platform as the Decayed, Missing, and/or Filled Proportion (DMFP). Interproximal alveolar bone levels measured in millimeters from the cemento-enamel junction (CEJ) to the most apical crest of the interproximal alveolar bone, as an indicator of the tooth’s periodontal status. Interproximal calculus on cementum for each tooth on both bitewing and periapical radiographs, scored as either absent or present. Periapical Radiolucencies (PARL) on periapical radiographs that may or may not be associated with an endodontic root filling. PARL is scored as either absent or present. Margin discrepancy (MD) where a full or partial coverage crown or filling has a defective margin, an over contoured or under contoured restoration or an overhang where a restorative material extends beyond or over the margin apically. MD is scored as either absent or present. Note that this feature of the Overjet AI platform is not currently FDA cleared but was included in the analysis because it adds information about the quality of existing restorations. Developing the dataset For the purposes of this study, we used deidentified data from 2,558 dental practices, which were randomly divided into three categories: a training dataset (n = 1,808), a validation dataset (n = 254) and a test dataset (n = 496). The training dataset was further subdivided to calculate a treatment probability-weighted cost-function for four clinical conditions: dental caries on teeth without crowns; recurrent dental caries on teeth with crowns; alveolar bone level (ABL) and periodontal probing depth (PD); and periapical radiolucency (PARL). For each patient in the training dataset, we included clinical findings from their most recent dental radiographs, along with treatments provided in the 12 months following the latest radiographs, as documented in the patient record using CDT codes. The average cost associated with each CDT code was calculated across all clinics. Additionally, periodontal pocket depth measurements for each tooth were extracted from the patient records, with the maximum probing depth per tooth serving as an indicator of periodontal status. Each data subset was constructed by applying filtering criteria. Initially, Overjet’s MLA determined the teeth as positive for specific findings and negative for others. Subsequently, the teeth were required to have received a specified set of treatments within one year of detecting a clinical finding being detected on a radiograph, as documented by CDT codes extracted from the patient records. Any treatments provided outside the primary dental practice were not available for inclusion in the dataset. Table 1 provides an overview of the patient count, tooth count, along with the inclusion and exclusion criteria for the overall training dataset and subsets. For example, the caries subset includes teeth identified by Overjet’s MLA as positive for caries and negative for other clinical findings, such as margin discrepancies, calculus, root tips, bone levels exceeding 2.0 mm, PARL, implants, crowns, root tips, and bridges. Additionally, each tooth was required to have received treatment – such as a filling, crown, root canal therapy (RCT), extraction, or implant – within one year from the time of detection, as indicated by CDT codes in the patient’s electronic record, to remain in the dataset. These filtering criteria ensured that teeth included in each dataset were treated primarily due to conditions detected by Overjet AI. Table 1 The number of patients, number of teeth, inclusion and exclusion criteria for the training data set and each data subset for the four specific clinical conditions. Training Data Subset Number of Patients Number of Teeth Overjet AI Positive Findings Overjet AI Findings Exclusions Treatments provided within 12 months of the dental radiographs Overall 321,530 524,298 all not applicable not applicable Caries 292,521 454,111 caries crown, ABL > 2mm, RCT, implant, bridge, PARL, calculus, root tips filling, crown, RCT, extraction and implants Alveolar Bone Level (ABL) and Probing Depth (PD) 6,556 42,951 ABL > 2mm, and the greatest (worst) PD measure for that tooth from the PMS crown, caries, RCT, implant, bridge, PARL, calculus, root tips Scaling and Root Planing (SRP), extraction, implants, and advanced bone level treatments PARL 7,103 7,619 PARL crown, ABL > 2mm, RCT, implant, bridge, caries, calculus, root tips RCT, extraction and implants Crown Recurrent Caries 18,078 19,617 crown + caries ABL > 2mm, RCT, implant, bridge, calculus, root tips crown, extractions and implants Table 2 summarizes patient age and gender distribution across the overall training dataset and within each data subset for the four specific clinical conditions. Patients within the caries subset were slightly younger than those in the overall training dataset. In contrast, patients with the remaining clinical conditions were older, on average, which aligns with the increased prevalence of these conditions with advancing age. Table 2 Summary of patient age (median, mean, standard deviation) and gender distribution for the overall training dataset and subsets defined by four specific clinical conditions. Overall Training Dataset Caries Crown with Recurrent Caries PARL ABL and PD Median patient age 38 years 36 years 51 years 45 years 50 years Mean Age (std. dev) 42.7 years 16.5 std. dev. 40.0 years 16.0 std. dev. 51.0 years 17.0 std. dev. 47.0 years 17.0 std. dev. 50.9 years 16.3 std. dev. Female 183,544 57.1% 166,787 57.0% 11,104 61.4% 3,972 56% 3263 49.8% Male 132,389 41.2% 120,395 41.2% 6,806 37.7% 3,021 42.5% 3280 50.0% Unknown 5567 1.7% 5,308 1.8% 168 0.9% 110 1.5% 13 0.2% Development of a “treatment probability weighted cost-function” to calculate the OS-B tooth scores. This research uses multiple data inputs to derive a novel treatment probability-weighted cost function for determining an individual tooth score. Using tooth-specific treatments administered within 12 months after the dental radiographs and the tooth’s state as calculated by Overjet's MLA, we developed a function to estimate treatment costs based on the tooth's clinical condition. The tooth score is based on the treatment cost needed to restore the tooth. The scoring acknowledges that dental restorations cannot perfectly replicate original tooth health. Higher treatment costs correspond to a lower tooth score, and lower costs correspond to a higher score. Once the individual tooth scores are calculated, the patient’s Oral Score Basic (OS-B) is determined by averaging the tooth scores of 28 individual teeth, excluding third molars. The treatment probability-weighted cost function integrates both the likelihood and cost of various dental treatments indicated for specific clinical conditions. The clinical state of the tooth determines a distribution of possible treatments. The estimated treatment cost is calculated by multiplying the cost of each treatment by its associated probability. Finally, this expected treatment cost is used to adjust the tooth’s health score by subtracting the weighted cost from the base score of 100 (representing a healthy tooth). For example, a score of 100 is assigned to a healthy tooth that exhibits no restorations or pathology. As clinical findings are detected, the score decreases accordingly. For example, a tooth exhibiting initial caries or radiolucent areas of demineralization would have a higher score than a tooth with more extensive caries requiring more invasive and expensive treatment. Conversely, a tooth with extensive caries is assigned a lower score due to the likelihood of needing a multi-surface or full coverage restoration to return it to a state of health. To illustrate how a tooth is scored using the treatment probability-weighted cost-function, we initially focused on the caries data subset, employing the DMFP as a metric for coronal caries severity. Within our training dataset, caries emerged as the most common clinical finding, affecting 85.2% of patients and 67.3% of teeth. Figure 4 A and 4 B plot the probability of treatment and treatment cost against the DMFP value of a tooth with caries, respectively. At low DMFP, the treatment cost is relatively low because only a small proportion of coronal tooth structure is compromised by demineralization or caries and a filling is the most prevalent treatment. As the DMFP increases treatment cost increases, as a larger portion of the tooth is compromised, necessitating more extensive interventions such as crowns, root canals, or extraction and placement of implants. These treatments are more invasive and expensive, leading to higher overall treatment costs. Figure 4 B illustrates the relationship between a tooth’s DMFP and its treatment cost for the next 12 months. We approximate this relationship using a second-degree polynomial of the form (green curve in Fig. 4 B). $$\:\varvec{C}\varvec{o}\varvec{s}\varvec{t}\:=\:\varvec{a}\varvec{*}\varvec{D}\varvec{M}\varvec{F}\varvec{P}\:+\:\varvec{b}\varvec{*}{\varvec{D}\varvec{M}\varvec{F}\varvec{P}}^{2}\:$$ Values of a and b are obtained using the least squares regression algorithm. The tooth score is calculated by subtracting points from 100, with the deduction proportional to treatment costs over the next 12 months. This is mathematically realized by linearly scaling the polynomial via the following 2 constraints: no points are deducted when the DMFP is 0, and 100 points are deducted when the DMFP = 1. Figure 4 C plots tooth score as a function of DMFP. Figure 5 A illustrates the variation in treatment costs as a function of DMFP. At lower DMFP values, indicating less compromised coronal tooth structure, fillings are the most common treatment, with costs ranging from $ 200 to $ 600. As DMFP increases, the likelihood of full-coverage restorations (crowns) and extractions followed by dental implant placements, also increases, resulting in higher associated costs, as shown in Fig. 5 B. Consequently, the cost distribution shifts upward, and for DMFP values exceeding 0.8, extraction and implant placement become the most likely treatment, with typical costs ranging from $ 3,000 to $ 4,000. A tooth's score after restoration depends on two factors: the severity of the decay and compromised coronal tooth structure and the understanding that dental treatments cannot fully restore a tooth to perfect health. Our research estimates that restored teeth regain approximately 80% of their original health status. The severity is measured by the tooth's Average DMFP which is defined as $$\:{DMFP}_{average}=\frac{{\sum\:}_{DMFP=0}^{1.0}P\left(treatment\:\right|\:DMFP)*\:DMFP}{P\left(treatment\right)\:}\:$$ Here \(\:P\left(treatment\:\right|\:DMFP)\:\) denotes the probability of treatment for a given DMFP, derived from Fig. 5 B. For example, the average DMFP for a crown treatment is 0.59. A tooth with this level of decay loses 50 points from its score. After crown placement, the tooth recovers 80% of these lost points, meaning only 10 points (20% of 50) are permanently deducted. This scoring system reflects that while restorative treatments significantly improve tooth function, they cannot achieve the same level of health as an original, undamaged tooth. Table 3 includes the determination of weightings for four types of restorations: 1) a full coverage restoration (crown); 2) a root canal treatment; 3) filling; and 4) an extraction and placement of a dental implant. Table 3 Determination of weightings various restorations based on the tooth’s DMFP. Restoration Average DMFP Points Deducted due to Average DMFP Points Deducted due to restoration = 0.2 * (Points deducted due to Average DMFP) Crown 0.59 50 10 RCT 0.7 61.5 12.3 Filling 0.32 [23, 50] [4.6, 10] Implant (Extraction) 0.72 64.5 12.9 The number of points deducted for a filling depends on its size, with a minimum deduction of 4.6 points and a maximum of 10 points. Here we capped the point deductions for fillings to that of a crown treatment because filling treatments generally retain more original coronal tooth structure as compared to a crown. We used a simple weighted average technique to determine point deductions for PARL and recurrent caries in the presence of a crown. For each of these conditions we obtained the probability distribution of different treatment types, and used the DMFP-based point deduction for each of those treatments together with the probabilities as the weights, to find the average point deductions. Table 4 provides a summary of the treatment distributions and corresponding point deductions for PARL and recurrent caries under crown restorations, as represented by the following formula: $$\:{TS}_{condition}={\sum\:}_{t=1}^{n}P\left(t\right){TS}_{t}$$ Where \(\:{TS}_{condition}\) represents either PARL or recurrent caries under crown, \(\:P\left(t\right)\) denotes the probability of a given treatment for the condition, and \(\:{TS}_{t}\) is the DMFP-based point deduction for the treatment performed for the condition. Table 4 Summary of treatment distribution and tooth score point deductions for PARL and recurrent caries associated with a crown restoration. Condition Root Canal (RCT) Crown Extraction + Implant Points Deduction PARL 59% – 41% 63.3 Crown Recurrent Caries -- 82% 18% 52.6 Points are deducted when a tooth's bone levels exceed 2.0 mm, where a measurement ≤ 2.0 mm is considered healthy. The deduction amount is proportional to the treatment cost at that bone level. Figures 6 A and 6 B illustrate the treatment probability and associated costs over the next 12 months as a function of a tooth's bone level. Like methodology used in caries analysis, a first-degree polynomial is used to approximate the relationship between treatment cost and bone level of a tooth (red curve in Fig. 6 B) $$\:\varvec{C}\varvec{o}\varvec{s}\varvec{t}\:=\:\varvec{a}\varvec{*}\varvec{B}\varvec{L}\:+\:\varvec{b}\:$$ The values of \(\:a\) and \(\:b\) are determined using the least squares regression algorithm. This first-order polynomial is then linearly scaled based on two constraints: no points are deducted when the bone level is less than or equal to 2.0 mm, and 63.5 points are deducted when the bone level reaches 6.71 mm. Like the Average DMFP, the Average Bone Level ( \(\:BL\) ) for extraction is 6.71 mm. We propose that the tooth score for a tooth requiring implant treatment, whether due to elevated bone levels or severe caries, should be the same. Figure 6 C plots tooth score as a function of bone level. Interproximal calculus on a tooth’s cementum typically requires scaling and root planing (SRP) treatment, with an associated cost equal to that of a tooth displaying a DMFP of 0.07, as seen in Fig. 4 B. According to the relationship between DMFP and tooth score (Fig. 4 C), 4.5 points are deducted from a score of 100 at this DMFP. Therefore, the presence of interproximal calculus results in a 4.5-point deduction. Similarly, a tooth typically requires SRP treatment when its probing depth exceeds 4 mm. Following the same point deduction approach as for interproximal calculus, 4.5 points are deducted when the probing depth surpasses 4 mm. Point deductions due to Margin Discrepancy (MD) vary based on its type. If the margin discrepancy occurs on a filling, the deduction is based on the tooth's DMFP. If the MD occurs on a crown, we assume that the tooth requires crown replacement, leading to a deduction of 50 points. A deduction of 100 points is applied when a tooth is missing or when only a root tip remains. Table 5 summarizes the point deductions for each clinical condition. Table 5 Summary of point deductions for each clinical condition. Condition Points Deduction Missing Tooth 100 Root Tip 100 PARL 63.3 Crown Recurrent Caries 52.6 Caries 60.41*DMFP + 39.59*DMFP2 Bone Level (BL) 13.67*BL − 27.33 PD > 4mm or Interproximal Calculus on cementum 4.5 Margin Discrepancy (on filling) (60.41*DMFP + 39.59*DMFP2) Margin Discrepancy (on Crown) 50 Filling [4.6, 10] Crown 10 RCT 12.3 Implant 12.9 The previous sections explored how each of the eight clinical findings affects individual tooth scores. Each finding, based on its nature, deducts a specific number of points from an ideal score of 100. When multiple findings are present, each deduction is calculated separately and then they are combined, as illustrated in Fig. 7 . The total deduction is subtracted from 100 to yield the final tooth score, while missing teeth and root tips are automatically assigned a score of zero. Since multiple conditions can often be addressed with a single restorative or endodontic procedure, treatment costs are non-additive. Thus, deductions for decay, MD, and PARL are combined by taking the maximum value among these findings. Similarly, deductions for elevated probing depth and interproximal calculus are also combined using the the maximum value, as both conditions are typically treated together through SRP. Bone level deductions are treated independently from other findings, reflecting their distinct nature and specific treatment requirements. Restorative deductions (crowns and fillings) are only applied if there is no concurrent MD or decay, as restorations are automatically accounted for by the DMFP when these conditions are present. Figure 7 provides an illustration of the calculation process for individual tooth scores, while Figs. 8 and 9 demonstrate the application of these calculations in patient cases. Results The dataset and subsets developed for this study are large, geographically dispersed, and generally represent the population of patients who seek care at dental practices across the United States. There are slightly more females than males, which is expected because females have slightly higher annual dental visit rates as compared to males. For example, the 2020 National Health Interview Survey (NHIS) indicates that 69.4% of females visit the dentist annually as compared to 64.2% of males [ 10 ]. The distribution of clinical findings as seen in Table 1 approximate epidemiological studies of the prevalence of these conditions [ 11 ]. Correlation of OS-B with Tooth Treatment Cost Prior work by Marcus et al. [ 3 ] to develop an Oral Health Status Index (OHSI) used a paired preference technique and data from 232 simulated adult patient cases to create 315 pairs; 12 dentists were asked to choose the healthier patient in each pair. This information was then used to determine weights for each clinical finding. The scores of all 32 teeth were summed to generate the overall oral score. We compared the two dental scoring systems, the OHSI tooth level score and the new OS-B tooth score, by examining how well they predict future treatment costs. We analyzed data from 124,583 teeth across 36,164 patients in 454 clinics not involved in OS-B's development. The study used CDT codes to determine treatment provided within that dental practice within 12 months of the date of the dental radiographs. We calculated both OHSI and OS-B scores for each tooth and compared them to treatment costs using Pearson correlation coefficients [ 12 ]: OHSI Score: -0.134 OS-B Score: -0.441 The negative correlations indicate that healthier teeth (higher scores) require less expensive treatments. OS-B showed significantly stronger predictive power (-0.441) compared to OHSI (-0.134), representing a 200% improvement. This improved accuracy stems from OS-B's ability to account for disease severity. For instance, while OHSI deducts the same 2.4 points for both minor and severe cavities, OS-B assigns different scores based on caries severity as measured by the DMFP, with the understanding that more severe cavities result in higher treatment costs. Impact Analysis of Clinical Findings OS-B evaluates tooth health using nine clinical findings, each weighted differently to calculate the final tooth score. To understand the importance of each finding, we performed a leave-one-out analysis, removing one component at a time and measuring how this affects the score's ability to predict future treatment costs. Results (correlation with future treatment costs): Complete OS-B Score: -0.441 Without Caries: -0.224 Without Bone Loss: -0.446 Without PARL: -0.440 Without Restorations: -0.418 Removing the caries component caused the most significant drop in predictive power (from − 0.441 to -0.224). This makes sense clinically as caries is a common, treatable condition that often requires expensive procedures (fillings, root canals, extractions and implants). In contrast, removing other components had minimal impact. Bone loss, for example, barely affected the correlation (-0.446). Similarly, existing restorations without active disease (-0.418) typically do not need immediate treatment. OS-B Scores: Age and Gender Patterns Analysis of OS-B scores demonstrates predictable patterns across age and gender demographics. As expected, oral health scores progressively decline with age, reflecting the cumulative impact of dental diseases over time. Gender-based analysis reveals a consistent pattern where women maintain marginally higher OS-B scores compared to men across all age groups. This gender disparity aligns with established national health data, which documents men's increased susceptibility to oral health challenges, including higher rates of periodontal disease, oral cancer, and dental trauma, often attributed to less rigorous oral hygiene practices and fewer dental visits [ 13 ]. These demographic trends in OS-B scores are visually represented in Fig. 10 . Discussion Our study represents a significant advancement in oral health assessment through the application of artificial intelligence and computer vision to analyze radiographic and clinical data from 2,558 U.S. dental practices. The novel treatment probability-weighted cost function provides a more sophisticated approach to quantifying oral health compared to previous methodologies. The OS-B addresses key limitations of previous scoring systems, notably the Oral Health Status Index [ 3 ]. While OHSI was valuable, its development was constrained by limited clinical examiners and sample size. Additionally, OHSI's binary categorization of complex conditions like dental caries failed to capture disease severity, which is a crucial determinant of treatment needs and costs. Our validation demonstrates OS-B's superior predictive power for future treatment costs (correlation coefficient − 0.441 versus − 0.134 for OHSI), representing a 200% improvement. Impact analysis identified dental caries as the strongest predictor of future treatment costs, affecting 85.2% of patients and 67.3% of teeth in our dataset. However, this finding may partially reflect methodological constraints in periodontal assessment, which was limited to interproximal bone levels and pocket depth measurements. The OS-B demonstrated expected demographic trends across age and gender, aligning with established epidemiological patterns. However, several limitations warrant acknowledgment, including the reliance on radiographic findings from patients with dental visits. The cost-based weighting considers CDT codes for care that was delivered to each patient. However, we do not take into account care that was recommended and not provided, nor do we know why that treatment was not completed. We also did not consider any dental care that was provided by a dental specialist or other dental practitioner beyond the practice data available for investigation. However, because the dataset is derived from many dental practices across the US, it is likely to be representative of general dental care provided to patients in the US as compared to studies that include a smaller number of patients or care provided by a more limited panel of clinicians. While the OS-B represents a significant advancement, it is limited by its reliance on radiographic findings from patients with dental visits and limited periodontal measures and does not account for soft tissue conditions, measures of oral function or other patient-reported oral health measures. This research focused on adult patients and was not intended to be applicable to dental patients under the age of 21 years. Future iterations should aim to incorporate these factors, be extended to other age groups, and undergo additional clinical validation in various patient groups or populations. This research should also be expanded to focus on risk indicators, including bio-behavioral variables as well as information about the patient's medical conditions and medications. Future research should also explore the relationship of the score to dental practice type, as well as to additional provider and patient characteristics. Conclusion To the best of our knowledge, OS-B represents the first large-scale data-driven approach to summarize the health status of individual teeth as well as provide a patient-level score. Our approach leverages dental healthcare costs as an objective measure to quantify the severity of various conditions, which were incorporated into the current definition of OS-B. Except for probing depth measurements, OS-B can be automatically calculated based on a detailed analysis of patients’ dental radiographs using our Overjet AI platform. OS-B shows good trends at the population level such as decreasing with age, showing some differences between men and women. Our approach of using treatment cost for each tooth as a basis paves the way to an oral score with multiple potential applications and benefits. While the current iteration of OS-B shows considerable promise, the current OS-B does not account for treatment planning and the nuanced process of prioritizing treatment delivery, as well as patient treatment acceptance. Future iterations and clinical validation of the Oral Score should explore how AI and large-scale data can further enhance the OS-B, evolving it into advanced versions that are not only applicable at the population level but also serve as a personalized monitoring tool, placing patients at the center of their oral health management. We strongly believe that the robust evidence presented in this research suggests that AI and large-scale data will profoundly impact the improvement of oral health, with tools like the Oral Score playing a pivotal role in centering care around the patient. Declarations Ethical considerations This study was approved by Advarra IRB for protocol, “Overjet Inc. - 2020OJV4, Performance Analysis of Computer Vision Algorithms on Detection of Dental-Based Diseases - A Pilot study to Establish Feasibility (Pro00042845)”. The data used in this study was collected by dental practices and/or practitioners who contract with Overjet, Inc. ("Overjet") for use of Overjet's dental diagnostic SaaS product. Pursuant to the terms of agreement between Overjet and its customers, customers are responsible for obtaining the requisite consents from patients to enable Overjet to deliver and develop its product. Patient data was de-identified in compliance with HIPAA regulations before use and analysis in this study. Competing Interests All authors are full-time employees of Overjet, Inc. Author Contribution S.K.Y., M.G., and V.L. contributed equally to this work. N.S., A.S., T.A.D., and W.I. have jointly supervised this work. All authors reviewed the manuscript. Data Availability The datasets generated during and/or analyzed during the current study are not publicly available because the underlying data is subject to confidentiality obligations required by the data owner and/or protected health information under HIPAA. Data that is not subject to HIPAA may be available from the corresponding author on reasonable request provided that the data owner has granted permission to make the data available. References Nikias, M. M., Lollecity, W. A. & Fink, R. An empirical approach to developing multidimensional oral health status profiles. J. Public. Health Dent. 38 , 148–158 (1978). Nikias, M. K., Sollecito, W. A. & Fink, R. An oral health index based on ranking of oral status profiles by panels of dental professionals. J. Public. Health Dent. 39 , 16–26 (1979). Marcus, M., Koch, A. L. & Gershen, J. A. A proposed index of oral health status: a practical application. J. Am. Dent. Assoc. 107 , 729–733 (1983). Burke, F. J. & Wilson, N. H. Measuring oral health: an historical view and details of a contemporary oral health index (OHX). Int. Dent. J. 45 , 358–370 (1995). Burke, F. J. T. et al. Evaluation of an oral health scoring system by dentists in general dental practice. Br. Dent. J. 194 , 215–218 (2003). discussion 205. PreViser | Dental Risk and Periodontal Disease Analysis Software & PreViser https://www.previser.com/ Changing Practice Modalities Among U.S. Dentists. https://www.ada.org/resources/research/health-policy-institute/dental-practice-research/practice-modalities-among-us-dentists Oral health. https://www.who.int/health-topics/oral-health Overjet, I. https://www.overjet.com Cha, A. E. & Cohen, R. A. Dental Care Utilization Among Adults Aged 18–64: United States, 2019 and 2020. NCHS Data Brief. 1–8 (2022). Oral Health in America. Advances and Challenges | National Institute of Dental and Craniofacial Research. (2021). https://www.nidcr.nih.gov/research/oralhealthinamerica Fleetwood, D. & QuestionPro Pearson Correlation Coefficient: Calculation + Examples. (2020). https://www.questionpro.com/blog/pearson-correlation-coefficient/ Lipsky, M. S., Su, S., Crespo, C. J. & Hung, M. Men and Oral Health: A Review of Sex and Gender Differences. Am. J. Men’s Health . 15 , 15579883211016360 (2021). Additional Declarations Competing interest reported. All authors are full-time employees of Overjet, Inc. Cite Share Download PDF Status: Published Journal Publication published 01 Jul, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 18 Feb, 2025 Reviews received at journal 09 Feb, 2025 Reviewers agreed at journal 07 Feb, 2025 Reviews received at journal 04 Feb, 2025 Reviewers agreed at journal 23 Jan, 2025 Reviews received at journal 12 Jan, 2025 Reviewers agreed at journal 12 Jan, 2025 Reviewers agreed at journal 12 Jan, 2025 Reviewers agreed at journal 12 Jan, 2025 Reviewers invited by journal 12 Jan, 2025 Editor assigned by journal 06 Jan, 2025 Editor invited by journal 30 Dec, 2024 Submission checks completed at journal 27 Dec, 2024 First submitted to journal 01 Nov, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5375490","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":395018799,"identity":"4b5efa3c-7729-4a65-b904-5fa43f3f5464","order_by":0,"name":"Sri Kalyan Yarlagadda","email":"","orcid":"","institution":"Overjet, Inc.","correspondingAuthor":false,"prefix":"","firstName":"Sri","middleName":"Kalyan","lastName":"Yarlagadda","suffix":""},{"id":395018800,"identity":"b1d7cf25-5923-4002-8047-a4fa5b4dbf2a","order_by":1,"name":"Navid Samavati","email":"","orcid":"","institution":"Overjet, Inc.","correspondingAuthor":false,"prefix":"","firstName":"Navid","middleName":"","lastName":"Samavati","suffix":""},{"id":395018801,"identity":"28a80550-b8cf-49a8-b9e2-c5c50566ea9d","order_by":2,"name":"Mina Ghorbanifarajzadeh","email":"","orcid":"","institution":"Overjet, Inc.","correspondingAuthor":false,"prefix":"","firstName":"Mina","middleName":"","lastName":"Ghorbanifarajzadeh","suffix":""},{"id":395018804,"identity":"93cbb605-0d4a-47eb-929b-d06321b8ebea","order_by":3,"name":"Vlada Levinta","email":"","orcid":"","institution":"Overjet, Inc.","correspondingAuthor":false,"prefix":"","firstName":"Vlada","middleName":"","lastName":"Levinta","suffix":""},{"id":395018805,"identity":"92730ea6-deb5-42ac-a771-9a515fe2a884","order_by":4,"name":"Alireza Sojoudi","email":"","orcid":"","institution":"Overjet, Inc.","correspondingAuthor":false,"prefix":"","firstName":"Alireza","middleName":"","lastName":"Sojoudi","suffix":""},{"id":395018808,"identity":"b7701e53-2253-48a9-9c47-11e8e6242e6b","order_by":5,"name":"Wardah Inam","email":"","orcid":"","institution":"Overjet, Inc.","correspondingAuthor":false,"prefix":"","firstName":"Wardah","middleName":"","lastName":"Inam","suffix":""},{"id":395018809,"identity":"37013d4a-d2cb-4fa9-9a04-82df855cea89","order_by":6,"name":"Teresa A. Dolan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAxElEQVRIiWNgGAWjYLCChAoIBSJ4CKoGq0g4A9VygGgtjG1Q3gFi3GTP3vvsw8N5Non90gcePv7AYCdD2Bae48YzErelJc7sS0g2OMCQTITDJNKYGRK3HTY2OMOQJnGAgZkILfLPgFrmHDa2P8OQ/uMAQz0xtrABtTQcljPgYUgDev8wEVrOAB2WcCxNTuIMQ7LEGYPjhLWwtx9jZvxRY8PD38OT+KGiotqeoBZkCxMYGAxI0QC08ABp6kfBKBgFo2DEAABmZDL7dtRZZQAAAABJRU5ErkJggg==","orcid":"","institution":"Overjet, Inc.","correspondingAuthor":true,"prefix":"","firstName":"Teresa","middleName":"A.","lastName":"Dolan","suffix":""}],"badges":[],"createdAt":"2024-11-01 20:53:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5375490/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5375490/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-07484-7","type":"published","date":"2025-07-01T15:57:47+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":72612458,"identity":"9ca04151-1683-4157-b105-7feeb76b614c","added_by":"auto","created_at":"2024-12-30 10:29:55","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":122801,"visible":true,"origin":"","legend":"\u003cp\u003eThis map illustrates the geographic distribution of the 2,558 dental practices whose data were used to develop the OS-B. These practices are in every U.S. state as well as Puerto Rico.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/3733e228dbd76440db1c5caa.png"},{"id":72612463,"identity":"29905f20-d7ba-4182-981d-96767dcfa84a","added_by":"auto","created_at":"2024-12-30 10:29:55","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":231880,"visible":true,"origin":"","legend":"\u003cp\u003eThis figure illustrates the clinical findings on dental radiographs as they appear in their original state and as analyzed by the Overjet AI platform. Images A and C are the original radiographs; images B and D are analyzed radiographs by the Overjet AI platform. Image B has segmentations in white to represent enamel, purple for pulp, blue for restorations including implant restorations, red for caries, marginal discrepancies in yellow box, calculus in an orange box, and millimeter bone level measurements in green, yellow and red corresponding to the value measured. Tooth numbers are presented in pink. Image D. In addition to identifying caries and measuring bone levels, this radiograph includes a PARL on tooth number 10.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/cb7544f93c69702640e19ba6.png"},{"id":72612472,"identity":"a3a9eaa4-9585-4357-b2b6-742334a5aaa2","added_by":"auto","created_at":"2024-12-30 10:29:56","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":163223,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eTop Row:\u003c/em\u003e The teeth illustrate 2-dimensional segmentations similar to the radiograph. From the top left, in image A illustrates the location of the CEJ on the tooth. The portion of the tooth identified as above or coronal to the dotted line is defined as the coronal portion of the tooth, and the area of the tooth below or apical to the CEJ is considered the root portion of the tooth. Image B depicts in 2 dimensions how the tooth is segmented to calculate DMFP by identifying the proportion of the coronal portion of the tooth that is decayed (red), missing (orange), and filled (green). Image C shows a standard bitewing radiograph without AI generated predictions. Image D shows the AI-analyzed image illustrating the decayed, missing and filled segmentations on a radiograph. The DMFP calculation for tooth number 30 is 0.71. \u003cem\u003eBottom Row:\u003c/em\u003e Image E illustrates the anatomical landmarks that are used to measure the interproximal alveolar bone level: CEJ and crest of bone. The distance between these two points is the reported bone level (BL). This measurement is analyzed on the mesial and distal of each tooth on the radiograph and can be seen on the bone level image (F).\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/39d145ac192c5f81467a9e42.png"},{"id":72612768,"identity":"35a3ca60-c1e3-420c-832e-b4a2f3780326","added_by":"auto","created_at":"2024-12-30 10:37:56","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":80851,"visible":true,"origin":"","legend":"\u003cp\u003e(A) The figure illustrates the relationship between a tooth’s DMFP and treatment distribution. At a low DMFP, fillings are the most likely treatment. As the DMFP increases, the likelihood of more invasive treatments – such as crowns, root canals, or extractions – also rises. The probability of a tooth being crowned peaks at a DMFP of 0.68 before declining, while extractions demonstrate a positive correlation with increasing DMFP. (B) Figure B displays the relationship between treatment cost and DMFP for the same set of teeth. (C) Figure C represents the tooth score and the corresponding point deductions as a function of DMFP.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/01c26b940b218747ba75fcc5.png"},{"id":72612480,"identity":"ac96ec11-7935-46da-82c4-a82c744c32b6","added_by":"auto","created_at":"2024-12-30 10:29:56","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":89402,"visible":true,"origin":"","legend":"\u003cp\u003e(A) Shows the variation of treatment cost as compared to the DMFP of teeth in caries dataset. At each DMFP, the minimum, maximum and mean costs are indicated by markers (-) at bottom, top and in between respectively. At each DMFP, treatments are heavily concentrated around the blue blobs. (B) Displays the treatment distribution relative to DMFP for the same set of teeth in the caries dataset.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/7d01e8b31620bf898c6ae4dd.png"},{"id":72612474,"identity":"4406f968-1911-4325-bab7-05bff80ba597","added_by":"auto","created_at":"2024-12-30 10:29:56","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":72425,"visible":true,"origin":"","legend":"\u003cp\u003e(A) Displays the treatment distribution relative to bone level of teeth in the AL \u0026amp; PD dataset. (B) Illustrates the relationship between treatment cost and bone level for the same set of teeth. (C) Visualizes the tooth score and the corresponding point deductions as a function of bone level.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/83cec2d04fba46e77ba20db1.png"},{"id":72612459,"identity":"a1dae794-22a9-474c-9557-36ae15b53496","added_by":"auto","created_at":"2024-12-30 10:29:55","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":51185,"visible":true,"origin":"","legend":"\u003cp\u003eOS-B is calculated by first determining individual tooth scores, as shown in this figure, and then averaging these scores across the 28 permanent teeth, excluding third molars. Missing teeth and root tips are assigned a score of zero.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/4969218e52b48c021e665947.png"},{"id":72612766,"identity":"2b49296c-6e59-4672-b77f-e26ee0621ac5","added_by":"auto","created_at":"2024-12-30 10:37:55","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":328835,"visible":true,"origin":"","legend":"\u003cp\u003ePatient A’s original full-mouth X-rays (FMX) without AI predictions can be compared to the AI-analyzed FMX. Using AI predictions and the calculated Oral Score for each tooth, the odontogram provides individual tooth scores and an overall Oral Score. Patient A has an overall Oral Score of 65.5, impacted by findings including PARLs, bone levels, caries, calculus, MD, RCTs, and extensive restorative treatments.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/fbf613e4ac8c0baf12500156.png"},{"id":72612767,"identity":"79478106-1ff6-4eaf-bf1c-274f4776124a","added_by":"auto","created_at":"2024-12-30 10:37:56","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":294595,"visible":true,"origin":"","legend":"\u003cp\u003ePatient B’s original FMX and AI-analyzed FMX illustrate predications used to calculate each tooth’s Oral Score. The odontogram indicates an overall Oral Score of 98.7, primarily attributed to six restorations. This patient most likely has a prior history of dental decay that was successfully treated with dental restorations, restoring the patient’s health to an improved but not perfect OS-B of 98.7.\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/9285096f8f116b249c7008cd.png"},{"id":72612765,"identity":"4b349774-f745-425d-9e87-d69fb6d9d5e2","added_by":"auto","created_at":"2024-12-30 10:37:55","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":60066,"visible":true,"origin":"","legend":"\u003cp\u003eOS-B distribution analysis. (A) Shows the distribution of Oral Score Basic (OS-B) across four age groups: 21-40, 41-60, 61-80, and 81-100. Each age group has a violin shape representing the distribution of OS-B scores, with a mean line and 2 lines showing maximum and minimum. (B) Shows the distribution of OS-B scores for male (M) and female (F) patients. Each gender has a violin shape showing the range of OS-B scores, with a line indicating the mean and others for the maximum and minimum scores.\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/e170a7542900032993dd2ceb.png"},{"id":86179090,"identity":"8942224f-effc-4533-bcba-8eb145d2c87c","added_by":"auto","created_at":"2025-07-07 16:15:36","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2611938,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5375490/v1/3222fef9-c1be-40e4-bfdf-9b3088d5b096.pdf"}],"financialInterests":"Competing interest reported. All authors are full-time employees of Overjet, Inc.","formattedTitle":"Development and Validation of an AI-Enabled Composite Oral Score Using Large-Scale Dental Data","fulltext":[{"header":"Introduction","content":"\u003cp\u003e Oral health is a critical component of overall health and well-being yet quantifying it comprehensively has remained a challenge. Over the past five decades, numerous oral health scores have been developed to summarize oral health status and to measure the impact of healthcare interventions. Notable examples include the work of Nikkias et al. [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e], the Index of Oral Health Status by Markus et al. [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e], the Oral Health Index published by Burke and Wilson [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] that was later modified and developed by Denplan (Winchester, UK) and renamed the Oral Health Score [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. More recently, commercial products such as Previser have emerged as an evidence-based risk score for oral diseases [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. While these previous efforts have been valuable, they are constrained by limited sample sizes and often rely on binary disease classifications, failing to capture the nuanced complexity of oral health conditions. The advent of artificial intelligence (AI) and advanced computer vision techniques powered by deep learning, now presents an unprecedented opportunity to revolutionize oral health assessment.\u003c/p\u003e \u003cp\u003eThe development of a more sophisticated oral health score is imperative, driven by several significant healthcare trends. The ongoing transformation from fee-for-service to value-based care models necessitates robust outcome measures. An AI-derived Oral Health Score could precisely quantify changes in oral health related to clinical interventions, enabling more accurate assessment of care effectiveness. Concurrently, the shift in dental practice modality, with an increasing rate of dentists affiliating with dental support organizations and practicing in groups [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e], provides an opportunity to measure and monitor services provided and their impact on health status. Moreover, the growing consumer interest in health monitoring and management calls for accessible tools that empower individuals. A consumer-friendly oral health score could play a crucial role in early detection and prevention, potentially reducing the need for invasive and costly treatments. Gamification of such a score could further engage and motivate consumers to better manage their oral health. Finally, private or public payers of care would benefit from an objective clinical outcome measure that could be utilized in population risk analysis, plan design, and provider network assessment. These factors collectively underscore the need for a comprehensive, AI-driven oral health score that can serve multiple stakeholders in the healthcare ecosystem.\u003c/p\u003e \u003cp\u003eThis research aims to address these needs by developing and validating a novel, AI-enabled composite oral health score. Our methodology leverages large-scale data extraction from dental practices to derive a treatment probability-weighted tooth health score which is then averaged to create an aggregate oral score. Because this initial version specifically focuses on dental indicators of health, we refer to the score as Oral Health - Basic (OH-B). This foundational research sets the groundwork for future expansions into more holistic measures of oral health to include oral function and aesthetics, for example, in a way that is consistent with the World Health Organization\u0026rsquo;s broad definition of oral health [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWe hypothesize that AI-driven analysis can yield a more consistent, objective, and scalable approach to oral health scoring, based on the most comprehensive and accurate data available. Our study utilizes AI analysis of dental radiographs from an unprecedented sample of 343,297 patients across 2,558 dental practices in the United States. This extensive dataset allows for a more nuanced and comprehensive assessment of oral health than ever before possible.\u003c/p\u003e \u003cp\u003e By developing this innovative scoring system, we aim to provide a valuable tool for clinicians, researchers, and policymakers to better understand, monitor, and improve oral health outcomes. Additionally, the OS-B has the potential to empower consumers in managing their oral health, ultimately contributing to improved overall health and reduced healthcare costs.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eThe development of the OS-B included defining the clinical components of the score, developing a test dataset and subsets, and developing a novel treatment probability weighted cost-function to calculate a weighted individual tooth score from each of the patient\u0026rsquo;s 28 permanent teeth, excluding third molars. The adult human dentition typically includes up to 32 teeth, including four third molars (wisdom teeth). Contemporary dental public health research increasingly adopts 28-tooth frameworks for population-level studies. This approach provides a more consistent and comparable metric across diverse demographic groups, minimizing confounding variables associated with third molar variability. Individual tooth scores were then averaged into a mouth-level summary score called OS-B. Once constructed, we conducted a preliminary validation of OS-B on the test dataset and compared the OS-B to the Marcus et. al Oral Health Status Index (OHSI) [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe OS-B is built using data from 2,558 dental practices across the United States who used the Overjet, Inc. Practice Application [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] and includes data from 321,530 adult patients who were 21 years of age or older (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). All patient data were deidentified in accordance with HIPAA guidelines to ensure confidentiality.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe clinical condition of the 28 permanent teeth was assessed using findings from the Overjet AI platform and its proprietary, FDA-cleared Machine Learning Algorithms (MLA) along with periodontal probing depth data from patient electronic records. Overjet\u0026rsquo;s algorithms detect and segment clinical conditions on bitewing and periapical radiographs, and Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e provide examples of how these clinical findings are noted on dental radiographs.\u003c/p\u003e \u003cp\u003eThe dental conditions analyzed by the Overjet AI platform include:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTooth status as either present, missing, or a root tip, which is defined as a tooth with more than 95 percent of the anatomical crown either missing or decayed.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eRadiolucencies on the tooth structure indicative of demineralization and/or dental caries.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eThe type and extent of dental restorations on an individual tooth including radiographic evidence of full and partial coverage crowns, fillings, root canals, and/or the presence of a dental implant in place of the tooth.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eThe percentage of the tooth\u0026rsquo;s coronal tooth structure that is decayed, missing and/or filled, calculated by the Overjet platform as the Decayed, Missing, and/or Filled Proportion (DMFP).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eInterproximal alveolar bone levels measured in millimeters from the cemento-enamel junction (CEJ) to the most apical crest of the interproximal alveolar bone, as an indicator of the tooth\u0026rsquo;s periodontal status.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eInterproximal calculus on cementum for each tooth on both bitewing and periapical radiographs, scored as either absent or present.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePeriapical Radiolucencies (PARL) on periapical radiographs that may or may not be associated with an endodontic root filling. PARL is scored as either absent or present.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eMargin discrepancy (MD) where a full or partial coverage crown or filling has a defective margin, an over contoured or under contoured restoration or an overhang where a restorative material extends beyond or over the margin apically. MD is scored as either absent or present. Note that this feature of the Overjet AI platform is not currently FDA cleared but was included in the analysis because it adds information about the quality of existing restorations.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eDeveloping the dataset\u003c/h2\u003e \u003cp\u003eFor the purposes of this study, we used deidentified data from 2,558 dental practices, which were randomly divided into three categories: a training dataset (n\u0026thinsp;=\u0026thinsp;1,808), a validation dataset (n\u0026thinsp;=\u0026thinsp;254) and a test dataset (n\u0026thinsp;=\u0026thinsp;496). The training dataset was further subdivided to calculate a treatment probability-weighted cost-function for four clinical conditions:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003edental caries on teeth without crowns;\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003erecurrent dental caries on teeth with crowns;\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ealveolar bone level (ABL) and periodontal probing depth (PD); and\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eperiapical radiolucency (PARL).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eFor each patient in the training dataset, we included clinical findings from their most recent dental radiographs, along with treatments provided in the 12 months following the latest radiographs, as documented in the patient record using CDT codes. The average cost associated with each CDT code was calculated across all clinics. Additionally, periodontal pocket depth measurements for each tooth were extracted from the patient records, with the maximum probing depth per tooth serving as an indicator of periodontal status.\u003c/p\u003e \u003cp\u003eEach data subset was constructed by applying filtering criteria. Initially, Overjet\u0026rsquo;s MLA determined the teeth as positive for specific findings and negative for others. Subsequently, the teeth were required to have received a specified set of treatments within one year of detecting a clinical finding being detected on a radiograph, as documented by CDT codes extracted from the patient records. Any treatments provided outside the primary dental practice were not available for inclusion in the dataset.\u003c/p\u003e \u003cp\u003eTable 1 provides an overview of the patient count, tooth count, along with the inclusion and exclusion criteria for the overall training dataset and subsets. For example, the caries subset includes teeth identified by Overjet\u0026rsquo;s MLA as positive for caries and negative for other clinical findings, such as margin discrepancies, calculus, root tips, bone levels exceeding 2.0 mm, PARL, implants, crowns, root tips, and bridges. Additionally, each tooth was required to have received treatment \u0026ndash; such as a filling, crown, root canal therapy (RCT), extraction, or implant \u0026ndash; within one year from the time of detection, as indicated by CDT codes in the patient\u0026rsquo;s electronic record, to remain in the dataset. These filtering criteria ensured that teeth included in each dataset were treated primarily due to conditions detected by Overjet AI.\u003c/p\u003e\u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eThe number of patients, number of teeth, inclusion and exclusion criteria for the training data set and each data subset for the four specific clinical conditions.\u003c/p\u003e\u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTraining Data Subset\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber of Patients\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNumber of Teeth\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eOverjet AI Positive Findings\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eOverjet AI Findings Exclusions\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eTreatments provided within 12 months of the dental radiographs\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eOverall\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e321,530\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e524,298\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eall\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003enot applicable\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003enot applicable\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCaries\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e292,521\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e454,111\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ecaries\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003ecrown, ABL\u0026thinsp;\u0026gt;\u0026thinsp;2mm, RCT, implant, bridge, PARL, calculus, root tips\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003efilling, crown, RCT, extraction and implants\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAlveolar Bone Level (ABL) and Probing Depth (PD)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e6,556\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e42,951\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eABL\u0026thinsp;\u0026gt;\u0026thinsp;2mm, and the greatest (worst) PD measure for that tooth from the PMS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003ecrown, caries, RCT, implant, bridge, PARL, calculus, root tips\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eScaling and Root Planing (SRP), extraction, implants, and advanced bone level treatments\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePARL\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7,103\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7,619\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePARL\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003ecrown, ABL\u0026thinsp;\u0026gt;\u0026thinsp;2mm, RCT, implant, bridge, caries, calculus, root tips\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eRCT, extraction and implants\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCrown Recurrent Caries\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e18,078\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e19,617\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ecrown\u0026thinsp;+\u0026thinsp;caries\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eABL\u0026thinsp;\u0026gt;\u0026thinsp;2mm, RCT, implant, bridge, calculus, root tips\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003ecrown, extractions and implants\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e\u003cp\u003eTable 2 summarizes patient age and gender distribution across the overall training dataset and within each data subset for the four specific clinical conditions. Patients within the caries subset were slightly younger than those in the overall training dataset. In contrast, patients with the remaining clinical conditions were older, on average, which aligns with the increased prevalence of these conditions with advancing age.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSummary of patient age (median, mean, standard deviation) and gender distribution for the overall training dataset and subsets defined by four specific clinical conditions.\u003c/p\u003e\u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"11\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOverall Training Dataset\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCaries\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eCrown with Recurrent Caries\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003ePARL\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eABL and PD\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eMedian patient\u0026nbsp;age\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e38\u003c/p\u003e \u003cp\u003eyears\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e36\u003c/p\u003e \u003cp\u003eyears\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e51\u003c/p\u003e \u003cp\u003eyears\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e45\u003c/p\u003e \u003cp\u003eyears\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e50\u003c/p\u003e \u003cp\u003eyears\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eMean Age (std. dev)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e42.7 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e16.5 std. dev.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e40.0\u003c/p\u003e \u003cp\u003eyears\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e16.0 std. dev.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e51.0\u003c/p\u003e \u003cp\u003eyears\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e17.0 std. dev.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e47.0\u003c/p\u003e \u003cp\u003eyears\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e17.0 std. dev.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e50.9\u003c/p\u003e \u003cp\u003eyears\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e16.3 std. dev.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eFemale\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e183,544\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e57.1%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e166,787\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e57.0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e11,104\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e61.4%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e3,972\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e56%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e3263\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e49.8%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eMale\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e132,389\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e41.2%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e120,395\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e41.2%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e6,806\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e37.7%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e3,021\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e42.5%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e3280\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e50.0%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eUnknown\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5567\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.7%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5,308\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.8%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e168\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.9%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e110\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e1.5%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0.2%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eDevelopment of a \u0026ldquo;treatment probability weighted cost-function\u0026rdquo; to calculate the OS-B tooth scores.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThis research uses multiple data inputs to derive a novel treatment probability-weighted cost function for determining an individual tooth score. Using tooth-specific treatments administered within 12 months after the dental radiographs and the tooth\u0026rsquo;s state as calculated by Overjet's MLA, we developed a function to estimate treatment costs based on the tooth's clinical condition. The tooth score is based on the treatment cost needed to restore the tooth. The scoring acknowledges that dental restorations cannot perfectly replicate original tooth health. Higher treatment costs correspond to a lower tooth score, and lower costs correspond to a higher score. Once the individual tooth scores are calculated, the patient\u0026rsquo;s Oral Score Basic (OS-B) is determined by averaging the tooth scores of 28 individual teeth, excluding third molars.\u003c/p\u003e \u003cp\u003eThe treatment probability-weighted cost function integrates both the likelihood and cost of various dental treatments indicated for specific clinical conditions. The clinical state of the tooth determines a distribution of possible treatments. The estimated treatment cost is calculated by multiplying the cost of each treatment by its associated probability. Finally, this expected treatment cost is used to adjust the tooth\u0026rsquo;s health score by subtracting the weighted cost from the base score of 100 (representing a healthy tooth).\u003c/p\u003e \u003cp\u003eFor example, a score of 100 is assigned to a healthy tooth that exhibits no restorations or pathology. As clinical findings are detected, the score decreases accordingly. For example, a tooth exhibiting initial caries or radiolucent areas of demineralization would have a higher score than a tooth with more extensive caries requiring more invasive and expensive treatment. Conversely, a tooth with extensive caries is assigned a lower score due to the likelihood of needing a multi-surface or full coverage restoration to return it to a state of health.\u003c/p\u003e \u003cp\u003eTo illustrate how a tooth is scored using the treatment probability-weighted cost-function, we initially focused on the caries data subset, employing the DMFP as a metric for coronal caries severity. Within our training dataset, caries emerged as the most common clinical finding, affecting 85.2% of patients and 67.3% of teeth.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA and \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB plot the probability of treatment and treatment cost against the DMFP value of a tooth with caries, respectively. At low DMFP, the treatment cost is relatively low because only a small proportion of coronal tooth structure is compromised by demineralization or caries and a filling is the most prevalent treatment. As the DMFP increases treatment cost increases, as a larger portion of the tooth is compromised, necessitating more extensive interventions such as crowns, root canals, or extraction and placement of implants. These treatments are more invasive and expensive, leading to higher overall treatment costs.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB illustrates the relationship between a tooth\u0026rsquo;s DMFP and its treatment cost for the next 12 months. We approximate this relationship using a second-degree polynomial of the form (green curve in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB).\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:\\varvec{C}\\varvec{o}\\varvec{s}\\varvec{t}\\:=\\:\\varvec{a}\\varvec{*}\\varvec{D}\\varvec{M}\\varvec{F}\\varvec{P}\\:+\\:\\varvec{b}\\varvec{*}{\\varvec{D}\\varvec{M}\\varvec{F}\\varvec{P}}^{2}\\:$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eValues of \u003cem\u003ea\u003c/em\u003e and \u003cem\u003eb\u003c/em\u003e are obtained using the least squares regression algorithm. The tooth score is calculated by subtracting points from 100, with the deduction proportional to treatment costs over the next 12 months. This is mathematically realized by linearly scaling the polynomial via the following 2 constraints: no points are deducted when the DMFP is 0, and 100 points are deducted when the DMFP\u0026thinsp;=\u0026thinsp;1. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC plots tooth score as a function of DMFP.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA illustrates the variation in treatment costs as a function of DMFP. At lower DMFP values, indicating less compromised coronal tooth structure, fillings are the most common treatment, with costs ranging from \u003cspan\u003e$\u003c/span\u003e200 to \u003cspan\u003e$\u003c/span\u003e600. As DMFP increases, the likelihood of full-coverage restorations (crowns) and extractions followed by dental implant placements, also increases, resulting in higher associated costs, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB. Consequently, the cost distribution shifts upward, and for DMFP values exceeding 0.8, extraction and implant placement become the most likely treatment, with typical costs ranging from \u003cspan\u003e$\u003c/span\u003e3,000 to \u003cspan\u003e$\u003c/span\u003e4,000.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eA tooth's score after restoration depends on two factors: the severity of the decay and compromised coronal tooth structure and the understanding that dental treatments cannot fully restore a tooth to perfect health. Our research estimates that restored teeth regain approximately 80% of their original health status.\u003c/p\u003e \u003cp\u003eThe severity is measured by the tooth's Average DMFP which is defined as\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:{DMFP}_{average}=\\frac{{\\sum\\:}_{DMFP=0}^{1.0}P\\left(treatment\\:\\right|\\:DMFP)*\\:DMFP}{P\\left(treatment\\right)\\:}\\:$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eHere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:P\\left(treatment\\:\\right|\\:DMFP)\\:\\)\u003c/span\u003e\u003c/span\u003edenotes the probability of treatment for a given DMFP, derived from Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB. For example, the average DMFP for a crown treatment is 0.59. A tooth with this level of decay loses 50 points from its score. After crown placement, the tooth recovers 80% of these lost points, meaning only 10 points (20% of 50) are permanently deducted. This scoring system reflects that while restorative treatments significantly improve tooth function, they cannot achieve the same level of health as an original, undamaged tooth. Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e includes the determination of weightings for four types of restorations: 1) a full coverage restoration (crown); 2) a root canal treatment; 3) filling; and 4) an extraction and placement of a dental implant.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDetermination of weightings various restorations based on the tooth\u0026rsquo;s DMFP.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRestoration\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAverage DMFP\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePoints Deducted due to Average DMFP\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePoints Deducted due to restoration\u0026thinsp;=\u0026thinsp;0.2 * (Points deducted due to Average DMFP)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCrown\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.59\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eRCT\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e61.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e12.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eFilling\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[23, 50]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e[4.6, 10]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eImplant (Extraction)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e64.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e12.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe number of points deducted for a filling depends on its size, with a minimum deduction of 4.6 points and a maximum of 10 points. Here we capped the point deductions for fillings to that of a crown treatment because filling treatments generally retain more original coronal tooth structure as compared to a crown.\u003c/p\u003e \u003cp\u003eWe used a simple weighted average technique to determine point deductions for PARL and recurrent caries in the presence of a crown. For each of these conditions we obtained the probability distribution of different treatment types, and used the DMFP-based point deduction for each of those treatments together with the probabilities as the weights, to find the average point deductions. Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e provides a summary of the treatment distributions and corresponding point deductions for PARL and recurrent caries under crown restorations, as represented by the following formula:\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:{TS}_{condition}={\\sum\\:}_{t=1}^{n}P\\left(t\\right){TS}_{t}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{TS}_{condition}\\)\u003c/span\u003e\u003c/span\u003e represents either PARL or recurrent caries under crown, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:P\\left(t\\right)\\)\u003c/span\u003e\u003c/span\u003e denotes the probability of a given treatment for the condition, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{TS}_{t}\\)\u003c/span\u003e\u003c/span\u003e is the DMFP-based point deduction for the treatment performed for the condition.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary of treatment distribution and tooth score point deductions for PARL and recurrent caries associated with a crown restoration.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCondition\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRoot Canal (RCT)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCrown\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eExtraction\u0026thinsp;+\u0026thinsp;Implant\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ePoints Deduction\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePARL\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e59%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ndash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e41%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e63.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCrown Recurrent Caries\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e--\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e82%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e18%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e52.6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003ePoints are deducted when a tooth's bone levels exceed 2.0 mm, where a measurement \u0026le; 2.0 mm is considered healthy. The deduction amount is proportional to the treatment cost at that bone level. Figures\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA and \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB illustrate the treatment probability and associated costs over the next 12 months as a function of a tooth's bone level.\u003c/p\u003e \u003cp\u003eLike methodology used in caries analysis, a first-degree polynomial is used to approximate the relationship between treatment cost and bone level of a tooth (red curve in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB)\u003cdiv id=\"Equd\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\n$$\\:\\varvec{C}\\varvec{o}\\varvec{s}\\varvec{t}\\:=\\:\\varvec{a}\\varvec{*}\\varvec{B}\\varvec{L}\\:+\\:\\varvec{b}\\:$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThe values of \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:a\\)\u003c/span\u003e\u003c/span\u003e and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:b\\)\u003c/span\u003e\u003c/span\u003e are determined using the least squares regression algorithm. This first-order polynomial is then linearly scaled based on two constraints: no points are deducted when the bone level is less than or equal to 2.0 mm, and 63.5 points are deducted when the bone level reaches 6.71 mm. Like the Average DMFP, the Average Bone Level (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:BL\\)\u003c/span\u003e\u003c/span\u003e) for extraction is 6.71 mm. We propose that the tooth score for a tooth requiring implant treatment, whether due to elevated bone levels or severe caries, should be the same. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC plots tooth score as a function of bone level.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eInterproximal calculus on a tooth\u0026rsquo;s cementum typically requires scaling and root planing (SRP) treatment, with an associated cost equal to that of a tooth displaying a DMFP of 0.07, as seen in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB. According to the relationship between DMFP and tooth score (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC), 4.5 points are deducted from a score of 100 at this DMFP. Therefore, the presence of interproximal calculus results in a 4.5-point deduction. Similarly, a tooth typically requires SRP treatment when its probing depth exceeds 4 mm. Following the same point deduction approach as for interproximal calculus, 4.5 points are deducted when the probing depth surpasses 4 mm.\u003c/p\u003e \u003cp\u003ePoint deductions due to Margin Discrepancy (MD) vary based on its type. If the margin discrepancy occurs on a filling, the deduction is based on the tooth's DMFP. If the MD occurs on a crown, we assume that the tooth requires crown replacement, leading to a deduction of 50 points. A deduction of 100 points is applied when a tooth is missing or when only a root tip remains. Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e summarizes the point deductions for each clinical condition.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary of point deductions for each clinical condition.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCondition\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePoints Deduction\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMissing Tooth\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRoot Tip\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePARL\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e63.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCrown Recurrent Caries\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e52.6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCaries\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e60.41*DMFP\u0026thinsp;+\u0026thinsp;39.59*DMFP2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBone Level (BL)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13.67*BL \u0026minus;\u0026thinsp;27.33\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePD\u0026thinsp;\u0026gt;\u0026thinsp;4mm or Interproximal Calculus on cementum\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMargin Discrepancy (on filling)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(60.41*DMFP\u0026thinsp;+\u0026thinsp;39.59*DMFP2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMargin Discrepancy (on Crown)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFilling\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e[4.6, 10]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCrown\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRCT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eImplant\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe previous sections explored how each of the eight clinical findings affects individual tooth scores. Each finding, based on its nature, deducts a specific number of points from an ideal score of 100. When multiple findings are present, each deduction is calculated separately and then they are combined, as illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e. The total deduction is subtracted from 100 to yield the final tooth score, while missing teeth and root tips are automatically assigned a score of zero.\u003c/p\u003e \u003cp\u003eSince multiple conditions can often be addressed with a single restorative or endodontic procedure, treatment costs are non-additive. Thus, deductions for decay, MD, and PARL are combined by taking the maximum value among these findings. Similarly, deductions for elevated probing depth and interproximal calculus are also combined using the the maximum value, as both conditions are typically treated together through SRP. Bone level deductions are treated independently from other findings, reflecting their distinct nature and specific treatment requirements. Restorative deductions (crowns and fillings) are only applied if there is no concurrent MD or decay, as restorations are automatically accounted for by the DMFP when these conditions are present. Figure\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e provides an illustration of the calculation process for individual tooth scores, while Figs.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e and \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e demonstrate the application of these calculations in patient cases.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eThe dataset and subsets developed for this study are large, geographically dispersed, and generally represent the population of patients who seek care at dental practices across the United States. There are slightly more females than males, which is expected because females have slightly higher annual dental visit rates as compared to males. For example, the 2020 National Health Interview Survey (NHIS) indicates that 69.4% of females visit the dentist annually as compared to 64.2% of males [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. The distribution of clinical findings as seen in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e approximate epidemiological studies of the prevalence of these conditions [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003eCorrelation of OS-B with Tooth Treatment Cost\u003c/h3\u003e\n\u003cp\u003ePrior work by Marcus et al. [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e] to develop an Oral Health Status Index (OHSI) used a paired preference technique and data from 232 simulated adult patient cases to create 315 pairs; 12 dentists were asked to choose the healthier patient in each pair. This information was then used to determine weights for each clinical finding. The scores of all 32 teeth were summed to generate the overall oral score.\u003c/p\u003e \u003cp\u003eWe compared the two dental scoring systems, the OHSI tooth level score and the new OS-B tooth score, by examining how well they predict future treatment costs. We analyzed data from 124,583 teeth across 36,164 patients in 454 clinics not involved in OS-B's development. The study used CDT codes to determine treatment provided within that dental practice within 12 months of the date of the dental radiographs.\u003c/p\u003e \u003cp\u003eWe calculated both OHSI and OS-B scores for each tooth and compared them to treatment costs using Pearson correlation coefficients [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eOHSI Score: -0.134\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eOS-B Score: -0.441\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThe negative correlations indicate that healthier teeth (higher scores) require less expensive treatments. OS-B showed significantly stronger predictive power (-0.441) compared to OHSI (-0.134), representing a 200% improvement. This improved accuracy stems from OS-B's ability to account for disease severity. For instance, while OHSI deducts the same 2.4 points for both minor and severe cavities, OS-B assigns different scores based on caries severity as measured by the DMFP, with the understanding that more severe cavities result in higher treatment costs.\u003c/p\u003e\n\u003ch3\u003eImpact Analysis of Clinical Findings\u003c/h3\u003e\n\u003cp\u003eOS-B evaluates tooth health using nine clinical findings, each weighted differently to calculate the final tooth score. To understand the importance of each finding, we performed a leave-one-out analysis, removing one component at a time and measuring how this affects the score's ability to predict future treatment costs.\u003c/p\u003e \u003cp\u003eResults (correlation with future treatment costs):\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eComplete OS-B Score: -0.441\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eWithout Caries: -0.224\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eWithout Bone Loss: -0.446\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eWithout PARL: -0.440\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eWithout Restorations: -0.418\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eRemoving the caries component caused the most significant drop in predictive power (from \u0026minus;\u0026thinsp;0.441 to -0.224). This makes sense clinically as caries is a common, treatable condition that often requires expensive procedures (fillings, root canals, extractions and implants). In contrast, removing other components had minimal impact. Bone loss, for example, barely affected the correlation (-0.446). Similarly, existing restorations without active disease (-0.418) typically do not need immediate treatment.\u003c/p\u003e\n\u003ch3\u003eOS-B Scores: Age and Gender Patterns\u003c/h3\u003e\n\u003cp\u003eAnalysis of OS-B scores demonstrates predictable patterns across age and gender demographics. As expected, oral health scores progressively decline with age, reflecting the cumulative impact of dental diseases over time. Gender-based analysis reveals a consistent pattern where women maintain marginally higher OS-B scores compared to men across all age groups. This gender disparity aligns with established national health data, which documents men's increased susceptibility to oral health challenges, including higher rates of periodontal disease, oral cancer, and dental trauma, often attributed to less rigorous oral hygiene practices and fewer dental visits [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. These demographic trends in OS-B scores are visually represented in Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eOur study represents a significant advancement in oral health assessment through the application of artificial intelligence and computer vision to analyze radiographic and clinical data from 2,558 U.S. dental practices. The novel treatment probability-weighted cost function provides a more sophisticated approach to quantifying oral health compared to previous methodologies. The OS-B addresses key limitations of previous scoring systems, notably the Oral Health Status Index [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. While OHSI was valuable, its development was constrained by limited clinical examiners and sample size. Additionally, OHSI's binary categorization of complex conditions like dental caries failed to capture disease severity, which is a crucial determinant of treatment needs and costs. Our validation demonstrates OS-B's superior predictive power for future treatment costs (correlation coefficient \u0026minus;\u0026thinsp;0.441 versus \u0026minus;\u0026thinsp;0.134 for OHSI), representing a 200% improvement.\u003c/p\u003e \u003cp\u003eImpact analysis identified dental caries as the strongest predictor of future treatment costs, affecting 85.2% of patients and 67.3% of teeth in our dataset. However, this finding may partially reflect methodological constraints in periodontal assessment, which was limited to interproximal bone levels and pocket depth measurements. The OS-B demonstrated expected demographic trends across age and gender, aligning with established epidemiological patterns. However, several limitations warrant acknowledgment, including the reliance on radiographic findings from patients with dental visits.\u003c/p\u003e \u003cp\u003eThe cost-based weighting considers CDT codes for care that was delivered to each patient. However, we do not take into account care that was recommended and not provided, nor do we know why that treatment was not completed. We also did not consider any dental care that was provided by a dental specialist or other dental practitioner beyond the practice data available for investigation. However, because the dataset is derived from many dental practices across the US, it is likely to be representative of general dental care provided to patients in the US as compared to studies that include a smaller number of patients or care provided by a more limited panel of clinicians.\u003c/p\u003e \u003cp\u003eWhile the OS-B represents a significant advancement, it is limited by its reliance on radiographic findings from patients with dental visits and limited periodontal measures and does not account for soft tissue conditions, measures of oral function or other patient-reported oral health measures. This research focused on adult patients and was not intended to be applicable to dental patients under the age of 21 years. Future iterations should aim to incorporate these factors, be extended to other age groups, and undergo additional clinical validation in various patient groups or populations. This research should also be expanded to focus on risk indicators, including bio-behavioral variables as well as information about the patient's medical conditions and medications. Future research should also explore the relationship of the score to dental practice type, as well as to additional provider and patient characteristics.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eTo the best of our knowledge, OS-B represents the first large-scale data-driven approach to summarize the health status of individual teeth as well as provide a patient-level score. Our approach leverages dental healthcare costs as an objective measure to quantify the severity of various conditions, which were incorporated into the current definition of OS-B. Except for probing depth measurements, OS-B can be automatically calculated based on a detailed analysis of patients\u0026rsquo; dental radiographs using our Overjet AI platform. OS-B shows good trends at the population level such as decreasing with age, showing some differences between men and women. Our approach of using treatment cost for each tooth as a basis paves the way to an oral score with multiple potential applications and benefits.\u003c/p\u003e \u003cp\u003eWhile the current iteration of OS-B shows considerable promise, the current OS-B does not account for treatment planning and the nuanced process of prioritizing treatment delivery, as well as patient treatment acceptance. Future iterations and clinical validation of the Oral Score should explore how AI and large-scale data can further enhance the OS-B, evolving it into advanced versions that are not only applicable at the population level but also serve as a personalized monitoring tool, placing patients at the center of their oral health management. We strongly believe that the robust evidence presented in this research suggests that AI and large-scale data will profoundly impact the improvement of oral health, with tools like the Oral Score playing a pivotal role in centering care around the patient.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthical considerations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by Advarra IRB for protocol, \u0026ldquo;Overjet Inc. - 2020OJV4, Performance Analysis of Computer Vision Algorithms on Detection of Dental-Based Diseases - A Pilot study to Establish Feasibility (Pro00042845)\u0026rdquo;. The data used in this study was collected by dental practices and/or practitioners who contract with Overjet, Inc. (\u0026quot;Overjet\u0026quot;) for use of Overjet\u0026apos;s dental diagnostic SaaS product. Pursuant to the terms of agreement between Overjet and its customers, customers are responsible for obtaining the requisite consents from patients to enable Overjet to deliver and develop its product. Patient data was de-identified in compliance with HIPAA regulations before use and analysis in this study.\u003c/p\u003e\u003cp\u003e\u003ch2\u003eCompeting Interests\u003c/h2\u003e\u003cp\u003eAll authors are full-time employees of Overjet, Inc.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eS.K.Y., M.G., and V.L. contributed equally to this work. N.S., A.S., T.A.D., and W.I. have jointly supervised this work. All authors reviewed the manuscript.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe datasets generated during and/or analyzed during the current study are not publicly available because the underlying data is subject to confidentiality obligations required by the data owner and/or protected health information under HIPAA. Data that is not subject to HIPAA may be available from the corresponding author on reasonable request provided that the data owner has granted permission to make the data available.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eNikias, M. M., Lollecity, W. A. \u0026amp; Fink, R. An empirical approach to developing multidimensional oral health status profiles. \u003cem\u003eJ. Public. Health Dent.\u003c/em\u003e \u003cb\u003e38\u003c/b\u003e, 148\u0026ndash;158 (1978).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNikias, M. K., Sollecito, W. A. \u0026amp; Fink, R. An oral health index based on ranking of oral status profiles by panels of dental professionals. \u003cem\u003eJ. Public. Health Dent.\u003c/em\u003e \u003cb\u003e39\u003c/b\u003e, 16\u0026ndash;26 (1979).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarcus, M., Koch, A. L. \u0026amp; Gershen, J. A. A proposed index of oral health status: a practical application. \u003cem\u003eJ. Am. Dent. Assoc.\u003c/em\u003e \u003cb\u003e107\u003c/b\u003e, 729\u0026ndash;733 (1983).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBurke, F. J. \u0026amp; Wilson, N. H. Measuring oral health: an historical view and details of a contemporary oral health index (OHX). \u003cem\u003eInt. Dent. J.\u003c/em\u003e \u003cb\u003e45\u003c/b\u003e, 358\u0026ndash;370 (1995).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBurke, F. J. T. et al. Evaluation of an oral health scoring system by dentists in general dental practice. \u003cem\u003eBr. Dent. J.\u003c/em\u003e \u003cb\u003e194\u003c/b\u003e, 215\u0026ndash;218 (2003). discussion 205.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePreViser | Dental Risk and Periodontal Disease Analysis Software \u0026amp; PreViser \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.previser.com/\u003c/span\u003e\u003cspan address=\"https://www.previser.com/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChanging Practice Modalities Among U.S. Dentists. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ada.org/resources/research/health-policy-institute/dental-practice-research/practice-modalities-among-us-dentists\u003c/span\u003e\u003cspan address=\"https://www.ada.org/resources/research/health-policy-institute/dental-practice-research/practice-modalities-among-us-dentists\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOral health. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.who.int/health-topics/oral-health\u003c/span\u003e\u003cspan address=\"https://www.who.int/health-topics/oral-health\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOverjet, I. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.overjet.com\u003c/span\u003e\u003cspan address=\"https://www.overjet.com\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCha, A. E. \u0026amp; Cohen, R. A. Dental Care Utilization Among Adults Aged 18\u0026ndash;64: United States, 2019 and 2020. \u003cem\u003eNCHS Data Brief.\u003c/em\u003e 1\u0026ndash;8 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOral Health in America. Advances and Challenges | National Institute of Dental and Craniofacial Research. (2021). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.nidcr.nih.gov/research/oralhealthinamerica\u003c/span\u003e\u003cspan address=\"https://www.nidcr.nih.gov/research/oralhealthinamerica\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFleetwood, D. \u0026amp; QuestionPro Pearson Correlation Coefficient: Calculation\u0026thinsp;+\u0026thinsp;Examples. (2020). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.questionpro.com/blog/pearson-correlation-coefficient/\u003c/span\u003e\u003cspan address=\"https://www.questionpro.com/blog/pearson-correlation-coefficient/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLipsky, M. S., Su, S., Crespo, C. J. \u0026amp; Hung, M. Men and Oral Health: A Review of Sex and Gender Differences. \u003cem\u003eAm. J. Men\u0026rsquo;s Health\u003c/em\u003e. \u003cb\u003e15\u003c/b\u003e, 15579883211016360 (2021).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-5375490/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5375490/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis research introduces Oral Score Basic (OS-B), a novel Artificial Intelligence (AI) derived methodology designed to provide a comprehensive, objective assessment of individual teeth and overall oral health. Leveraging data from more than 340,000 patients across 2,558 U.S. dental practices, OS-B combines radiographic findings and periodontal probing depths with a treatment probability-weighted cost function to quantify the severity of dental conditions. The OS-B score aims to address limitations in prior oral health scoring systems by incorporating nuanced clinical data, accounting for disease severity, and providing a scalable, data-driven approach to measuring oral health. This score was developed using Overjet\u0026rsquo;s FDA-cleared AI platform, which detects dental conditions using bitewing and periapical radiographs, providing a detailed analysis of each tooth. OS-B\u0026rsquo;s effectiveness was validated by demonstrating a strong correlation between tooth scores and treatment costs, surpassing the predictive power of previous scoring systems. This research presents a foundational framework for AI-enabled oral health scoring, with potential applications in value-based care, population risk analysis, and consumer health management. Future iterations may expand to include additional dimensions of oral health beyond clinical conditions such as risk factors and measures of oral function and esthetics, further enhancing the score\u0026rsquo;s clinical utility and patient engagement.\u003c/p\u003e","manuscriptTitle":"Development and Validation of an AI-Enabled Composite Oral Score Using Large-Scale Dental Data","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-12-30 10:29:50","doi":"10.21203/rs.3.rs-5375490/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-02-18T17:34:09+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-02-10T04:49:02+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"205974202245418416197729587868739036380","date":"2025-02-07T15:47:23+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-02-04T10:55:38+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"293086881829493555240449000430209294644","date":"2025-01-24T04:55:06+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-01-13T04:56:29+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"75301104019177910298695221895265928334","date":"2025-01-12T13:24:32+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"217755171726545475171470208220229623787","date":"2025-01-12T12:22:49+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"23943592390177173955483838800056745488","date":"2025-01-12T12:00:37+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-01-12T11:38:14+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-01-06T16:58:38+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2024-12-31T03:31:17+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-12-27T14:00:24+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2024-11-01T20:51:23+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"61da4407-cf76-47a4-ba40-33fc7e9072ce","owner":[],"postedDate":"December 30th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":42093845,"name":"Health sciences/Health care/Dentistry/Dental conditions"},{"id":42093846,"name":"Health sciences/Health care/Dentistry/Dental public health"}],"tags":[],"updatedAt":"2025-07-07T16:03:35+00:00","versionOfRecord":{"articleIdentity":"rs-5375490","link":"https://doi.org/10.1038/s41598-025-07484-7","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-07-01 15:57:47","publishedOnDateReadable":"July 1st, 2025"},"versionCreatedAt":"2024-12-30 10:29:50","video":"","vorDoi":"10.1038/s41598-025-07484-7","vorDoiUrl":"https://doi.org/10.1038/s41598-025-07484-7","workflowStages":[]},"version":"v1","identity":"rs-5375490","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5375490","identity":"rs-5375490","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}