Face Validation of an Artificial Intelligence Driven Tool for Clinical Triaging

Highlights

• Artificial intelligence technology has potential to enhance clinical care in dentistry through clinical triage and care prioritization, but evidence is required on real-world practical applications.
• This study demonstrated an artificial intelligence driven tool that had excellent agreement on clinical triage decisions when compared with dentists.
• The face-validated artificial intelligence driven tool should be explored for broader implementation in Australian public oral healthcare.

Abstract

Objectives

The adoption of artificial intelligence (AI) technology holds significant potential to drive innovation for value-based healthcare for Victorian public oral healthcare, Australia. This pilot study compared the clinical triage decision outcome of experienced dentists in prioritizing consumers with urgent dental care at Your Community Health, a community health service in Victoria, Australia, with those generated by the CoTreat AI-driven tool (CoTreat Pty Ltd, Australia).

Methods

Consecutive sampling of consumers were recruited for the 6-month pilot. Five clinical criteria were established to determine consumers with urgent care: (1) requires urgent extraction causing loss of upper or lower anterior teeth, (2) multiple teeth (≥15% of remaining dentition) with moderate/extensive dental caries, (3) abscessed tooth/teeth, (4) root caries, and (5) missing upper and lower anterior teeth. Face validation was undertaken to compare the agreement between the triage output returned by CoTreat (index test) and 2 experienced dentists (reference standard) in determining urgent care.

Results

A total of 173 consumers were triaged in this pilot study (mean age 61.9 years [SD 18.]). One-third (n = 57) were identified needing urgent care. CoTreat achieved 98.3% agreement with dentist clinical triage outcome (Cohen’s k = 0.96), 100% sensitivity, and 97.5% specificity, with positive predictive value of 94.7% and negative predictive value of 100%. The most common triage criteria were missing anterior teeth (60.3%) and abscesses (20.7%).

Conclusions

CoTreat demonstrated excellent agreement with dentists’ clinical triage decisions. AI-driven tools, such as CoTreat, can advance value-based healthcare implementation by identifying consumers with urgent care needs for Victorian public oral healthcare.

Introduction

Globally, appropriate resource allocation for universal access to essential oral healthcare remains a significant gap in government public healthcare expenditure, particularly in low- and middle- income countries.1,2 High income countries face similar challenges, such as Australia, where 60% of the costs for dental services are paid for by individuals and their families in 2024.3 For adult populations accessing various means-tested Australian state and territory government public oral healthcare, finite funding and stretched resources have meant that many eligible adults experience excessive waiting times. Consumers typically are not seen within the clinical benchmark of 24 months for nonurgent general dental care.4 In Victoria, Australia, the typical length for the adult dental waitlist is 3 years.4,5Eligible adults utilizing Australian state and territory government public oral healthcare are commonly offered comprehensive oral healthcare on a “first come, first served” basis.4,6 Although all 8 independent models of Australian public oral healthcare, provided by the states and territories, offer emergency and comprehensive oral healthcare, each jurisdiction use different triaging processes and may involve a brief oral health assessment by dental practitioners.4Some Australian jurisdictions are embracing value-based healthcare to make the shift from service delivery outputs and focus on measuring outcomes and optimizing value.7 In particular, Victoria8 and New South Wales9 have developed and agreed on statewide patient-reported outcome measures for public oral healthcare. These outcome measures have the potential to inform how to best triage consumers in a more efficient and equitable manner within resource constraints so that consumers with greater oral health needs are offered fast-tracked access to comprehensive oral healthcare, as opposed to being registered and need to wait on the adult dental waitlist.Previous research has shown there are variations in the diagnosis and care planning between dental practitioners, particularly dental caries management,10,11 which can lead to unwarranted variation and low value care.1 An accurate triage process can improve the efficiency and effectiveness of Australian public oral healthcare, by reorientating resources for consumers based on clinical urgency and oral health needs. Additionally, a detailed oral health assessment can inform service providers and dental practitioners to support tailored preventive interventions to promote oral health and shared decision making with consumers. Rapid technological advances in artificial intelligence (AI) in healthcare also presents significant opportunities to promote safety and quality in healthcare,12 including dentistry.There have been broad healthcare applications in using on AI in dentistry.13-15 Most AI in dentistry primarily use dental radiography and oro-facial image scans, ie, clinical photographs,14 to support oral disease diagnosis, risk prediction and clinical decision making.13,14 Other AI applications in dentistry have included patient management and care planning.13 Despite the potential for AI to transform oral healthcare, there are challenges for it to be fully adopted by the dental profession into routine clinical practice.16 Key areas to enhance the uptake of AI in dentistry include demonstrating value, respect, and privacy protections and maintaining trustworthiness and assuring generalisability.16In 2024, Your Community Health (YourCH), a not-for-profit community health service in Victoria, partnered with CoTreat Pty Ltd to integrate an AI-powered platform into its clinical workflow to enhance triage processes for comprehensive oral healthcare. CoTreat uses a proprietary supervised machine learning algorithm to generate “Observational Diagnostics” based on visual inputs, such as clinical photographs and dental radiographs (eg, bitewings and orthopantomographs). These diagnostics classify findings into macrocategories (ie, dental caries, bone loss) and visual subcategories (ie, radiolucency), enabling structured, AI-assisted clinical decision making. The platform supports evidence-based care by mapping diagnostic findings to recommended therapeutic options, continuously updated through automated evidence reviews. Additionally, CoTreat leverages generative AI for follow-up support and treatment plan clarification. Since 2022, these features have been deployed in both chairside and mobile applications, producing over 10 000 individual diagnostic reports across Victoria, primarily in the private sector.17For this venture, the CoTreat platform was repurposed for Australian public oral healthcare for the local context for service provision provided by YourCH. Specifically, the use of Observational Diagnostics was relevant to support the clinical triage process YourCH had implemented (usual care) from 2020 (see Appendix Fig. 1 in Supplemental Materials).To achieve a more equitable resource allocation, reduce dental waitlist times, improve health outcomes, and enhance consumer experiences, an accurate assessment of oral health needs was identified as high priority area to improve Victorian public oral healthcare by YourCH. Therefore, the aim of this study is to evaluate and face validate the use CoTreat as a clinical triage tool for prioritization to lay the foundations for further implementation.

Methods

This retrospective study received approval for exemption from ethical review by Deakin University Human Research Ethics Committee (DUHREC ID 2024/HE000561).

Sample Size Calculation

A priori sample size was estimated based on recommendations for diagnostic accuracy study,18 following the formula for a single-test design of new diagnostic test. Type 1 error was set at 5%. The calculations were based on the results from a study investigating an AI-based tool in the diagnosis of oral diseases on panoramic dental radiographs. Resulted sensitivity for missing teeth served for the new sample size calculation (88.5%). Prevalence of 36.8% was calculated on the occurrence of the missing teeth among the different diagnostic scenarios.19 Including a marginal error of 10%, sample size resulted in 106 and 62 individuals, respectively, for sensitivity and specificity estimations.

Study Design and Participants

This study is reported following the Standards for Reporting of Diagnostic Accuracy (STARD) reporting guidelines while also considering the core principles of the STARD-AI protocol for reporting diagnostic accuracy studies.20 Face validation study assessed the performance of the CoTreat clinical triage tool to determine clinical priority was undertaken into 2 stages with consecutive sampling. In this study, face validity as adapted from the psychology literature,21 was subjectively referred to as the use of the CoTreat to support clinical triage from the perspective of 2 senior dentists and YourCH, the service provider. Face validation infers whether the CoTreat AI-driven platform is appropriate and suitable to support clinical triage for urgent care. Although the consecutive sampling approach can introduce bias, it is the most appropriate recruitment process because it reflects the real-world scenario, in which all consumers on the adult public dental waitlist was offered the intervention.

Stage 1

First, the clinical triage criteria for prioritizing consumers seeking comprehensive oral healthcare at YourCH was formalized within CoTreat’s AI-driven platform. Two senior dentists and the leadership dental team at YourCH met with CoTreat to discuss and confirmed the corresponding decision rules for urgent care whereby consumers had to meet at least 1 of the following conditions:

• requires urgent extraction causing loss of upper or lower anterior teeth
• multiple teeth (≥15% of remaining dentition) with moderate/extensive caries
• abscessed tooth/teeth
• root caries
• missing upper and lower teeth anterior teeth.The 5 clinical criteria for triage decisions had been in use at YourCH since 2020 and were based on the severity of dental conditions as agreed by YourCH and were not weighted, ie, any individual that met at least 1 of the 5 criteria were prioritized for requiring urgent care. Stage 1 included a calibration exercise, in which individual case discussion was conducted with 2 senior dentists (I.C. and J.T.) and CoTreat, with a target sample size of 30 consumers before the clinical triage tool face validation process (stage 2).

Stage 2

After the calibration exercise (stage 1), consumers (18 years and older) seeking comprehensive oral healthcare at YourCH were offered and gave verbal consent to participate in the pilot implementation of the AI-driven triage process. Consumers who were seeking emergency care were excluded. CoTreat was used for the in-person brief oral health assessment performed by dental practitioners (2 dentists, I.C. and T.M.) within a 6-month period between February and July 2024.

Test Methods

The reference standard was the clinical triage decision outcome of the in-person dental appointment, in which the consumer has a brief oral health assessment by 2 senior dentists employed by YourCH. To ensure appropriate diagnostic quality of the clinical photos, intraoral radiographs, and the orthopantomogram taken, in-person training was provided by CoTreat with 2 senior dentists from YourCH involved in this study. These 2 senior dentists were both blinded to the output from the CoTreat AI-powered platform, which typically has a turnaround time of 12-24 hours. Hence, CoTreat’s triage decision was not known to human dental practitioners and vice versa. CoTreat has an in-built image quality assurance monitoring system to reject clinical photos and dental radiographs of poor diagnostic quality.A typical orthodontic set of 5 clinical photos were taken (anterior teeth in occlusion, 2 buccal left and right photos, and 2 upper and lower occlusal photos), along with 2 left and right bitewing intraoral radiographs and an orthopantomogram. These clinical photographs and digital dental radiographs were uploaded into the CoTreat AI-powered platform, generate Observational Diagnostics, which served as the basis for the automated decision making.17The index test was the clinical triage decision outcome returned by CoTreat, which is supported by a human element by having a second-stage oversight by dentists employed by CoTreat, which was not available to the assessors of the reference standard when the triage decision was made and hence did not affect care pathway trajectory of consumers.

Statistical Analysis

Diagnostic performance of the AI-driven triage tool was assessed by comparing its classification outcomes with those of the reference standard, clinical judgments made by senior dentists (I.C. and T.M.) after the in-person brief oral health assessments. Agreement between the AI-driven tool and human assessors was quantified using Cohen’s Kappa coefficient, with values ≥ 0.90 interpreted as indicating excellent agreement. Additional diagnostic accuracy metrics included sensitivity (true-positive rate), specificity (true-negative rate), and overall accuracy, as well as positive predictive value and negative predictive value, calculated using well-established formulae.22 Statistical analysis was performed using the R software (version 4.3.3) via the RStudio integrated development environment.In a secondary analysis for exploratory purposes, the prevalence and severity of certain dental conditions in the cohort were analyzed using CoTreat existing Observational Diagnostics reporting system. Specifically, the diagnostic outcomes of interest were dental caries (International Caries Detection and Assessment System ≥ stage 3 for visual assessment and ≥RB-grade dental radiolucency) as per the International Caries Classification and Management System.23 and preliminary diagnosis of at least stage II periodontitis,24 as determined by periodontal bone loss.

Results

Stage 1

For stage 1, 30 consumers were included for case discussion with senior dentists I.C. and J.T., whereby agreement on the clinical triage decision was based on the clinical criteria for prioritization for general care. The mean age were 59.5 years (SD 18.5) and 19 were female (63.3%).In the calibration exercise, the level of agreement on triage decision was 73.3%. In cases in which the clinical priority decision was in conflict (n = 8), images from patients were assessed to identify improvements to the clinical priority criteria and interpretation for stage 2 (see Appendix Table 1 in Supplemental Materials).

Stage 2

After the calibration exercise, 173 consumers participated at stage 2 and had a mean age of 61.9 years (SD 18.2). Image quality was adequate for all patients (n = 173) to enable triage decisions. Overall, 57 consumers (33%) were triaged as requiring urgent care and offered general care, whereas 116 (67%) were registered with the adult dental waitlist. CoTreat output returned an overall triage decision accuracy of 98.3% (Cohen’s k = 0.96). Sensitivity and specificity were 100% and 97.5%, respectively, with a positive predictive value was 94.7% and a negative predictive value of 100%.Of the 33% of patients triaged for urgent care, the following triage criteria were identified (multiple criteria were detected for 10.4% of patients; Appendix Table 2 in Supplemental Materials): 32.8% required urgent extraction causing loss of upper or lower anterior teeth, 6.9% had multiple teeth (≥15% of remaining dentition) with moderate/extensive caries, 20.7% had abscessed tooth/teeth, 22.4% had root caries, and 60.3% had missing upper and lower teeth anterior teeth. Overall, the dental caries prevalence was 76.3% (95% CI 70.0; 82.7) and the prevalence for preliminary diagnosis of moderate/severe periodontitis was 71.1% (95% CI 64.3; 77.9).

Discussion

Our study demonstrates that the CoTreat AI-powered platform can be repurposed as a triage tool, offering high agreement when compared with dental practitioners, to determine the clinical criteria for prioritizing consumers accessing Victorian oral healthcare. CoTreat has potential to enhance resource allocation by distinguishing urgent and nonurgent oral health needs of consumers using objective observations. The triage decision accuracy of 98.3% observed in this study suggest that CoTreat can reliably complement clinical judgment by correctly identifying patients with severe conditions, such as abscesses, multiple moderate/severe caries or root caries, or those requiring urgent treatments related to compromised/missing anterior teeth. The findings from this study also showed that consumers in our study had a higher prevalence of dental caries and moderate/severe periodontitis when compared with the national oral disease surveillance data report at 32.2% and 51.1%, respectively, for adults aged between 55 and 74 years.25In Victoria, Australia, there is no standardized triage process for eligible adult consumers accessing general care. It is the only Australian jurisdiction that has a Priority Access Policy4 that exempts consumers meeting the eligibility criteria from being placed on the adult dental waitlist. For adults, these are Aboriginal and Torres Strait Islander people, people who are homeless or at risk of homelessness, pregnant women, refugees and asylum seekers, and people registered with mental health or disability services with a letter of recommendation from their case manager.26 Although populations tend to have greater oral health needs,27 the Priority Access Policy do not consider patient-reported outcomes measures, or clinical observations of at the individual level. AI-driven support tools, such as CoTreat, can support more nuanced decisions for general care offered by Victorian public oral healthcare.Applications of AI for the purpose of triage has had greater research visibility with hospital-based environments, particularly in emergency departments.28-30 The 2024 systematic review of prospective studies investigating the use of AI tools used in emergency departments showed there was a triage prediction range of 80.5% to 99.1%.30 Other outcomes reported included impacts increased speed for triage, issues on overtriage and undertriage, mistriage factors, and patient care and prognosis outcomes.30 However, there is limited published literature explicitly evaluating the role of AI in dentistry for the purpose of clinical triage.31 Most recent literature reviews on AI applications in dentistry found that majority of studies were primarily focused on diagnostic performance using of a single source of data (clinical photos or dental radiographs) rather than a combination.15,32Evidence suggests the potential benefits of AI in clinical triage, particularly within emergency medicine. AI-supported systems can enable objective, consistent, and rapid risk stratification and the earlier identification of patients requiring emergency intervention compared with traditional manual processes.33,34 Despite the increasing integration and validation of these approaches in emergency care, their application within dentistry remains largely unexplored.16 This technological gap provided the rationale for the present study, which aimed to investigate how this concept could be translated to the dental setting to support more rapid and objective identification of patients presenting with urgent needs on public dental waitlists. In addition, AI in dentistry has potential for improving oral health literacy for consumers and improving oral health self-care behaviors, such as oral hygiene,35 but no studies to our knowledge have reported the benefits of these AI-enabled applications on oral health outcomes.Presently, YourCH and CoTreat have codeveloped consumer facing information, whereby the virtual oral health assessment reports are communicated with consumers via SMS and a website link directing consumers to their own digital oral health information, including clinical photographs and dental radiographs. Plans are underway to further codesign the AI-driven oral healthcare pathway with consumers to ensure more effective resource allocation, reduce adult public dental waitlist times, improving oral health outcomes and consumer experience. At present, CoTreat is positioned primarily as a decision-support aid, designed to assist dental practitioners in standardizing clinical triage decisions, facilitating routine data collection on oral health outcomes, and supporting training and auditing processes. The interim goal in utilizing CoTreat within YourCH is therefore to augment, rather than replace, clinical judgment. Over time, as the platform matures, there is potential for elements of the triage process to be increasingly automated.In the longer term, CoTreat may help enable a broader role for nondental health professionals (eg, dental assistants, nurse practitioners, and pharmacists) in supporting oral healthcare delivery, particularly in rural and remote settings where access to fixed dental services is limited. This work is aligned with Oral Health Victoria’s 3 value-based healthcare principles to develop, implement, and evaluate models of care: (1) care is codesigned with the person or population, (2) prevention and early intervention are prioritized, and (3) consistent measurement of health outcomes and costs are embedded.36 Reorientating oral healthcare systems with a population health approach using AI, as shared in this pilot study, is also aligned with the World Health Organization’s Global strategy and action plan on oral health 2023 to 2030, specifically Principle 6: Optimizing digital technologies for oral health.37 In 2025, the use of CoTreat is fully embedded at YourCH to support the clinical triage decisions for urgent care with adults coming off the dental waitlist, and utilizes oral health therapists rather than dentists, thereby shifting the expertise of dentists to provide more complex care needs to consumers.

Limitations

Although this pilot study was conducted with methodological rigor, it was not exempt from limitations. A possible drawback of our study design is the use of nonprobabilistic sampling. Although consecutive sampling can introduce selection bias and limit generalizability, it was chosen as the most appropriate recruitment approach because it reflects real-world conditions, in which all consumers on the adult public dental waitlist were offered the intervention. Additionally, our consumer target population who participated in the AI-driven pilot study was small and is unlikely representative of the population eligible for Victorian public oral healthcare. Further research and application of CoTreat as a triage tool is required with diverse population groups is required to assure generalizability.Another important limitation in our study is the reliance on the CoTreat’s AI software to report the epidemiological diagnostic outcomes for dental caries and moderate/severe periodontitis because they were not verified clinically by dental practitioners in-person. Potentially, the CoTreat’s AI software may be overdiagnosing or underdiagnosing these 2 oral diseases. For dental caries, previous research suggests calibration for dental caries diagnosis and using AI with good quality clinical photographs does not significantly affect data quality.38-40 However, this may prove more difficult for diagnosing periodontitis using the new periodontal disease classification for periodontitis,24 given that it relies on additional information that cannot be derived from CoTreat’s Observational Diagnostics, such as whether there is tooth loss due to periodontitis and tooth probing depths for the staging criteria and the grading criteria (eg, direct evidence of bone loss or clinical attachment loss relative to time and modifiable risk factors related to smoking and diabetes).Finally, it remains unknown if the use of AI for the purpose of triage for clinical priority in Victorian public oral healthcare may potentially be excluding consumers who are risk-adverse being exposed to AI or do not have digital health literacy to participate. These health equity considerations must be taken into account when thinking about broader intervention implementation and scale-up of this pilot study. Other implementation considerations should be evaluated, such CoTreat’s shared decision making algorithm from both consumer and dental practitioner perspectives. The PROLIFERATE_AI framework provides an evidence-based approach to do this, by capturing 5 key constructs: (1) Comprehension, (2) Emotional response, (3) Barriers, 4) Motivations, and (5) Optimization Strategies.29

Conclusions

Under the leadership of YourCH, the CoTreat AI-driven clinical triage tool demonstrated high levels of agreement to determine clinical priority for consumers seeking comprehensive oral healthcare through Victorian oral healthcare. Preliminary insights indicate CoTreat had face validity as a clinical triage tool for urgent care. The integration of CoTreat at YourCH to deliver Victorian public oral healthcare can help optimize resource allocation in a more equitable way and supports value-based healthcare in routine measurement of oral health outcomes. AI-driven models of care in dentistry hold promising applications to improve patient outcomes, consumer experiences, and automate oral disease surveillance.

Supplemental Material (2)

Author Disclosures PDF (1.50 MB)

Supplemental Material Document (139.35 KB)

References

1. Nguyen, T.M. ∙ Bridge, G. ∙ Hall, M. ...
   Is value-based healthcare a strategy to achieve universal health coverage that includes oral health? An Australian case studyJ Public Health Policy. 2023; 44(2):310-324
   Google Scholar
2. Winkelmann, J. ∙ Listl, S. ∙ van Ginneken, E. ...
   Universal health coverage cannot be universal without oral healthLancet Public Health. 2023; 8(1):e8-e10
   Google Scholar
3. Oral Health and dental care in Australia. Australian Institute of Health and Welfare
   https://www.aihw.gov.au/reports/dental-oral-health/oral-health-and-dental-care-in-australia/contents/summary
   Date accessed: June 7, 2025
   Google Scholar
4. Introducing competition and informed user choice into human services [inquiry report]. Productivity Commission, Commonwealth of Australia
   https://www.pc.gov.au/inquiries/completed/human-services/reforms/report
   Date accessed: June 7, 2025
   Google Scholar
5. Access to public dental services in Victoria
   Victorian Auditor General’s Office, Victorian Government Printer
   https://www.audit.vic.gov.au/report/access-public-dental-services-victoria?section=
   Date accessed: June 7, 2025
   Google Scholar
6. Jones, K.
   An evaluation of the discriminant and predictive validity of relative social disadvantage as screening criteria for priority access to public general dental care, in Australia
   BMC Health Serv Res. 2014; 14:106
   Google Scholar
7. Wise, S. ∙ Hall, J. ∙ Haywood, P. ...
   Paying for value: options for value-based payment reform in Australia
   Aust Health Rev. 2022; 46(2):129-133
   Google Scholar
8. Hegde, S. ∙ McKee, S. ∙ Cole, D. ...
   Experiences and learnings from developing and implementing a co-designed value-based healthcare framework within Victorian public oral health sector
   Aust Health Rev. 2024; 48:134-141
   Google Scholar
9. Chen, R. ∙ Ajwani, S. ∙ Christian, B. ...
   The development of a new oral health patient reported outcome measure: the New South Wales public dental services approach
   J Patient Rep Outcomes. 2024; 8(1):98
   Google Scholar
10. Innes, N.P.T. ∙ Schwendicke, F.
    Restorative thresholds for carious lesions: systematic review and meta-analysis
    J Dent Res. 2017; 96(5):501-508
    Google Scholar
11. Keys, T. ∙ Burrow, M.F. ∙ Rajan, S. ...
    Carious lesion management in children and adolescents by Australian dentists
    Aust Dent J. 2019; 64(3):282-292Google Scholar
12. Howell, M.D.
    Generative artificial intelligence, patient safety and healthcare quality: a review
    BMJ Qual Saf. 2024; 33(11):748-754
    Google Scholar
13. Ahmed, N. ∙ Abbasi, M.S. ∙ Zuberi, F. ...
    Artificial intelligence techniques: analysis, application, and outcome in dentistry—a systematic review
    Biomed Res Int. 2021; 2021, 9751564
    Google Scholar
14. Mahdi, S.S. ∙ Battineni, G. ∙ Khawaja, M. ...
    How does artificial intelligence impact digital healthcare initiatives? A review of AI applications in dental healthcare
    Int J Inf Manag Data Insights. 2023; 3(1), 100144
    Google Scholar
15. Chen, W. ∙ Dhawan, M. ∙ Liu, J. ...
    Mapping the use of artificial intelligence–based image analysis for clinical decision-making in dentistry: a scoping review
    Clin Exp Dent Res. 2024; 10(6), e70035
    Google Scholar
16. Schwendicke, F. ∙ Samek, W. ∙ Krois, J.
    Artificial intelligence in dentistry: chances and challenges
    J Dent Res. 2020; 99(7):769-774
    Google Scholar
17. Raj, R. ∙ Rajappa, R. ∙ Murthy, V. ...
    Observational diagnostics: the building block of AI-powered visual aid for dental practitioners
    Bioengineering (Basel). 2024; 12(1):9
    Google Scholar
18. Akoglu, H.
    User’s guide to sample size estimation in diagnostic accuracy studies
    Turk J Emerg Med. 2022; 22(4):177-185
    Google Scholar
19. Zhu, J. ∙ Chen, Z. ∙ Zhao, J. ...
    Artificial intelligence in the diagnosis of dental diseases on panoramic radiographs: a preliminary study
    BMC Oral Health. 2023; 23(1):358
    Google Scholar
20. Sounderajah, V. ∙ Ashrafian, H. ∙ Golub, R.M. ...
    Developing a reporting guideline for artificial intelligence-centred diagnostic test accuracy studies: the STARD-AI protocol
    BMJ Open. 2021; 11(6), e047709
    Google Scholar
21. Allen, M.S. ∙ Robson, D.A. ∙ Iliescu, D.
    Face validity: a critical but ignored component of scale construction in psychological assessment
    Eur J Psychol Assess. 2023; 39(3):153-156
    Google Scholar
22. Trevethan, R.
    Sensitivity, specificity, and predictive values: foundations, pliabilities, and pitfalls in research and practice
    Front Public Health. 2017; 5:307
    Google Scholar
23. Ismail, A.I. ∙ Pitts, N.B. ∙ Tellez, M. ...
    The international caries classification and management system (ICCMSTM) an example of a caries management pathway
    BMC Oral Health. 2015; 15(suppl 1):S9
    Google Scholar
24. Tonetti, M.S. ∙ Greenwell, H. ∙ Kornman, K.S.
    Staging and grading of periodontitis: framework and proposal of a new classification and case definition
    J Periodontol. 2018; 89(suppl 1):S159-S172
    Google Scholar
25. Australia’s Oral Health
    National study of adult Oral Health 2017-18. Australian Research Centre for Population
    https://www.adelaide.edu.au/arcpoh/national-study/report/Australias_Oral_Health_2017-18.pdf
    Date accessed: June 6, 2025
    Google Scholar
26. Access to Victoria’s public dental care services. Victorian Department of Health
    https://www.health.vic.gov.au/dental-health/access-to-victorias-public-dental-care-services
    Date accessed: June 6, 2025
    Google Scholar
27. Australia’s national Oral Health plan 2015-2024COAG Health Council
    https://www.health.gov.au/resources/publications/healthy-mouths-healthy-lives-australias-national-oral-health-plan-2015-2024?language=en
    Date accessed: June 6, 2025
    Google Scholar
28. Weisberg, E.M. ∙ Chu, L.C. ∙ Fishman, E.K.
    The first use of artificial intelligence (AI) in the ER: triage not diagnosisEmerg Radiol. 2020; 27(4):361-366
    Google Scholar
29. Pinero de Plaza, M.A. ∙ Lambrakis, K. ∙ Marmolejo-Ramos, F. ...
    Human-centred AI for emergency cardiac care: evaluating RAPIDx AI with PROLIFERATE_AI
    Int J Med Inform. 2025; 196, 105810
    Google Scholar
30. Yi, N. ∙ Baik, D. ∙ Baek, G.
    The effects of applying artificial intelligence to triage in the emergency department: a systematic review of prospective studies
    J Nurs Scholarsh. 2025; 57(1):105-118
    Google Scholar
31. Shafi, I. ∙ Fatima, A. ∙ Afzal, H. ...
    A comprehensive review of recent advances in artificial intelligence for dentistry e-health
    Diagnostics (Basel). 2023; 13(13):2196Google Scholar
32. Semerci, Z.M. ∙ Yardımcı, S.
    Empowering modern dentistry: the impact of artificial intelligence on Patient Care and clinical decision making
    Diagnostics (Basel). 2024; 14(12):1260Google Scholar
33. Almulihi, Q.A. ∙ Alquraini, A.A. ∙ Almulihi, F.A.A. ...
    Applications of artificial intelligence and machine learning in emergency medicine triage - a systematic review
    Med Arch. 2024; 78(3):198-206
    Google Scholar
34. Da’Costa, A. ∙ Teke, J. ∙ Origbo, J.E. ...
    AI-driven triage in emergency departments: a review of benefits, challenges, and future directions
    Int J Med Inform. 2025; 197, 105838
    Google Scholar
35. Thorat, V. ∙ Rao, P. ∙ Joshi, N. ...
    Role of artificial intelligence (AI) in patient education and communication in dentistry
    Cureus. 2024; 16(5), e59799
    Google Scholar
36. Lin, C.E. ∙ Nguyen, T.M. ∙ McGrath, R. ...|
    Dental Health Services Victoria value-based health care principles for oral health models of care
    J Public Health Dent. 2023; 83(3):325-328
    Google Scholar
37. Global strategy and action plan on oral health 2023-2030. World Health Organization
    https://iris.who.int/bitstream/handle/10665/376623/9789240090538-eng.pdf?sequence=1
    Date accessed: June 6, 2025
    Google Scholar
38. Christian, B. ∙ Amezdroz, E. ∙ Calache, H. ...
    Examiner calibration in caries detection for populations and settings where in vivo calibration is not practical
    Community Dent Health. 2017; 34(4):248-253
    Google Scholar
39. Kühnisch, J. ∙ Meyer, O. ∙ Hesenius, M. ...
    Caries detection on intraoral images using artificial intelligence
    J Dent Res. 2022; 101(2):158-165
    Google Scholar
40. Casas K, DiPede L, Toema S, Ogwo C.
    Assessing teledentistry versus in-person examinations to detect dental caries: a systematic review and meta-analysis.
    JDR Clin Transl Res. 2026:11(1):4-15.
    Google Scholar