REVIEW ARTICLE

https://doi.org/10.5005/jp-journals-10063-0149
CODS Journal of Dentistry
Volume 15 | Issue 2 | Year 2023

Advanced Dentistry: Transforming Patient Care with Artificial Intelligence

Ayush Ahluwalia¹ https://orcid.org/0000-0001-8575-1118, Ayushi Gautam² https://orcid.org/0009-0001-9307-146X, Sahil S Thakar³ https://orcid.org/0000-0002-8686-5309

^1,2Department of Dentistry, Himachal Dental College, Sundernagar, Mandi, Himachal Pradesh, India

³Department of Public Health Dentistry, Himachal Dental College, Sundernagar, Mandi, Himachal Pradesh, India

Corresponding Author: Ayush Ahluwalia, Department of Dentistry, Himachal Dental College, Sundernagar, Mandi, Himachal Pradesh, India, Phone: +91 9816909604, e-mail: ayush.ahluwalia@yahoo.com

Received: 15 March 2024; Accepted: 10 April 2024; Published on: 17 May 2024

ABSTRACT

Artificial intelligence (AI) has already revolutionized the fundamental operations of various sectors today, yet its integration into the healthcare sector remains in its infancy. This integration holds tremendous potential to bring about transformative effects, particularly in the advancement of patient-centered care. In a critical sector like health care, where even the smallest decision can shape treatment outcomes significantly, the integration of AI reveals its complexity. While on the one hand, AI holds promise in aiding doctors to reach accurate diagnoses faster, on the contrary, it presents its own set of challenges. This review aims to outline the current issues in the healthcare sector, explore the role of AI in addressing these challenges, summarize recent advances in AI within dentistry, and examine the key challenges in this integration.

How to cite this article: Ahluwalia A, Gautam A, Thakar SS. Advanced Dentistry: Transforming Patient Care with Artificial Intelligence. CODS J Dent 2023;15(2):64–69.

Source of support: Nil

Conflict of interest: None

Keywords: Artificial intelligence, Black box, Dentistry, Heuristics, Partial least squares and artificial neural network, Patient care

INTRODUCTION

Artificial intelligence (AI) is widely regarded as having originated as a field at the Dartmouth Summer Research Project on AI in 1956.¹ However, it only evolved into a practical tool in recent years, driven by significant advancements in three critical components of modern AI technology—big data, computational power, and AI algorithms.² The majority of AI applications in the field of medicine and dentistry are centered around diagnosis, largely due to the availability of data. As AI continues to evolve, ongoing innovations are driving its application across various healthcare domains, including decision-making, treatment planning, disease prognosis, and prediction of treatment outcomes.

HEALTHCARE TODAY

The Robot Doctor

In the current healthcare landscape, there is a prevalent tendency among doctors to focus primarily on treating the disease rather than addressing the holistic needs of the patient.³ Providing healthcare demands a delicate equilibrium wherein medical requirements are met alongside recognizing and tending to the emotional well-being of the patient. Achieving this balance necessitates sufficient consultation time, and any compromise in this regard can significantly diminish the quality of care delivered.⁴ Shockingly, in India, primary care consultants spend an average of only 2 minutes with each patient.⁵ This perceived lack of empathy and rushed demeanor, rather than a deficiency in technical expertise, frequently gives rise to negative perceptions of doctors.⁶

The Inaccurate Doctor

Instances where evidence is obtained from inadequate examination, especially when conducted by stressed and burned-out doctors⁷—particularly in the aftermath of COVID-19⁸—or due to a lack of sound knowledge and clinical experience, as seen with students.⁹ Language barriers between patients and doctors, especially in multilingual countries like India, consultations with quacks,¹⁰ rare or unusual clinical presentations, or issues with malfunctioning medical equipment can lead to unnecessary patient anxiety, potentially resulting in the misdiagnosis of conditions leading to unwarranted treatments.¹¹

Doctors in Debt

In India, the path of obtaining a medical or dental degree reflects a landscape where private colleges outnumber government colleges, bringing about substantial fees for the majority of aspiring doctors who choose to train in these private institutions. Progressing from graduation to establishing a clinic or hospital adds additional substantial costs for land, infrastructure, equipment, and other resources. This journey frequently leads doctors to accumulate debts by the time they commence practicing. The ensuing financial strain puts doctors under considerable pressure to maximize returns from their practice to meet the demands of loan repayment.¹² Additionally, the oversupply of dentists in India has led to the private sector emerging as the main source of employment for the majority of dental graduates.¹³ This trend reinforces a financial incentive, given that out-of-pocket/fee-for-service stands as the predominant payment method¹⁴; it tends to encourage dentists to incline toward overtreatments.¹⁵

Rules of Thumb

The dual processing theory postulates that human cognition operates via two systems—system I, which is heuristic and relies on intuitive and experiential processes, and system II, which is analytical and uses a rule-based, deliberative method for reasoning and decision-making.¹⁶ In situations of uncertainty, our tendency to lean heavily on heuristic processing intensifies. This becomes particularly problematic in critical environments like healthcare settings, where rapid decision-making is crucial. As heuristics are quickly constructed from fragments of memory, they provide the benefit of speed and efficiency, often leading to mostly accurate decisions. However, there are instances when they can result in systematic cognitive errors.

Outlined below are several heuristics chosen to illustrate their relevance to clinical judgments and decision-making in dentistry, showcasing the theoretical implications of each bias.

Availability Heuristic

“Tendency to make likelihood predictions based on what can easily be remembered.”¹⁷

Considering the extensive range of diseases prevalent in the population, it’s unrealistic for any doctor to memorize each one. Thus, during diagnosis, they depend on a subset of easily recalled diseases, heightening the risk of misdiagnosis.

Representativeness Heuristic

“…the degree of correspondence between a sample and a population that makes us think an event is likely if it seems representative of a larger class.”¹⁸

For instance, when doctors exclusively seek out characteristic manifestations of a disease based on familiar patterns, they may disregard atypical variants.

Confirmation Bias

“...tendency to look for evidence “confirming” a diagnosis rather than disconfirming evidence to refute it.”¹⁹

For instance, an endodontist might attribute a patient’s sharp tooth pain solely to dental issues like a cavity or cracked tooth. Consequently, they may concentrate solely on supporting evidence, such as abnormalities on an intraoral periapical radiograph, overlooking evidence suggesting alternative causes like neuralgia.

Overconfidence Heuristic

“The overconfidence bias describes the fact that most people perceive themselves and their abilities as above average.”²⁰

For instance, inexperienced doctors, such as students, often overestimate their knowledge when making diagnoses, which can result in treatment errors.

HEALTHCARE TOMORROW

The human elements of health care—empathy, active listening, communication, and thorough physical examination—form the foundation of a meaningful patient–doctor relationship. Integration of AI presents a significant opportunity to reintroduce the essential “human touch” in this relationship. By leveraging natural language processing neural networks, manual entry into electronic health records can be eliminated during patient visits, enabling clinicians to prioritize patient needs,²¹ allowing them to invest more time in understanding their patients, enhancing communication, and cultivating empathy. This enables tailored care that addresses the individual needs of each patient.

Artificial intelligence systems can provide personalized, evidence-based recommendations to doctors and patients in real-time, overcoming cognitive biases. By offering second opinions, both doctors and patients can reevaluate the initial diagnosis, empowering patients to make informed decisions and thus reducing the risk of unnecessary treatments. To accomplish this, the system must analyze all relevant patient data, including chief complaints, physical findings, comorbidities, medications, allergies, and lab and imaging tests. It then cross-references this data with a vast clinical database to formulate a differential diagnosis, recommend additional testing, and propose a treatment strategy. While such comprehensive systems are not yet widely accessible, companies like “flow health” are actively developing end-to-end services aimed at addressing cognitive biases that can cloud medical judgment.²²

Artificial intelligence-powered second opinions have the potential to become more commonplace in health care, providing patients with the ability to upload their medical data like radiographs or laboratory tests to get opinions on potential diagnoses and treatment plans. A noteworthy instance in veterinary care involves OpenAI’s ChatGPT-4, an AI-powered chatbot, which played a pivotal role in saving the life of a dog suffering from a tick-borne disease. Dissatisfied with their veterinarian’s diagnosis, the dog’s owner turned to the chatbot, providing detailed information on the dog’s blood test results and overall condition. After analyzing the data, the AI proposed an underlying condition leading to anemia. Upon confirmation by another veterinarian, the diagnosis was validated, enabling prompt treatment and a swift recovery for the dog.²³

Advanced Dentistry

Advanced AI algorithms are driving remarkable progress in the fields of medicine and dentistry, effectively reducing human errors encountered in daily practice. Meanwhile, state-of-the-art AI-powered surgical robotic systems are surpassing human precision, further elevating the standards of minimally invasive surgical procedures.²⁴

The increasing count of the Food and Drug Administration-approved algorithms²⁵ presents a promising outlook for the integration of AI-driven solutions into routine medical and dental procedures. With rigorous testing and the assurance of safety, these algorithms hold the potential to establish a new standard in healthcare practices.

Listed below are several fields in dentistry selected to illustrate the current and future applications of AI.

ORTHODONTICS

Cephalometric Analysis

Cephalometric analysis, particularly landmarking on lateral cephalograms, is fundamental to orthodontic diagnosis, treatment planning, and assessing treatment outcomes. Traditional manual landmarking, while effective, is time-consuming, relies heavily on experience, and can yield inconsistent results among orthodontists. Commercial solutions like ORCA Dental AI’s CephX not only offer immediate cephalometric tracing and analysis but also provide instant airway analysis, making it an all-in-one solution.²⁶ Moreover, studies have highlighted the utilization of convolutional neural networks (CNNs) for automated landmarking on posteroanterior cephalograms, facilitating the assessment of mandibular deviation.²⁷

Treatment Planning

Several AI-driven software solutions, such as SmileCloud, analyze tooth shape and alignment data to devise optimal treatment plans for achieving an esthetically appealing smile. Furthermore, these programs serve as intelligent tools for planning and guiding simulated orthodontic surgeries.²⁸

PROSTHODONTICS

Implant Fabrication

Lerner et al.²⁹ integrated AI with CAD software to craft implant-supported monolithic zirconia crowns, which were then cemented to individual hybrid abutments. This approach not only lowers prosthetic surgery costs and minimizes errors but also significantly streamlines the workflow.

PERIODONTOLOGY

Diagnosis

Rana et al.³⁰ reported an advanced machine learning classifier designed to differentiate gingival inflammation from healthy gum tissue. The classifier produces a pixelwise segmentation of areas predicted to contain gingival inflammation. Employing intraoral images, the system can facilitate early detection of gingival inflammation in point-of-care settings, thereby aiding in the prevention of severe periodontal diseases.

Dental Implants

At times, dentists may face challenges in addressing patient issues related to dental implants when they’re unfamiliar with the specific implant system. Hence, there arises a necessity for a system capable of identifying a patient’s implant system with minimal data, independent of the dentist’s expertise and experience. Takahashi et al.³¹ effectively conducted a study utilizing deep learning (DL)-based object detection software to discern implant systems from panoramic radiographs. Cha et al.³² utilized a region-based CNN on periapical radiographs to assess bone loss accurately and aid in the diagnosis of the severity of peri-implantitis.

ENDODONTICS

Determination of Root Canal Morphology

For successful root canal treatment, understanding the root canal morphology of the tooth being treated is essential. Hiraiwa et al.³³ utilized AI on panoramic radiographs, achieving an 87% accuracy in diagnosing single or multiple distal roots in mandibular first molars. In a study by Lahoud et al.,³⁴ AI demonstrated comparable accuracy and greater efficiency than human evaluators in determining root canal morphology through three-dimensional (3D) tooth segmentation.

Commercial AI software companies like Diagnocat have emerged, aiding practitioners in analyzing patients’ cone-beam computed tomography (CBCT) to identify root canal morphology.³⁵ These software solutions can automatically segment teeth and generate 3D Standard Tessellation Language (STL) models, providing dentists with valuable resources for further analysis.

Determination of Working Length

Determining the apical limit of the root canal system is a pivotal stage in root canal treatment. Studies have highlighted that even a small discrepancy in working length measurements can significantly impact treatment success rates, with a millimeter loss potentially reducing success rates by 14% in teeth with apical periodontitis.³⁶ Using a cadaver model, Saghiri et al.³⁷ used AI to determine working length measurements. Their findings revealed a remarkable 100% match with actual measurements posttooth extraction.

ORAL AND MAXILLOFACIAL SURGERY

Orthognathic Surgery

Preoperative planning is paramount for achieving successful outcomes in orthognathic surgery. This includes comprehensive steps, such as virtual surgical planning, aided by advanced technologies like 3D imaging and printing, to enhance surgeons’ ability to visualize anatomical structures; anticipating complications, such as blood loss during the procedure, is also essential. To aid with this, Stehrer et al.³⁸ leveraged a machine-learning algorithm to predict blood loss during orthognathic surgery.

Third Molar Impaction

One crucial area where AI algorithms have shown promise is in optimizing the management of impacted teeth. A study by Vranckx et al.³⁹ presented and validated an AI-driven auto-angulation tool designed to automatically segment mandibular molars on panoramic radiographs and extract the molar orientations in order to predict the third molars’ eruption potential. Additionally, CNN models have been used to evaluate the difficulty of extracting mandibular third molars based on 2D panoramic radiographic images.⁴⁰

PEDODONTICS

Supernumerary Tooth Identification

The oversight of supernumerary teeth on panoramic radiographs often results from the screening abilities of inexperienced dental personnel, such as students. Ahn et al.⁴¹ employed a DL model to identify mesiodens in primary or mixed dentition, implying its potential to assist clinicians with limited clinical experience in achieving more precise diagnoses.

ORAL DIAGNOSIS AND RADIOLOGY

Oral Cancer

Automated DL methods have reached accuracy levels that are on par with oral cancer specialists for diagnosing oral cancer.⁴² Moreover, various studies have demonstrated the effectiveness of AI in distinguishing oral lesions. For instance, Wang et al.⁴³ utilized a partial least squares and artificial neural network classification algorithm to discriminate premalignant and malignant tissues from benign tissues.

Periapical Lesions

Detecting periapical lesions via radiographs presents a considerable challenge, even for experienced clinicians, due to the subjective interpretation involved. While CBCT offers higher accuracy in diagnosing periapical lesions than traditional radiographs,⁴⁴ there is a risk of clinicians, especially those not trained as oral radiologists, overlooking subtle density changes in CBCT volumes.⁴⁵ To overcome these challenges, Setzer et al. implemented a DL system, which demonstrated a 93% accuracy in lesion detection with a specificity of 88%.⁴⁶ Another study by Abdolali et al.⁴⁷ proposed a model based on asymmetry analysis to automatically segment radicular cysts, dentigerous cysts, and keratocysts. Moreover, multiple research groups have demonstrated the efficacy of AI-based detection in identifying periapical lesions, showing comparable or superior performance to that of experienced specialists.⁴⁸

Dental Caries

Swift and precise detection play crucial roles in administering timely prevention and treatment for patients with dental caries. Inaccurate diagnoses may result in the gradual progression of lesions through the enamel, dentin, and potentially to the pulp tissue, culminating in severe pain and eventual loss of tooth function. Lee et al.⁴⁹ proposed a caries detection model using a pretrained GoogLeNet Inception v3 CNN to identify caries in maxillary premolars and molars, achieving high diagnostic performance.

CHALLENGES

While many sectors are already moving toward the integration of AI into their operations, the healthcare industry lags behind. This discrepancy can be attributed, in part, to barriers within the healthcare sector, such as its dependence on patents and exclusivity, which raise questions concerning the protection of intellectual property without hindering progress. Discussed below are key challenges that must first be addressed before AI can be widely adopted by the healthcare sector.

Data Challenge

To date, much of AI research has focused on structured data, like well-organized images. However, in medical contexts, there’s an abundance of unstructured data lacking clear labels or annotations. Errors in labeling or annotating such data can result in nonsensical outputs from neural networks. Achieving accurate predictions necessitates a significant volume of data to overcome the challenges of distinguishing signals from noise.

In his book, American cardiologist and author Eric Topol recounts a case involving a patient with an uncommon manifestation of coronary artery blockage, where an AI algorithm might overlook stenting as a viable treatment option due to the scarcity of supporting evidence.⁵⁰ This instance highlights the potential pitfalls of relying solely on AI for medical decision-making, particularly in cases where data might be scarce or nonexistent, such as in the context of rare or unusual disease presentations.

Machine Bias

Another substantial challenge that has come to the forefront is the issue of machine bias. A notable example is the 2015 incident involving Google Photos AI, where black individuals were inaccurately labeled as gorillas.⁵¹

Unfortunately, bias is already deeply ingrained in the medical research system, with clinical studies often lacking representative samples, particularly from racial to ethnic minorities.⁵² This underrepresentation results in biased data, hindering accurate diagnoses and perpetuating healthcare disparities within these populations.

The Black Box

Understanding how AI systems make decisions or predictions is challenging due to their opacity, often referred to as the black box phenomenon. This challenge is particularly pronounced when employing DL algorithms with complex, nonlinear computations across multiple layers. In medicine, we already encounter black boxes. Take lithium, for instance. While its precise biochemical mechanism in influencing mood remains elusive, the drug is still sanctioned for treating bipolar disorder. Similarly, aspirin, the most extensively used medication in history, operated without a fully understood mechanism for over 70 years.⁵³ While patients may prioritize treatment outcomes over understanding mechanisms, it is essential to refrain from embracing a black box approach in critical fields like healthcare.

Liability Challenge

Traditionally, accountability for medical malpractice has rested with doctors, hospitals, or healthcare systems. However, with the introduction of AI in health care, questions about accountability for decisions made by machines have arisen. The black box nature of algorithms makes it challenging, or even impossible, to explain how AI arrived at specific decisions, thereby complicating the assignment of accountability for any adverse outcomes.

Hacking

In the era of hacking, all operations using AI need to be concerned about the threat of bad data corrupting their system, as AI’s susceptibility to hacking, due to its unique training process, renders it more vulnerable to attacks compared to other software.⁵⁴

Consider the possible scenario where a dental radiograph assessment algorithm is hacked, resulting in mislabeling of caries progression. The repercussions of this error could impact 100 or even 1,000 patients, leading to unnecessary root canal therapy or extractions. In contrast, the same mistake made by a clinician would affect only a handful of patients at most. The stakes are even higher in critical situations, such as with a diabetes algorithm recommending insulin doses based on various factors like glucose levels, sleep, and physical activity. A hack in this algorithm could result in inaccurate insulin dosage recommendations, potentially causing hypoglycemic coma or even fatalities for numerous patients.

Trust Problem

Trust in healthcare systems is typically built upon anticipated traits, such as transparency, controllability, and reliability. Specifically, reliability, which ensures tasks are executed predictably and consistently, is a significant concern, given the potential for variations in output when algorithms encounter new data.⁵⁵ Furthermore, improving trust in AI may come at the cost of eroding a deeper sense of moral trust. If we equate trust solely with reliability and accuracy, as AI performance improves, people may begin to trust machines more than human doctors, whose technical accuracy might end up being inferior to machines.⁵⁶

CONCLUSION

The integration of AI into dentistry and medicine remains in its infancy. Ongoing research across various dental specialties is actively shaping the trajectory of oral healthcare. Presently, data shows the adoption of AI by healthcare professionals to enhance medical imaging, aid in diagnosis, optimize treatment planning, and beyond. The healthcare landscape is on the brink of a transformative shift driven by AI, with several pioneering AI-powered solutions already emerging and many more on the horizon. At this pivotal moment, integrating AI into undergraduate and postgraduate curricula becomes imperative. Equipping future dental practitioners with early AI literacy will foster greater adoption of such AI-based tools in everyday practice, enabling patients to leverage the advantages they provide.

However, it is crucial to embed in the curriculum an understanding of the rationale behind AI and the principles of machine learning, including the concepts of model training, validation, and testing, along with the myriad challenges inherent in deploying AI in healthcare. This ensures a comprehensive understanding of AI’s limitations and prevents blind reliance on such programs. Nevertheless, as this integration matures, AI programs will evolve to become increasingly reliable for widespread adoption. This opens avenues for new regulations aimed at enhancing safety and reliability. A concise overview of the advantages and disadvantages of AI can be outlined as follows:

Advantages

Restoring the crucial doctor–patient relationship.
Minimizing diagnostic errors caused by doctor bias.
Relieving doctors from mundane administrative tasks.
Allowing doctors to prioritize patients’ holistic needs.
Empowering patients to make informed treatment decisions.
Enabling advanced diagnosis and treatment planning for improved outcomes.
Saving valuable time by facilitating swift yet accurate diagnoses.

Disadvantages

Vulnerability to hacking and glitches.
Potential liability concerns for mistreatment.
Black box nature of AI.
Ethical issues since personal medical data is used for training.
Lack of empathy and nuanced understanding.
Underrepresentation of minority populations in training data.
Insufficient availability of structured data.
High cost associated with deploying and setting up AI solutions.

ORCID

Ayush Ahluwalia https://orcid.org/0000-0001-8575-1118

Ayushi Gautam https://orcid.org/0009-0001-9307-146X

Sahil S Thakar https://orcid.org/0000-0002-8686-5309

REFERENCES

1. Anyoha R. The History of Artificial Intelligence. Science in the News. Updated August 28, 2017. Accessed February 22, 2024. https://sitn.hms.harvard.edu/flash/2017/history-artificial-intelligence/

2. Ding H, Wu J, Zhao W, et al. Artificial intelligence in dentistry—a review. Front Dent Med 2023;4:1085251. DOI: 10.3389/fdmed.2023.1085251

3. Davidson CS. The caring physician: the life of Dr. Francis W. Peabody. N Engl J Med 1993;328:817–818. DOI: 10.1056/NEJM199303183281123

4. Mercer SW, Hasegawa H, Reilly D, et al. Length of consultations. Time and stress are limiting holistic care in Scotland. BMJ 2002;325(7374):1241. PMID: 12455103.

5. Irving G, Neves AL, Dambha-Miller H, et al. International variations in primary care physician consultation time: a systematic review of 67 countries. BMJ Open 2017;7(10):e017902. DOI: 10.1136/bmjopen-2017-017902

6. Singletary B, Patel N, Heslin M. Patient perceptions about their physician in 2 words: The good, the bad, and the ugly. JAMA Surg 2017;152(12):1169–1170. DOI: 10.1001/jamasurg.2017.3851

7. Kulkarni S, Dagli N, Duraiswamy P, et al. Stress and professional burnout among newly graduated dentists. J Int Soc Prev Community Dent 2016;6(6):535–541. DOI: 10.4103/2231-0762.195509

8. Kane L. ’I cry but no one cares’: Physician burnout & depression report 2023. Medscape. Updated January 20, 2023. Accessed February 22, 2024. https://www.staging.medscape.com/slideshow/2023-lifestyle-burnout-6016058?reg=1#29

9. Moreno T, Sanz JL, Melo M, et al. Overtreatment in restorative dentistry: Decision making by last-year dental students. Int J Environ Res Public Health 2021;18(23):12585. DOI: 10.3390/ijerph182312585

10. Hans MK, Hans R, Nagpal A. Quackery: a major loophole in dental practice in India. J Clin Diagn Res 2014;8(2):283. DOI: 10.7860/JCDR/2014/6820.4081

11. von Eckardstein KL, Keil M, Rohde V. Unnecessary dental procedures as a consequence of trigeminal neuralgia. Neurosurg Rev 2015;38(2):355–360. DOI: 10.1007/s10143-014-0591-1

12. Marathe S, Hunter BM, Chakravarthi I, et al. The Impacts of corporatisation of healthcare on medical practice and professionals in Maharashtra, India. BMJ Global Health 2020;5(2):e002026. DOI: 10.1136/bmjgh-2019-002026

13. Mohan M, Ravindran TKS. Unemployment and vulnerable financial situation among recent dental graduates of Kerala, India - results from a cross-sectional study. J Global Oral Health 2019;1(1):49–57. DOI: 10.25259/JGOH-12-2018

14. Ramanarayanan V, Janakiram C, Joseph J, et al. Oral health care system analysis: a case study from India. J Family Med Prim Care 2020;9(4):1950–1957. DOI: 10.4103/jfmpc.jfmpc_1191_19

15. Holden A. Do I really need this crown? Dentists admit feeling pressured to offer unnecessary treatments. The Conversation. Updated November 2, 2020. Accessed February 22, 2024. https://theconversation.com/do-i-really-need-this-crown-dentists-admit-feeling-pressured-to-offer-unnecessary-treatments-148638

16. Djulbegovic B, Hozo I, Beckstead J, et al. Dual processing model of medical decision-making. BMC Med Inform Decis Mak 2012;12:94. DOI: 10.1186/1472-6947-12-94

17. Whelehan DF, Conlon KC, Ridgway PF. Medicine and heuristics: cognitive biases and medical decision-making. Ir J Med Sci 2020;189(4):1477–1484. DOI: 10.1007/s11845-020-02235-1

18. Kahneman D, Tversky A. Prospect theory: an analysis of decision under risk. Econometrica 1979;47(2):263–291. DOI: 10.2307/1914185

19. Hussain A, Oestreicher J. Clinical decision-making: heuristics and cognitive biases for the ophthalmologist. Surv Ophthalmol 2017;63(1):119–124. DOI: 10.1016/j.survophthal.2017.08.007

20. Rehder O, Noack MJ, Zirkel C, et al. Recognition and prevention of cognitive biases and judgment errors in diagnostics and dental therapy. Dtsch Zahnärztl Z Int 2021;3:231–237. DOI: 10.3238/dzz-int.2021.0028

21. Nagy M, Sisk B. How will artificial intelligence affect patient-clinician relationships? AMA J Ethics 2020;22(5):E395–E400. DOI: 10.1001/amajethics.2020.395

22. Rowley R. AI as a way to overcome cognitive bias in physicians. CIO. Updated June 27, Accessed February 22, 2024. https://www.cio.com/article/230275/ai-as-a-way-to-overcome-cognitive-bias-in-physicians.html

23. Tsai A. How chatGPT4 save a dog’s life. Medium. Updated March 29, 2023. Accessed February 22, 2024. https://medium.com/@albertfetsai/how-chatgpt4-save-a-dogs-life-a8c67561f01f

24. Gordon C. AI innovations in healthcare. Forbes. Updated September 30, 2021. Accessed February 22, 2024. https://www.forbes.com/sites/cindygordon/2021/09/30/ai-innovations-in-healthcare/?sh=47c89b3d36ed

25. How artificial intelligence is accelerating innovation in healthcare. Goldman Sachs. Research. Updated April 26, 2023. Accessed February 22, 2024. https://www.goldmansachs.com/intelligence/pages/how-artificial-intelligence-is-accelerating-innovation-in-healthcare.html

26. Orthodontic Solutions. ORCA Dental AI. c2024. Accessed February 22, 2024. https://www.orca-ai.com/solutions/orthodontics/

27. Takeda S, Mine Y, Yoshimi Y, et al. Landmark annotation and mandibular lateral deviation analysis of posteroanterior cephalograms using a convolutional neural network. J Dent Sci 2021;16(3):957–963. DOI: 10.1016/j.jds.2020.10.012

28. Patcas R, Bernini DAJ, Volokitin A, et al. Applying artificial intelligence to assess the impact of orthognathic treatment on facial attractiveness and estimated age. Int J Oral Maxillofac Surg 2019;48(1):77–83. DOI: 10.1016/j.ijom.2018.07.010

29. Lerner H, Mouhyi J, Admakin O, et al. Artificial intelligence in fixed implant prosthodontics: a retrospective study of 106 implant-supported monolithic zirconia crowns inserted in the posterior jaws of 90 patients. BMC Oral Health 2020;20(1):80. DOI: 10.1186/s12903-020-1062-4

30. Rana A, Yauney G, Wong LC, et al. Automated segmentation of gingival diseases from oral images. IEEE Healthcare Innov Point Care Technol (HI-POCT) 2017:144–147. DOI: 10.1109/HIC.2017.8227605

31. Takahashi T, Nozaki K, Gonda T, et al. Identification of dental implants using deep learning-pilot study. Int J Implant Dent 2020;6(1):53. DOI: 10.1186/s40729-020-00250-6

32. Cha JY, Yoon HI, Yeo IS, et al. Peri-implant bone loss measurement using a region-based convolutional neural network on dental periapical radiographs. J Clin Med 2021;10(5):1009. DOI: 10.3390/jcm10051009

33. Hiraiwa T, Ariji Y, Fukuda M, et al. A deep-learning artificial intelligence system for assessment of root morphology of the mandibular first molar on panoramic radiography. Dentomaxillofac Radiol 2019;48(3):20180218. DOI: 10.1259/dmfr.20180218

34. Lahoud P, EzEldeen M, Beznik T, et al. Artificial intelligence for fast and accurate 3-dimensional tooth segmentation on cone-beam computed tomography. J Endod 2021;47(5):827–835. DOI: 10.1016/j.joen.2020.12.020

35. Study root canal morphology and anatomy with Diagnocat. Diagnocat. Updated March 19, 2024. Accessed February 22, 2024. https://diagnocat.com/camx/case-studies/study-root-canal-morphology-and-anatomy-with-diagnocat/

36. Chugal NM, Clive JM, Spångberg LS. Endodontic infection: some biologic and treatment factors associated with outcome. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2003;96(1):81–90. DOI: 10.1016/s1079-2104(02)91703-8

37. Saghiri MA, Garcia-Godoy F, Gutmann JL, et al. The reliability of artificial neural network in locating minor apical foramen: a cadaver study. J Endod 2012;38(8):1130–1134. DOI: 10.1016/j.joen.2012.05.004

38. Stehrer R, Hingsammer L, Staudigl C, et al. Machine learning based prediction of perioperative blood loss in orthognathic surgery. J Craniomaxillofac Surg 2019;47(11):1676–1681. DOI: 10.1016/j.jcms.2019.08.005

39. Vranckx M, Van Gerven A, Willems H, et al. Artificial intelligence (AI)-driven molar angulation measurements to predict third molar eruption on panoramic radiographs. Int J Environ Res Public Health 2020;17(10):3716. DOI: 10.3390/ijerph17103716

40. Yoo JH, Yeom HG, Shin W, et al. Deep learning based prediction of extraction difficulty for mandibular third molars. Sci Rep 2021;11(1):1954. DOI: 10.1038/s41598-021-81449-4

41. Ahn Y, Hwang JJ, Jung YH, et al. Automated mesiodens classification system using deep learning on panoramic radiographs of children. Diagnostics 2021;11(8):1477. DOI: 10.3390/diagnostics11081477

42. Fu Q, Chen Y, Li Z, et al. A deep learning algorithm for detection of oral cavity squamous cell carcinoma from photographic images: a retrospective study. E Clin Med 2020;27:100558. DOI: 10.1016/j.eclinm.2020.100558

43. Wang CY, Tsai T, Chen HM, et al. PLS-ANN based classification model for oral submucous fibrosis and oral carcinogenesis. Lasers Surg Med 2003;32(4):318–326. DOI: 10.1002/lsm.10153

44. Patel S, Dawood A, Mannocci F, et al. Detection of periapical bone defects in human jaws using cone-beam computed tomography and intraoral radiography. Int Endod J 2009;42(6):507–515. DOI: 10.1111/j.1365-2591.2008.01538.x

45. Aminoshariae A, Kulild J, Nagendrababu V, et al. Artificial intelligence in endodontics: current applications and future directions. J Endod 2021;47(9):1352–1357. DOI: 10.1016/j.joen.2021.06.003

46. Setzer FC, Shi KJ, Zhang Z, et al. Artificial intelligence for the computer-aided detection of periapical lesions in cone-beam computed tomographic images. J Endod 2020;46(7):987–993. DOI: 10.1016/j.joen.2020.03.025

47. Abdolali F, Zoroofi RA, Otake Y, et al. Automatic segmentation of maxillofacial cysts in cone-beam CT images. Comput Biol Med 2016;72:108–119. DOI: 10.1016/j.compbiomed.2016.03.014

48. Ekert T, Krois J, Meinhold L, et al. Deep learning for the radiographic detection of apical lesions. J Endod 2019;45(7):917–922.e5. DOI: 10.1016/j.joen.2019.03.016

49. Lee JH, Kim DH, Jeong SN, et al. Detection and diagnosis of dental caries using a deep learning-based convolutional neural network algorithm. J Dent 2018;77:106–111. DOI: 10.1016/j.jdent.2018.07.015

50. Topol E. Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again. New York: Basic Books; 2019.

51. Barr A. Google mistakenly tags black people as ‘gorillas,’ showing limits of algorithms. Wall Street J. Updated July 1, 2015. Accessed February 22, 2024. https://www.wsj.com/articles/BL-DGB-42522

52. Ma MA, Gutiérrez DE, Frausto JM, et al. Minority representation in clinical trials in the United States: trends over the past 25 years. Mayo Clin Proc 2021;96(1):264–266. DOI: 10.1016/j.mayocp.2020.10.027

53. Brouillette M. Deep learning is a black box, but health care won’t mind. MIT Technol Rev. Updated April 27, 2017. Accessed February 22, 2024. https://www.technologyreview.com/2017/04/27/242905/deep-learning-is-a-black-box-but-health-care-wont-mind/

54. Lohn A. Hacking poses risks for artificial intelligence. Center Sec Emerg Technol. Updated March 1, 2022. Accessed February 22, 2024. https://cset.georgetown.edu/article/hacking-poses-risks-for-artificial-intelligence/

55. Parikh RB, Obermeyer Z, Navathe AS. Regulation of predictive analytics in medicine. Science 2019;363(6429):810–812. DOI: 10.1126/science.aaw0029

56. DeCamp M, Tilburt JC. Why we cannot trust artificial intelligence in medicine. Lancet Digit Health 2019;1(8):e390. DOI: 10.1016/S2589-7500(19)30197-9

________________________
© The Author(s). 2023 Open Access. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by-nc/4.0/), which permits unrestricted use, distribution, and non-commercial reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.