1,000 digitized cases: How a Kazakhstani orthodontist launches an AI startup for complex dental diagnostics
Digitalization, automation, and neural networks have also made significant inroads into the field of medicine. For example, modern orthodontics has long transcended the boundaries of simple aesthetic tooth alignment, evolving into a high-tech discipline of comprehensive engineering of the dentofacial system.
Astana-based orthodontist Bagdad Nurmukhambetova, Master of Medicine and Head of the Orthodontic Department, shares insights into a new medical IT startup designed to bridge the gap between AI diagnostics and advanced gnathology. She explains how a structured database of over 1,000 digitized patient cases will automate routine orthodontic tasks, calculate bone-loss risks, and establish a new standardized, error-free protocol for treating complex temporomandibular joint (TMJ) disorders across the region.
The foundation of gnathological health: why do we need AI here?
One of the most pressing challenges in contemporary clinical practice is the absence of a unified, standardized treatment protocol, a problem that is particularly evident among early-career orthodontists across the CIS countries, says Bagdad Nurmukhambetova, Master of Medicine and Head of the Orthodontic Department.
— At the beginning of their professional careers, orthodontists frequently encounter conceptual uncertainty that leads to significant diagnostic and treatment-planning errors. The principal hidden threat to patient health is the development of premature occlusal contacts and occlusal interferences — early tooth contacts that force the mandible into an abnormal functional position. This fundamental and inseparable relationship between occlusion and the temporomandibular joint results in chronic TMJ overload, deformation of the articular disc, and muscular spasms.
In this context, artificial intelligence (AI) is not merely a tribute to the digital era but an essential adjunctive tool for precision diagnostics, capable of accurately localizing pathology and establishing a mathematically validated foundation for treatment from the very outset, she says.
«Today, a significant portion of a clinician’s time is spent not on treatment itself, but on analyzing large volumes of diagnostic data»
The evolution of dental diagnostics has driven the industry through an unprecedented digital transformation. Intraoral 3D scanning has replaced conventional plaster impressions, cone beam computed tomography (CBCT) has enabled visualization of previously inaccessible anatomical structures, and manual virtual design of clear aligners and occlusal splints has become the industry standard. However, this rapid technological progress has also exposed a new vulnerability — the enormous cognitive burden placed on the treating clinician, says Bagdad Nurmukhambetova, orthodontist.
— Today, a significant portion of a clinician’s time is spent not on treatment itself, but on analyzing large volumes of diagnostic data. My goal is to create a tool that helps clinicians analyze diagnostic studies more efficiently, identify potential risks, suggest treatment options, and automate routine stages of digital treatment planning. At the same time, the final clinical decision must always remain with the doctor.
As the orthodontist points out, modern treatment planning requires the analysis of enormous volumes of data, consuming a substantial portion of the clinician’s time. The manual review of hundreds of CBCT slices, correlation of these findings with intraoral 3D scans, cephalometric analysis of lateral cephalometric radiographs, and evaluation of photographic records inevitably carries the risk of subjective human error associated with fatigue and reduced concentration during repetitive diagnostic tasks.
Artificial intelligence as a systemic solution
The response to these clinical challenges has been the development of a new medical IT startup in Astana — an intelligent assistant designed specifically for orthodontists. The foundation for training the project’s neural networks is Big Data in gnathology: a structured clinical database comprising approximately 1,000 digitized patient cases.
— We are gradually building a structured clinical database. It includes CBCT scans, intraoral 3D scans, photographic records, lateral cephalometric radiographs, results of gnathological examinations, treatment plans, and their clinical outcomes. In the future, comprehensive datasets of this kind may serve as the foundation for developing intelligent clinical decision-support algorithms.
According to the founder, the functionality of the platform under development is focused on automating the most labor-intensive routine tasks. The system is designed to detect hidden pathologies, generate step-by-step treatment scenarios, and perform high-precision AI-driven modeling of customized occlusal splints and clear aligners, while leaving the final clinical decision entirely in the hands of the treating orthodontist.
— Modeling an occlusal splint for the treatment of temporomandibular joint (TMJ) dysfunction is an exceptionally complex and highly precise process. It is impossible to completely replace the clinician at this stage. However, an AI algorithm is capable of integrating and analyzing vast amounts of data — including MRI, CBCT, axiography findings, and clinical parameters — and comparing them with a growing repository of successful clinical cases to help clinicians determine the optimal physiological position of the mandibular condyle for each individual patient with the highest possible degree of precision.
The orthodontist as the principal architect: where AI steps back
Nurmukhambetova maintains that the purpose of implementing such software solutions is not to replace clinicians with artificial intelligence.
— The most complex decisions will always require clinical reasoning, professional experience, communication with the patient, and consideration of numerous individual factors. I see AI as an intelligent assistant that enables clinicians to analyze information more efficiently, reduce the likelihood of errors, and devote more time to their patients.
She is convinced that the orthodontist will continue to serve as «the principal architect of the patient’s smile». The most challenging clinical decisions—including interdisciplinary treatment planning, preparation for orthognathic surgery, patient communication, and the exercise of clinical judgment and empathy—will always rely on human expertise. The synergy between clinician and machine will make it possible to delegate computational and repetitive tasks to artificial intelligence, substantially reduce the incidence of clinical errors, and establish highly accurate gnathological treatment as a safe standard of care accessible to any dental practice. Only in the distant future might technology advance to the point where routine treatment procedures can be fully automated by robotic systems, leaving clinicians to focus on the most complex and controversial cases.
— The project is currently at the stage of developing the product architecture. In parallel, we are designing the platform concept, structuring the clinical database, and recruiting specialists who will be able to implement the project from a technical perspective.
The next phase of development will focus on expanding the technical team through the recruitment of highly specialized IT professionals, including neural network engineers, computer vision specialists, and data scientists, alongside leading clinical experts who will contribute to the training, validation, and calibration of the algorithms.