Predicting early orthodontic treatment results and development of the dentofacial system without orthodontic treatment in 3-12-year-old children

Relevance. Prognosis of the dentofacial system (DS) development in children with dentofacial deformities (DD) is an urgent medical and social problem since the prognosis of the DS development will allow timely prescription and provision of adequate therapy, which will significantly reduce the risks of severe DD development in children. Machine learning methods have proven to be a reliable tool for predicting a patient's health status and evaluating the effectiveness of treatment methods. Therefore, it seems interesting to use this modern toolkit to build predictive models that allow us to assess the change in the condition of DS in children with DD after orthodontic treatment (OT) at different ages or without OT.

Purpose. The study aimed to build a set of predictive models for assessing the severity of the dentofacial system condition in 3.5-4-year-old children after and without orthodontic treatment.

Material and methods. The study used the data on the DS of children aged 3-5 years (n=50), 6-9 years (n=100), 10-12 years (n=100) and 13-17 years (n =100). The author's program was developed in Python 3.11 using the sklearn, pandas, and xgb libraries in Anaconda to build the predictive models.

Results. We developed nine models of the DS condition in children aged 3-12 years, three of which make predictions for the DS development after the OT (one - in the group of 3 – 5-year-old children, the second – in the group of 6 – 9-year-old children and the third – in the group of 10 – 12-year-olds) and six models predict the development of the DS without OT. Three out of 6 models predict DS development without OT at 3-5 years: the first makes a prediction of the DS condition for 6-9 year-olds; the second – for 10-12 year-olds; the third – for 13-17-year-olds. The accuracy of the models ranges from 82 to 86%. Two models out of 6 predict the DS development for children with DD who did not receive OT at 6-9 years old: one – at 10-12 years old, the second – at 13-17 years old. The accuracy of the models ranges from 92 to 97%. The sixth model makes predictions of the DS condition in children aged 13-17 years who did not receive OT at the age of 10-12 years. The accuracy of the model is 94%. In addition, we built three models that predict the DS condition in 3.5-4 years after the OT: the first model predicts for 3–5-year-old children; the second – for 6–9-year-olds; and the third - for children of 10–12 years old. The accuracy of the models ranges from 82 to 90%.

Conclusion. All obtained models will be used to build a web application for predicting the DS state severity in children after the orthodontic treatment and without the latter.

1. Leontev VK, Kiselnikova LP, editors. Pediatric Therapeutic Dentistry: national leadership. Moscow: GEOTAR-Media, 2017.952 (In Russ.). Available from: https://www.elibrary.ru/item.asp?id=19552884

2. Sergeeva MV, Kiseleva EA, Kiseleva KS, Kostritsin AG. The structure of dentofacial anomalies among children and adolescents of Kuzbass. Dental Forum. 2019;(2):19-20 (In Russ.). Available from: http://www.dental-forum.ru/index.php?menu_id=163

3. Olesov EE, Kaganova OS, Fazilova TA, Mirgazizov MZ, Ilyin AA, Shugailov IA. Dynamics of structure and severity of dentoalveolar anomalies on the background of early orthodontic treatment during occlusion. Journal of Clinical Practice. 2019;10(3):9-25 (In Russ.). doi: 10.17816/clinpract10319-25

4. Voskanyan AR, Ayupova FS. Regional and age-related features of the prevalence and structure of dental anomalies in children of the Krasnodar region. Stomatology for All / Int. Dental Review. 2021;(4):21-23 (In Russ.). doi: 10.35556/idr-2021-4(97)21-23

5. Gusev AV, Novitskiy RE, Ivshin AA, Alekseev AA. Machine learning based on laboratory data for disease prediction. FARMAKOEKONOMIKA. Modern Pharmacoeconomics and Pharmacoepidemiology. 2021;14(4):581-592 (In Russ.). doi: 10.17749/2070-4909/farmakoekonomika.2021.115

6. Nevzorova VA, Plekhova NG, Priseko LG, Chernenko IN, Bogdanov DYu, Mokshina MV, et al. Machine learning for predicting the outcomes and risks of cardiovascular diseases in patients with hypertension: results of ESSE-RF in the Primorsky Krai. Russian Journal of Cardiology. 2020;25(3):3751 (In Russ.). doi: 10.15829/1560-4071-2020-3-3751

7. Ishankulov TA, Danilov GV, Pitskhelauri DI, Titov OYu, Ogurtsova AA, Buklina SB, et al. Prediction of postoperative speech dysfunctions in neurosurgery based on cortico-cortical evoked potentials and machine learning technology. Sovremennye tehnologii v medicine. 2022;14(1):25 (In Russ.). doi: 10.17691/stm2022.14.1.03

8. Sinotova SL, Limanovskaya OV, Plaksina AN, Makutina VA. Software application for predicting the health status of a child born with the use of assisted reproductive technologies, according to the mother's anamnesis. Modeling, Optimization and Information Technology. 2021;9(3) (In Russ.). doi: 10.26102/2310-6018/2021.34.3.008

9. Sinotova SL, Solodushkin SI, Plaksina AN, Makutina VA. An intelligent clinical decision support system for predicting the outcome of an assisted reproductive technology protocol at various stages of its implementation. Modeling, Optimization and Information Technology. 2022;10(2) (In Russ.). doi: 10.26102/2310-6018/2022.37.2.009

10. Tokmakova SI, Bondarenko OV, Lunitsyna YuV, Zhukova ES, Mokrenko EV, Gaidarova TA, et al. The study of mouthwashes’ effect on oral microbiota. Pediatric dentistry and dental prophylaxis. 2023;23(1):4-14 (In Russ.). doi: 10.33925/1683-3031-2023-561

11. Dolgalev AA, Muraev AA, Lyakhov PA, Lyakhova UA, Choniashvili DZ, Zolotyaev KE, et al. Determining the optimal neural network structure for the development of decision support programmes in dental implantation. Medical alphabet. 2022;(34):54-64 (In Russ.). doi: 10.33667/2078-5631-2022-34-54-64

12. Ho TK. The Random Subspace Method for Constructing Decision Forests. IEEE Transactions on Pattern Analysis and Machine Intelligence. 1998:20(8):832–844. doi: 10.1109/34.709601. S2CID 206420153

13. Chen T, Guestrin C. XGBoost: a scalable tree boosting system. 2016. Proceedings of the 22 nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2016:785-794. doi: 10.1145/2939672.2939785

14. Cover ThM, Hart PE. Nearest neighbor pattern classification. IEEE Transactions on Information Theory. 1967:13(1):21–27. doi: 10.1109/TIT.1967.1053964. S2CID 5246200

15. Cortes C, Vapnik V. Support-vector networks. Machine Learning. 1995:20(3):273–297. doi: 10.1007/BF00994018. S2CID 206787478

16. Tolles Ju., Meurer WJ. Logistic Regression Relating Patient Characteristics to Outcomes. JAMA. 2016:316(5):533–534. doi: 10.1001/jama.2016.7653

17. Geurts P, Ernst D, Wehenkel L. Extremely randomized trees. Mach Learn. 2006;63:3–42. doi: 10.1007/s10994-006-6226-1

18. Wolpert DH. Stacked Generalization. Neural Networks. 1992;5(2):241–259. doi: 10.1016/s0893-6080(05)80023-1

19. Littlestone N., Warmuth M. The Weighted Majority Algorithm. Information and Computation. 1994;108(2):212–261. doi: 10.1006/inco.1994.1009

20. Roskoshenko VV. Overcoming the class imbalance in modeling the credit defaul. Finance and Credit. 2019;25(11):2534-2561 (In Russ.). doi: 10.24891/fc.25.11.2534