LOGISTIC SCREENING MODELS FOR PREDICTING THE DEVELOPMENT OF CERVICAL LESIONS OF TEETH
DOI:
https://doi.org/10.35220/2523-420X/2026.3.7Keywords:
caries, erosion, oral fluid, cracks, prevention, predictorsAbstract
To develop logistic screening models for predicting the risk of cervical lesions in young patients. Research methods. There were examined and surveyed 272 patients (average age 24.3±6.9 years). Using the method of logistic regression analysis, independent predictors of the development of a wedge-shaped defect (WSD), cervical caries (CC) and enamel erosion (E) of teeth were identified. For prediction modeling, logistic regression was chosen, implemented in the Python programming language using scikit-learn libraries. Due to different prevalence of cervical lesions, class balancing (class_ weight=”balanced”) was used. To evaluate the models, the Receiver Operating Characteristic curve-Area under the curve (ROC-AUC) was calculated. Scientific novelty. The leading independent predictors of WSD were gingival recession and excessive pressure on a toothbrush, CC – the habit of drinking tea with lemon and a pathological bite, enamel E – the presence of enamel cracks on the vestibular surface of the teeth and occlusal attrition. The analysis of the effectiveness of the logistic model for predicting WSD of teeth showed maximum sensitivity (recall=1.00) and moderate, but insufficient discriminatory ability (ROCAUC= 0.677) that corresponds to a satisfactory level of class separation. The model for predicting CC of teeth also demonstrated high recall=0.98 and discriminatory ability (ROC-AUC=0.651) that is acceptable for early screening tools aimed at minimizing false negative results. The model for predicting enamel E of teeth showed recall=0.87 and ROC-AUC=0.604 that indicates insufficient discriminatory ability in conditions of a limited subsample. Obtaining some predictors with negative coefficients in the model for predicting CC is due to the limitations of the linear logistic regression approach and the characteristics of the studied sample. Conclusions. The developed logistic models can be used by a dentist to assess the individual risk of WSD, CC and enamel E of teeth in young patients and determine the most significant predictors with their subsequent correction
References
Li, Y., Xiang, Y., Ren, H., Zhang, C., Hu, Z., Leng, W., & Xia, L. (2024). Association between periodontitis and dental caries: a systematic review and meta-analysis. Clin Oral Investig, 28 (6), 306. https://doi.org/ 10.1007/s00784-024-05687-2
Alsubai, S. (2023). Enhancing prediction of tooth caries using significant features and multi-model classifier. PeerJ Comput Sci, 9, e1631. https://doi.org/ 10.7717/peerj-cs.1631
Kong, W., Ma, H., Qiao, F., Xiao, M., Wang, L.,Zhou, L., Chen, Y., Liu, J., Wang, Y., & Wu, L. (2024). Risk factors for noncarious cervical lesions: A case-control study. J Oral Rehabil, 51 (9), 1684–1691. https://doi.org/10.1111/joor.13772
Ogwo, C., Brown, G., Warren, J., Caplan, D., & Levy, S. (2024). Predicting dental caries outcomes in young adults using machine learning approach. BMC Oral Health, 24 (1), 529. https://doi.org/ 10.1186/s12903-024-04294-7
Hayes, M., Da Mata, C., McKenna, G., Burke, F.M., & Allen, P.F. (2017). Evaluation of the Cariogram for root caries prediction. J Dent, 62, 25–30. https://doi.org/10.1016/j.jdent.2017.04.010
Havsed, K., Hänsel Petersson, G., Isberg, P.E., Pigg, M., Svensäter, G., Foresight Research Consortium; & Rohlin, M. (2023). Multivariable prediction models of caries increment: a systematic review and critical appraisal. Syst Rev, 12 (1), 202. https://doi.org/ 10.1186/s13643-023-02298-y
Levrini, L., Di Benedetto, G., & Raspanti, M. (2014). Dental wear: A scanning electron microscope study. BioMed Research International, 10, Article ID 340425. https://doi.org/ 10.1155/ 2014/ 340425
Zabolotna, I.I., Bogdanova, T.L., Potapov, Yu.O., & Genzytska, O.S. (2023). Сorrelation of dentine hypersensitivity (DH) with manifestations of psycho-emotional stress, its features in patients with cervical teeth pathology. Protet Stomatol, 73 (2), 97–110. https://doi.org/ 10.5114/ps/ 168064
Chavance, M., Escolano, S., Romon, M., Basdevant, A., de Lauzon-Guillain, B., & Charles, M.A. (2010). Latent variables and structural equation models for longitudinal relationships: an illustration in nutritional epidemiology. BMC Med Res Methodol, 10, 37. https://doi.org/ 10.1186/ 1471-2288-10-37
Deng, K., Zonta, F., Yang, H., Pelekos, G., & Tonetti, M.S. (2024). Development of a machine learning multiclass screening tool for periodontal health status based on non-clinical parameters and salivary biomarkers. J Clin Periodontol, 51 (12), 1547–1560. https://doi.org/10.1111/jcpe.13856
Ramsay, D.S., Marilynn Rothen, M., Scott, J., & Cunha-Cruz, J. (2015). Tooth wear and the role of salivary measures in general practice patients. Clin Oral Investig, 19 (1), 85–95. https://doi.org/ 10.1007/s00784-014-1223-4
Gao, X., Jiang, S., Koh, D., & Hsu, C.Y. (2016). Salivary biomarkers for dental caries. Periodontol 2000, 70 (1), 128–141. https://doi.org/ 10.1111/prd.12100
Ibrahim, M.O. (2021). Dietary habits, eating practices and DMFT index among adults attending dental clinics in Amman, Jordan. Nutr Clín Diet Hosp, 41 (4), 48–54. https://doi.org/ 10.12873/ 414mohammed
Southward, K. (2015). A hypothetical role for vitamin K2 in the endocrine and exocrine aspects of dental caries. Med Hypotheses, 84 (3), 276–280. https://doi.org/10.1016/j.mehy.2015.01.011
Dioguardi, M., Polverari, D., Spirito. F., Iacovelli, G., Sovereto. D., Laneve, E., Caloro, G.A., Ballini, A., & Lo Muzio, L. (2023). Introspection of the Etiopathological Mechanisms Underlying Noncarious Cervical Lesions: Analysis of the Different Theories and Their Impact on the Mineralized Structures of the Tooth. Int J Dent, 2023, 8838314. https://doi.org/ 10.1155/ 2023/ 8838314
Femiano, F., Femiano, R., Femiano, L., Festa, V.M., Rullo, R., & Perillo, L. (2015). Noncarious cervical lesions: correlation between abfraction and wear facets in permanent dentition. Open Journal of Stomatology, 5, 152–157. https://doi.org/ 10.4236/ojst.2015.56021
