Sign in Subscribe
Machine learning

Machine Learning Predicts Treatment Outcomes for Skeletal Class III Malocclusion

Strategic Insights

— 1 min read

Researchers at the Hospital of Stomatology have developed and validated a predictive nomogram and deep learning model to determine the best treatment approach for patients with skeletal Class III malocclusion. The study, published in the BMC Oral Health journal, utilized computed tomography (CBCT) images of 313 patients to identify key predictors for selecting either an orthodontic-first approach (OFA) or a surgery-first approach (SFA). The team found that a Simple-CNN model achieved the best performance in automated classification, with a sensitivity of 0.979, specificity of 0.800, and accuracy of 0.909.

Key Takeaways:

  • The research team developed a predictive nomogram and a deep learning model to determine the treatment modality for patients with skeletal Class III malocclusion using CBCT images.
  • The nomogram included key predictors such as maxillary crowding, maxillary arch form asymmetry index (AI), mandibular AI, and U1-SN angle, which were found to be significant in predicting SFA selection.
  • The Simple-CNN model was found to be the most accurate in predicting treatment outcomes, with a sensitivity of 0.979 and specificity of 0.800.
  • The study demonstrated the effectiveness of the Simple-CNN model for automated classification of treatment approaches.
  • The researchers identified that patients with maxillary crowding, maxillary arch form asymmetry, and mandibular AI were more likely to require SFA.
  • The study's findings have implications for clinicians to make more informed decisions regarding treatment approaches for patients with skeletal Class III malocclusion.

Statistics:

  • 313 patients with skeletal Class III malocclusion were included in the study.
  • 191 patients underwent the orthodontic-first approach (OFA), while 122 patients underwent the surgery-first approach (SFA).
  • The nomogram showed calibration with an AUC of 0.995.
  • The Simple-CNN model achieved a sensitivity of 0.979, specificity of 0.800, precision of 0.960, accuracy of 0.909, and F1-score of 0.969, with an AUC of 0.896.

Sources:

  • CBCT radiomic features-based machine learning prediction models and nomogram for treatment decision-making regarding surgery-first approach in skeletal Class III malocclusion. BMC Oral Health, 2025,25(1):1-15. (BMC Oral Health - http://bmcoralhealth.biomedcentral.com)
  • Hospital of Stomatology, Guanghua School of Stomatology, Guangdong Provincial Key Laboratory of Stomatology, Sun Yast-sen University.

Read more