Do machine learning methods solve the main pitfall of linear regression in dental age estimation?

Area of Science:

• Forensic Anthropology
• Dental Age Estimation
• Machine Learning Applications

Background:

• Dental age estimation is vital in forensic and anthropological studies due to tooth preservation.
• Machine learning (ML) algorithms offer high accuracy in age estimation.
• The precision and error trends of ML methods in dental age estimation require thorough investigation.

Purpose of the Study:

• To compare ML-assisted age estimation methods with traditional techniques.
• To evaluate the performance of Random Forest, Support Vector Regression, K-Nearest Neighbors, and Gradient Boosting Method.
• To assess accuracy and precision against linear regression and Segmented Normal Bayesian Calibration.

Main Methods:

• Analysis of 1,949 orthopantomographs from South African children (ages 5-14).
• Comparison of ML models (Random Forest, SVR, KNN, GBM) against linear regression and SNBC.
• Evaluation using Mean Absolute Error, Root Mean Squared Error, Inter-Quartile Range, and error slope.

Main Results:

• ML methods slightly outperformed traditional models in accuracy.
• Gradient Boosting Method and Support Vector Regression showed the highest accuracy (MAE: 0.69, RMSE: 0.85).
• ML and linear regression exhibited significant residual bias; SNBC showed none. Gender did not significantly impact results.

Conclusions:

• ML methods provide high accuracy in dental age estimation.
• Systematic error trends in ML models necessitate careful evaluation.
• Future research should focus on improving accuracy while mitigating biases.