Machine learning and regression analysis for age estimation from the iliac crest based on computed tomographic...

Area of Science:

• Forensic Anthropology
• Radiology
• Biometrics

Background:

• Accurate age estimation is crucial for legal proceedings involving minors and young adults.
• The iliac crest is a reliable indicator of skeletal maturity in these age groups.
• The modified Risser method offers a novel approach to iliac crest age assessment.

Purpose of the Study:

• To evaluate the applicability of the modified Risser method for age estimation in the Indian population.
• To compare the accuracy of regression and machine learning models for this method.
• To establish the reliability of computed tomography-based iliac crest ossification for forensic age determination.

Main Methods:

• Computed tomography (CT) scans of the iliac crest were analyzed from individuals in the Indian population.
• Scans were scored using the modified Risser stages to assess ossification and fusion.
• Various regression and machine learning models were developed and tested for age estimation accuracy.

Main Results:

• The modified Risser method accurately reflects skeletal maturation changes in the iliac crest.
• Compound regression showed the lowest error among traditional models (inaccuracy: 4.78 years).
• Machine learning, specifically decision tree regression, significantly improved accuracy (inaccuracy: 1.88 years).

Conclusions:

• Machine learning models demonstrate statistically superior performance for forensic age estimation compared to traditional regression.
• The modified Risser method, when analyzed with machine learning, provides a reliable tool for age estimation in legal contexts.
• This study highlights the potential of CT-based iliac crest ossification for accurate age determination in the Indian population.