Seda Sertel Meyvaci; Handan Ankarali & Sena Demiroglu

This study aims to develop predictive models for estimating the humerus total length (HTL) from dry humerus specimens using 19 transverse and vertical segment measurements, thereby contributing to applications in anthropometry, forensic medicine, and clinical practice. A total of 26 dry humeri (11 right, 15 left) of unknown age and sex from the Anatomy Department of Bolu Abant Izzet Baysal University were analyzed. The total humeral length was measured manually while 19 morphometric segments were measured using a digital vernier caliper. Each segment was measured three times and mean values were used for analysis. Normality was assessed via Shapiro-Wilk tests and right-left comparisons were performed using Independent Samples T-Test. Pearson correlation analysis was conducted to identify relationships between segment lengths and HTL. Two predictive models were constructed: multiple linear regression (MLR) with backward variable selection and multivariate adaptive regression splines (MARS). Model performance was assessed via coefficient of determination (R2) and root mean square error (RMSE). No significant differences were observed between right and left humeri; therefore, analyses were performed collectively. Pearson correlation analysis indicated that 17 out of 19 segment measurements were significantly associated with HTL. The MLR model retained 9 measurements, achieving R2 = 0.984 and RMSE = 2.937. The MARS model included 5 segments with interaction terms, yielding R2 = 0.994 and RMSE = 1.755. The most influential measurements in MARS were ISUBH, DTUH, DTLH, PHTW, and SNUH. Both MLR and MARS models demonstrated high predictive accuracy for HTL, with MARS showing superior performance and efficiency. These models provide reliable tools for estimating humerus length in anthropometric, forensic and clinical contexts. Incorporating additional factors such as age, sex and body height in future studies may further enhance predictive performance.

KEY WORDS: Humerus morphometry; Bone length prediction; Anthropometry; Forensic identification; Clinical anatomy.

SERTEL MEYVACI, S.; ANKARALI, H. & DEMIROGLU, S. Modeling the entire length of the humerus from transverse and vertical segment lengths. Int. J. Morphol., 44(1):158-164,2026.