Personalized Prediction of Upper-Thigh Circumference Reduction After Circumferential Thigh Liposuction
Related Experiment Video
Updated: Nov 6, 2026
Ultrasonographic Evaluation of Breast Cancer-related Lymphedema
Published on: January 12, 2017
Jiyang Li1, Kaifang Hua1, Xinyu Zhang1
1Department of Body Contouring and Fat Grafting Center, Plastic Surgery Hospital and Institute, Chinese Academy of Medical Sciences & Peking Union Medical College, No. 33, BaDaChu Road, Shijingshan District, Beijing, 100144, People's Republic of China.
Background:
Operative endpoints in circumferential thigh liposuction are experience dependent. A patient-specific model relating routinely collected characteristics and planned lipoaspirate volume (LV) to expected circumference change could provide a structured adjunct for counseling and planning.
Objectives:
To develop an interpretable linear model for the postoperative change ratio of upper-thigh circumference (RC-upper) and characterize performance in repeated cross-validation and a same-center prospective temporal validation cohort.
Methods:
The model-development cohort comprised 452 women treated from July 2018 to June 2022. Stepwise multivariable linear regression identified a locked equation using BMI, body fat percentage (BFP), upper-thigh subcutaneous adipose thickness (T-upper), and LV. Author-supplied additional analysis evaluated the locked model using 100 repeats of stratified tenfold cross-validation. Temporal validation used 132 prospectively identified, non-overlapping same-center patients. Patient-level paired percentile bootstrap resampling (2,000 replicates) provided 95% confidence intervals.
Results:
The locked equation was RC-upper = - 0.9029 + 0.3309 × BMI - 0.1893 × BFP + 0.6670 × T-upper + 0.0015 × LV (apparent R2 = 0.7655). Repeated cross-validation yielded R2 = 0.7607 (95% CI, 0.7200-0.7945), MAE = 1.5173 percentage points (95% CI, 1.4171-1.6223), and RMSE = 1.8772 percentage points (95% CI, 1.7605-1.9926). Same-center temporal validation yielded R2 = 0.9680 (95% CI, 0.9537-0.9771), MAE = 0.8970 cm (95% CI, 0.7921-1.0145), and RMSE = 1.1036 cm (95% CI, 0.9736-1.2345).
Conclusions:
The model may support individualized counseling and planning in patients similar to the study cohort. It does not define a safe LV, replace surgeon judgment, or establish improved clinical outcomes. Full-process internal validation, multicenter external validation, and prospective impact evaluation remain necessary.
Level Of Evidence Iii:
This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors https://www.springer.com/00266 .
