가미태음조위탕 복용 성인의 체중 감량 중 골격근량 감소 요인에 대한 의사결정나무 분석

Decision Tree Analysis of Factors Associated with Skeletal Muscle Mass Loss during Weight Loss Using Gamitaeeumjowee-tang

Objectives: This retrospective observational study evaluated the factors associated with skeletal muscle mass (SMM) loss during weight loss program using Gamitaeeumjowee-tang. Methods: The study cohort comprised adults who completed a weight loss program between January and May, 2024. Statistical analyses were conducted in R 4.5.1 (P<0.05). K-means clustering classified participants into greater SMM loss (GSL) and lesser SMM loss (LSL) groups. Independent variables included baseline characteristics and weight loss-related factors. Pearson correlation analysis was performed to identify associations with SMM loss. Decision tree models identified key split variables under four variable sets, with cross-validation and pruning applied. Retrospective power analysis confirmed 80% accuracy and Cohen’s h≥0.5 as sufficient. Results: GSL group lost significantly more weight (-9.59±5.28% vs. -6.20±3.94%, P=0.03) and had a higher weekly weight loss (-0.93±0.46% vs. -0.61±0.37%, P=0.01). SMM decreased markedly in GSL (-1.39±0.65 kg, P<0.001) but slightly increased in LSL (0.02±0.55 kg, P=0.90). Fat mass (FM) loss was similar between groups (-17.03±12.53% vs. -16.98±10.16%, P=0.58). SMM loss correlated with weight loss, body mass index (BMI) loss, and weekly weight loss. Decision tree analysis identified weekly weight loss (≥0.86%) as the key split in most models. In baseline-only models, age<53, initial BMI≥25, higher initial SMM, and shorter height predicted greater loss of SMM. Accuracy (71 to 73%) and effect size (h=0.43 to 0.47) supported clinical relevance. Conclusions: A weekly weight loss<0.86% was associated with better preservation of SMM and a reduction pattern primarily driven by FM, indicating this rate may be favorable for effective and healthy weight loss.