Partial Least Squares Bootstrapping Results.

Background
This study aimed to identify patient segments of acceptability to vaginal dosage forms in women with the genitourinary syndrome of menopause and to identify the most important predictors and mediators of each segment using latent segment analysis.
Methods
A cross-sectional study included 351 peri- and postmenopausal women from two tertiary care hospitals in the United Arab Emirates. To select the best model, we ran a finite-mixture partial least squares segmentation (FIMIX-PLS). The number of resulting segments was used to run a partial least squares-predicted-oriented segmentation to assign cases to segments and maximize the segment-specific explained variance (R2) across all groups. Multi-group analysis was then performed to examine whether differences between segments were significant. Analyses were performed using the SmartPLS Software version 4.9,
Results
A two-segment model was identified (Entropy > 0.8, Corrected-Akaike’s Information Criterion = 1977.11; Bayesian Information Criterion = 1960.11), which indicated adequate and well-separated segments. The first and second segments had 179 and 171 of the cases, respectively, which were considered substantive representations of both segments. The model resulted in a weighted average R2 greater than that of the original sample, indicating better predictive relevance of the model. The multi-group analysis showed that the differences between the two segments were significant. Being on vaginal treatment was observed to be the variable that drove the partition of the segments. Perceived effectiveness for those who were not on treatment (segment 1) and those who were (segment 2) was predicted by affective attitude and intervention coherence, respectively,
Conclusion
The acceptability of vaginal treatments differs between experienced and anticipated users. Considering patient-related factors and previous treatment experiences can serve as a benchmark to improve patient acceptability of treatment.

背景
本研究旨在明确绝经泌尿生殖综合征（genitourinary syndrome of menopause）女性对阴道给药制剂（vaginal dosage forms）的可接受性患者群体，并采用潜段分析（latent segment analysis）识别各群体的关键预测因子与中介变量。

方法
本研究为横断面研究（cross-sectional study），纳入来自阿拉伯联合酋长国两家三级医院（tertiary care hospitals）的351名围绝经期与绝经后女性（peri- and postmenopausal women）。为筛选最优模型，我们运行了有限混合偏最小二乘分段法（finite-mixture partial least squares segmentation, FIMIX-PLS）。基于得到的群体数量，我们采用偏最小二乘预测导向分段法（partial least squares-predicted-oriented segmentation）对研究对象进行群体划分，以最大化所有分组的组内解释方差（R²）。随后开展多组分析（multi-group analysis），检验群体间差异是否具有统计学意义。所有分析均使用SmartPLS软件4.9版本完成。

结果
最终确定双群体模型（熵值>0.8，修正赤池信息准则=1977.11；贝叶斯信息准则=1960.11），表明群体划分充分且组间区分度良好。第一、第二群体分别包含179例与171例研究对象，均为具有实质意义的群体划分结果。该模型的加权平均决定系数（R²）高于原始样本，提示模型具备更优的预测效度。多组分析显示，双群体间差异具有统计学意义。是否接受阴道治疗是驱动群体划分的核心变量。未接受阴道治疗者（群体1）的感知疗效由情感态度（affective attitude）预测，而接受治疗者（群体2）的感知疗效则由干预契合度（intervention coherence）预测。

结论
绝经泌尿生殖综合征女性对阴道治疗的可接受性在既往有治疗经验者与预期使用者间存在差异。综合考虑患者相关因素与既往治疗经历，可作为提升患者治疗可接受性的参考基准。