基于生物力学表型的嗓音障碍治疗选择框架提案：一种临床决策模型

文献信息

中文摘要

目的： 传统的嗓音障碍器质性-功能性分类并未明确嗓音治疗的生物力学靶点。本文提出一种基于规则的生物力学表型分类，以支持候选治疗方法的选取，并说明其应如何验证。
方法： 将二维的器质性-张力性排列1,2改编为分类评分规则，仅使用视频喉频闪镜（VLS）和欧洲喉科学会-欧洲嗓音医学家联盟核心数据集中的GRBAS评级。输入包括组织轴（X，0-3）、活动度标志（M）、超内收和低内收子评分（H、L）以及非病变性声门裂隙标志（G）。空气动力学、声学和运动学测量被排除在分型之外，保留作为外部验证指标。通过穷举枚举检查逻辑一致性，并通过精确法和蒙特卡洛法推导验证样本量。
结果： 有序规则产生六种表型（功能亢进型、功能低下型、结构-僵硬型、结构-病变型、混合A型和混合B型）以及一个正常张力/未分类类别，并为7680种可允许输入组合中的每一种分配恰好一个标签（36个决策单元，无冲突）。分层信度样本105例（每类15例）使Fleiss' κ的95%置信区间半宽至多为0.082。在可行规模的连续队列中，验证性收敛效度对比无法获得足够把握度；因此指定了富集罕见表型的分阶段设计。
结论： 该评分规则在逻辑上穷尽且互斥，其验证性假设所依赖的测量不参与分型。其临床效度及表型-技术映射仍有待前瞻性检验。

英文摘要

OBJECTIVES: Traditional organic-functional classifications of voice disorders do not specify the biomechanical targets of voice therapy. This article proposes a rule-based classification of biomechanical phenotypes to support the selection of candidate therapeutic approaches, and specifies how it should be validated.
METHODS: The two-dimensional organicity-tonicity arrangement1,2 was adapted into a categorical rubric that uses only videolaryngostroboscopy (VLS) and GRBAS ratings from the European Laryngological Society-Union of the European Phoniatricians Core Dataset. Inputs are a tissue axis (X, 0-3), a mobility flag (M), hyperadductory and hypoadductory subscores (H, L), and a nonlesional glottal gap flag (G). Aerodynamic, acoustic, and kinematic measures were excluded from assignment and reserved as external validators. Logical consistency was checked by exhaustive enumeration, and validation sample sizes were derived by exact and Monte Carlo methods.
RESULTS: Ordered rules yield six phenotypes (Hyperfunctional, Hypofunctional, Structural-Rigid, Structural-Lesional, Mixed-A, and Mixed-B) plus a Eutonic/Unclassified category, and assign exactly one label to each of 7680 admissible input combinations (36 decision cells, no conflicts). A stratified reliability sample of 105 cases (15 per category) gives a 95% confidence-interval half-width for Fleiss' κ of at most 0.082. Confirmatory convergent-validity contrasts cannot be adequately powered in a consecutive cohort of feasible size; a staged design with enrichment of rare phenotypes is therefore specified.
CONCLUSIONS: The rubric is logically exhaustive and mutually exclusive, and its confirmatory validation hypotheses rely on measures that play no part in assignment. Its clinical validity and the phenotype-to-technique mapping remain to be tested prospectively.