Scholarlink

Hyposalivation in women presents diagnostic challenges due to overlapping etiologies including medications, hormonal changes and autoimmune conditions. This exploratory study investigates whether computational approaches can identify distinct hyposalivation phenotypes in women. We analyzed 196 women across 40 clinical variables including demographics, medications, dental health, and salivary flow measurements using unsupervised machine learning. K-means clustering with PCA preprocessing tentatively identified three clusters: medication and lifestyle-associated hyposalivation (n=141, 72%), optimal salivary function (n=18, 9%), and severe hyposalivation (n=37, 19%). Salivary flow parameters and dental health variables were primary cluster drivers. Results provide preliminary evidence that hyposalivation may comprise distinct phenotypes, offering proof-of-concept for computational phenotyping approaches that warrant validation in independent cohorts.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Exploratory Computational Phenotyping of Hyposalivation Etiologies in Women

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Kristina Lacasta,
  • María J. Rus,
  • Angela de la Cruz Gándara Alvarez,
  • Cristiane Cantiga-Silva,
  • Virginia Moreira Navarrete,
  • Carmen Dominguez Quesada,
  • Jose Javier Perez Venegas,
  • Juan Antonio Ortega,
  • Aurea Simon-Soro

摘要

Hyposalivation in women presents diagnostic challenges due to overlapping etiologies including medications, hormonal changes and autoimmune conditions. This exploratory study investigates whether computational approaches can identify distinct hyposalivation phenotypes in women. We analyzed 196 women across 40 clinical variables including demographics, medications, dental health, and salivary flow measurements using unsupervised machine learning. K-means clustering with PCA preprocessing tentatively identified three clusters: medication and lifestyle-associated hyposalivation (n=141, 72%), optimal salivary function (n=18, 9%), and severe hyposalivation (n=37, 19%). Salivary flow parameters and dental health variables were primary cluster drivers. Results provide preliminary evidence that hyposalivation may comprise distinct phenotypes, offering proof-of-concept for computational phenotyping approaches that warrant validation in independent cohorts.