Scholarlink

<p>With the increasing global prevalence of diabetes, predictive models are crucial for early interventions, including elderly participants who are often underrepresented in existing models. Existing models, however, are often derived from specific subpopulations or from individuals with prevalent diabetes, which limits their generalizability to the broader population. This study aimed to develop and validate a predictive model for type 2 diabetes mellitus (T2DM) onset using a large Japanese cohort including elderly participants. Data from the Shizuoka Kokuho Database, comprising over 2.5&#xa0;million people, was used. The analysis included 463,248 adults aged 40 and above who underwent health checkups. Participants were split into derivation (308,832) and validation datasets (154,416) in a 2:1 ratio. Predictive factors, identified using Cox proportional hazards models, included demographics, clinical parameters, and lifestyle factors. During a median follow-up period of 5.17 years, 16.9% of the derivation group and 17.0% of the validation group developed T2DM. The model assigns weighted scores to factors like age, sex, BMI, blood pressure, lipid profiles, liver enzymes, kidney function, and lifestyle habits. The model achieved a Harrell’s c-index of 0.656 (95% confidence interval, 0.652–0.659) in the validation dataset, indicating modest predictive performance. This model, based on routinely collected health check-up data, may facilitate risk stratification and guide preventive interventions.</p>

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

Development and validation of a type 2 diabetes mellitus prediction tool using a large Japanese regional insurance claims database

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月30日

查看全文
  • Tatsunori Satoh,
  • Eiji Nakatani,
  • Hiroyuki Ariyasu,
  • Chihiro Saito,
  • Yuri Oshiro,
  • Emi Ohata,
  • Hiroshi Itoh,
  • Kaori Hayashi,
  • Takeshi Usui

摘要

With the increasing global prevalence of diabetes, predictive models are crucial for early interventions, including elderly participants who are often underrepresented in existing models. Existing models, however, are often derived from specific subpopulations or from individuals with prevalent diabetes, which limits their generalizability to the broader population. This study aimed to develop and validate a predictive model for type 2 diabetes mellitus (T2DM) onset using a large Japanese cohort including elderly participants. Data from the Shizuoka Kokuho Database, comprising over 2.5 million people, was used. The analysis included 463,248 adults aged 40 and above who underwent health checkups. Participants were split into derivation (308,832) and validation datasets (154,416) in a 2:1 ratio. Predictive factors, identified using Cox proportional hazards models, included demographics, clinical parameters, and lifestyle factors. During a median follow-up period of 5.17 years, 16.9% of the derivation group and 17.0% of the validation group developed T2DM. The model assigns weighted scores to factors like age, sex, BMI, blood pressure, lipid profiles, liver enzymes, kidney function, and lifestyle habits. The model achieved a Harrell’s c-index of 0.656 (95% confidence interval, 0.652–0.659) in the validation dataset, indicating modest predictive performance. This model, based on routinely collected health check-up data, may facilitate risk stratification and guide preventive interventions.