Scholarlink

Objective <p>Both alkaline phosphatase (ALP) and serum creatinine (sCr) have been individually associated with adverse outcomes in patients with diabetes. This study investigates a novel composite indicator-ln[ALP × sCr]-to predict all-cause and CVD mortality risk among U.S. adults with diabetes.</p> Research design and methods <p>We analyzed data from 82,091 U.S. adults enrolled in NHANES (1999-2014), with mortality follow-up through December 31, 2019. A deep learning model identified ALP, sCr, and vitamin D as top mortality-related biomarkers. Based on these results, we derived a composite index, ln[ALP × sCr], to reflect integrated cardiac-renal dysfunction. Restricted cubic spline analysis was used to define risk thresholds. Cox proportional hazards models assessed the association between ln[ALP × sCr] and all-cause, cardiovascular, and diabetes-related mortality.</p> Results <p>Over a median follow-up of 11.4&#xa0;years, 4,839 T2DM patients in the highest quartile of ln[ALP × Scr] exhibited significantly elevated risks of all-cause (HR 1.47, 95% CI 1.18-1.82), cardiovascular (HR 1.44, 95% CI 1.01-2.04), and diabetes-related mortality (HR 2.50, 95% CI 1.58-3.96), compared to those in the lowest quartile. Mediation analysis indicated that serum vitamin D accounted for 24.3% of the association between the composite biomarker and all-cause mortality (<i>P</i> &lt; 0.001).</p> Conclusions <p>In this nationally representative cohort, ln[ALP × Scr] showed a J-shaped association with all-cause mortality. These findings highlight the potential utility of ln[ALP × Scr] as a simple, noninvasive biomarker for mortality risk assessment in individuals with diabetes. This study also illustrates the value of integrating AI-based feature selection with traditional epidemiologic modeling to enhance long-term mortality risk stratification and inform public health strategies.</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

From deep learning discovery to clinical validation: a new composite marker predicts mortality in type 2 diabetes

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年9月3日

查看全文
  • Jianjun Wu,
  • Dawei Yang,
  • Youqi Zhang,
  • Liu Ji,
  • Xiao Tong,
  • Guanghua Chen,
  • Fan Yang

摘要

Objective

Both alkaline phosphatase (ALP) and serum creatinine (sCr) have been individually associated with adverse outcomes in patients with diabetes. This study investigates a novel composite indicator-ln[ALP × sCr]-to predict all-cause and CVD mortality risk among U.S. adults with diabetes.

Research design and methods

We analyzed data from 82,091 U.S. adults enrolled in NHANES (1999-2014), with mortality follow-up through December 31, 2019. A deep learning model identified ALP, sCr, and vitamin D as top mortality-related biomarkers. Based on these results, we derived a composite index, ln[ALP × sCr], to reflect integrated cardiac-renal dysfunction. Restricted cubic spline analysis was used to define risk thresholds. Cox proportional hazards models assessed the association between ln[ALP × sCr] and all-cause, cardiovascular, and diabetes-related mortality.

Results

Over a median follow-up of 11.4 years, 4,839 T2DM patients in the highest quartile of ln[ALP × Scr] exhibited significantly elevated risks of all-cause (HR 1.47, 95% CI 1.18-1.82), cardiovascular (HR 1.44, 95% CI 1.01-2.04), and diabetes-related mortality (HR 2.50, 95% CI 1.58-3.96), compared to those in the lowest quartile. Mediation analysis indicated that serum vitamin D accounted for 24.3% of the association between the composite biomarker and all-cause mortality (P < 0.001).

Conclusions

In this nationally representative cohort, ln[ALP × Scr] showed a J-shaped association with all-cause mortality. These findings highlight the potential utility of ln[ALP × Scr] as a simple, noninvasive biomarker for mortality risk assessment in individuals with diabetes. This study also illustrates the value of integrating AI-based feature selection with traditional epidemiologic modeling to enhance long-term mortality risk stratification and inform public health strategies.