Scholarlink

<p>The electrical properties of the skin can reflect changes in its structure and physiological state, and bioimpedance analysis has been widely used to distinguish specific components of the human body, such as water, fat, or muscle tissue, instead of the traditional examinations with bulky equipment gradually. For the assessment of microangiopathy, a low-cost, simple, effective, and clinically interpretable model was proposed that relies on individual skin conductance data, collected from 28 patients with lower extremity arterial occlusion. The model demonstrated a specificity of 67.4% and a sensitivity of 82.9% in classifying healthy-affected sides. The severity estimates were consistent with the patient’s laser speckle and ankle-brachial index, with intra-class correlation coefficients of 0.43 and 0.55. Further, patient record data was combined to improve accuracy by about 15% in multimodal ensemble learning, indicating the potential for using the electrical properties of the skin to characterize surface microcirculation disorders.</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

A clinically interpretable model derived from skin conductance for assessing microangiopathy of the skin surface

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年3月28日

查看全文
  • Ximing Zhu,
  • ShihChun Chu,
  • Jie Fu,
  • Siyu Ao,
  • Tingting Liu,
  • Ziming Liao,
  • Jiamei Song,
  • Yuhong Wang,
  • Yanping Guo,
  • Qinghong Song,
  • Pengfei Liu,
  • Shuang Li,
  • Bing Chen,
  • Bin Yao

摘要

The electrical properties of the skin can reflect changes in its structure and physiological state, and bioimpedance analysis has been widely used to distinguish specific components of the human body, such as water, fat, or muscle tissue, instead of the traditional examinations with bulky equipment gradually. For the assessment of microangiopathy, a low-cost, simple, effective, and clinically interpretable model was proposed that relies on individual skin conductance data, collected from 28 patients with lower extremity arterial occlusion. The model demonstrated a specificity of 67.4% and a sensitivity of 82.9% in classifying healthy-affected sides. The severity estimates were consistent with the patient’s laser speckle and ankle-brachial index, with intra-class correlation coefficients of 0.43 and 0.55. Further, patient record data was combined to improve accuracy by about 15% in multimodal ensemble learning, indicating the potential for using the electrical properties of the skin to characterize surface microcirculation disorders.