Scholarlink

Assessing aortic stiffness (AS) is key for early cardiovascular disease detection, as it reflects arterial wall mechanics and predicts adverse events. Carotidfemoral Pulse Wave Velocity (cfPWV) is the gold standard for non-invasive AS assessment, while Central Aortic Systolic Pressure (CASP) provides complementary information on cardiac afterload. This study explores the predictive value of Pulse Transit Time (PTT) measured at five peripheral sites (carotid, temporal, radial, digital, tibial) using a synthetic in-silico dataset. The temporal site showed the highest correlation with cfPWV and CASP and was selected for feature extraction from its photoplethysmography waveform. A fully-connected neural network was trained to estimate both cfPWV and CASP from these features. On a held-out test set, the model achieved high accuracy for cfPWV (R \(^2\) = 0.99, MAE = 0.10 m/s, RMSE = 0.13 m/s) and for CASP (R \(^3\) = 0.84, MAE = 2.89 mmHg, RMSE = 3.78 mmHg). These results highlight the diagnostic relevance of the temporal artery and support its use in non-invasive cardiovascular monitoring and potential cerebrovascular applications.

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

Temporal Artery Photoplethysmography as a Surrogate for Central Hemodynamics: From Pulse Transit Time Analysis to Deep Learning Estimation

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ana P. Nuñez,
  • Eugenia Ipar,
  • Mariano D. Vidal,
  • Ricardo L. Armentano,
  • Leandro J. Cymberknop

摘要

Assessing aortic stiffness (AS) is key for early cardiovascular disease detection, as it reflects arterial wall mechanics and predicts adverse events. Carotidfemoral Pulse Wave Velocity (cfPWV) is the gold standard for non-invasive AS assessment, while Central Aortic Systolic Pressure (CASP) provides complementary information on cardiac afterload. This study explores the predictive value of Pulse Transit Time (PTT) measured at five peripheral sites (carotid, temporal, radial, digital, tibial) using a synthetic in-silico dataset. The temporal site showed the highest correlation with cfPWV and CASP and was selected for feature extraction from its photoplethysmography waveform. A fully-connected neural network was trained to estimate both cfPWV and CASP from these features. On a held-out test set, the model achieved high accuracy for cfPWV (R \(^2\) = 0.99, MAE = 0.10 m/s, RMSE = 0.13 m/s) and for CASP (R \(^3\) = 0.84, MAE = 2.89 mmHg, RMSE = 3.78 mmHg). These results highlight the diagnostic relevance of the temporal artery and support its use in non-invasive cardiovascular monitoring and potential cerebrovascular applications.