The article presents the development of BioRural, a low-cost portable device with Edge AI designed to autonomously detect health status by reading vital signs in rural areas of Ecuador. The device consists of biomedical sensors, a TTGO ESP32 module, and a multilayer perceptron (MLP) artificial neural network that infers health status at three levels: “normal,” “alert,” and “risk,” without requiring constant Internet connectivity. The MLP model, trained with Kaggle data, achieved an accuracy of 99.77% in training and 99.58% in validation, while field tests reached an accuracy of 86% and an average latency of 1.7 min for inference. Measurements of heart rate, oxygen saturation, and temperature showed an average difference of 0.15% in SpO \(_2\) , 0.72 BPM, and 2.26 \(^\circ \) C compared to commercial medical devices. The results support the viability of BioRural as an autonomous, efficient, and adaptable solution for primary health care in rural communities with limited access to medical services.

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

Low-Cost Edge AI Device for Autonomous Health Assessment in Rural Areas

  • Vilmer D. Criollo,
  • Carlos F. López,
  • David I. Sosa-Zuñiga,
  • Edison D. Mañay

摘要

The article presents the development of BioRural, a low-cost portable device with Edge AI designed to autonomously detect health status by reading vital signs in rural areas of Ecuador. The device consists of biomedical sensors, a TTGO ESP32 module, and a multilayer perceptron (MLP) artificial neural network that infers health status at three levels: “normal,” “alert,” and “risk,” without requiring constant Internet connectivity. The MLP model, trained with Kaggle data, achieved an accuracy of 99.77% in training and 99.58% in validation, while field tests reached an accuracy of 86% and an average latency of 1.7 min for inference. Measurements of heart rate, oxygen saturation, and temperature showed an average difference of 0.15% in SpO \(_2\) , 0.72 BPM, and 2.26 \(^\circ \) C compared to commercial medical devices. The results support the viability of BioRural as an autonomous, efficient, and adaptable solution for primary health care in rural communities with limited access to medical services.