Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Heart rate and respiratory rate prediction from noisy real-world smartphone based on Deep Learning methods

Created by
  • Haebom

Author

Ibne Farabi Shihab

Outline

This paper points out the limitations of previous studies on estimating biosignals, such as heart rate (HR) and respiration rate (RR), using mobile phone videos taken of fingertips during daily life, and proposes a new method to improve them. Previous studies mainly reported high accuracy based on data collected in a laboratory environment, but the accuracy was much lower in real daily life data. In response, the research team constructed a large-scale mobile phone video dataset taken during daily life and proposed a new 3D CNN model based on deep learning, which reduced the heart rate estimation error by 68% and the respiration rate estimation error by 75%. This suggests that the regression-based deep learning approach is effective for estimating heart rate and respiration rate.

Takeaways, Limitations

•
Takeaways:
◦
Clearly present the difficulties and limitations of estimating biosignals using mobile phone image data acquired in daily life.
◦
The problem of reduced applicability of existing algorithms to real environments is presented based on data, and a new solution based on deep learning is presented.
◦
We demonstrate that a novel method based on 3D CNN significantly improves the accuracy of heart rate and respiration rate estimation.
◦
To verify the effectiveness of regression-based deep learning approaches and suggest future research directions.
•
Limitations:
◦
Further validation of the generalization performance of the proposed method is needed.
◦
Lack of performance evaluation under various environments and conditions.
◦
Limitations on the size and diversity of the dataset.
◦
Additional research is needed for practical application in medical settings.
👍
Made with Slashpage