研究目的
To propose a cuff-less continuous method for measuring blood pressure using wrist and fingertip photo-plethysmograms, evaluating its accuracy and analyzing feature importance.
研究成果
The proposed method achieves accurate cuff-less blood pressure estimation with MAEs as low as 4.94/4.03 mmHg for SBP/DBP, meeting clinical standards, and demonstrates that feature selection improves performance, with potential for implementation in wearable devices.
研究不足
The study used a custom system not commercially available, and the sample size of 111 participants may limit generalizability; further validation on larger populations is needed.
1：Experimental Design and Method Selection:
The study involved designing a custom system for simultaneous PPG recording from wrist and fingertip, extracting morphological features from PPG and APG signals, and using multi-layer neural networks with genetic algorithms for feature selection and blood pressure estimation.
2：Sample Selection and Data Sources:
111 volunteers (100 males, 11 females, age 32.63 ± 15.82 years) participated, with PPG signals recorded for 2 minutes per participant, and reference BP measured using an Omron sphygmomanometer.
3：63 ± 82 years) participated, with PPG signals recorded for 2 minutes per participant, and reference BP measured using an Omron sphygmomanometer.
List of Experimental Equipment and Materials:
3. List of Experimental Equipment and Materials: Custom-made PPG recording system with infrared sensors, filters, amplifiers, and ADCs; Omron M6 Comfort sphygmomanometer for reference measurements.
4：Experimental Procedures and Operational Workflow:
Signals were filtered, features extracted (e.g., PTT, PWV, morphological parameters), and ANNs trained using leave-one-out cross-validation with feature selection via moving-backward and genetic algorithms.
5：Data Analysis Methods:
Mean absolute error (MAE) was used to evaluate performance, with statistical analysis including Bland-Altman plots and correlation coefficients.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
