| The concentration of hemoglobin,an important indicator in physical health,plays a significant role in the medical field of health assessment and disease prediction.At present,in clinical,invasive detection is the mainly method to detect hemoglobin.While,this method cannot achieve real-time monitoring of hemoglobin concentration,and the patient experience is poor.Non-invasive detection can well solve the above problems well.In this research,aiming to enrich non-invasive hemoglobin detection methods,based on photoplethysmography and modified Lambert’s law,using dual-wavelength reflective method,a non-invasive,real-time monitoring device has successfully fabricated for hemoglobin detection.The hardware of the monitoring device mainly consisted of five parts: the STM32F407 minimum system,PPG signal acquisition module,TFT-LCD display module,power management module,and SD card storage module.The PPG signal acquisition module used the MAX30100 sensor to acquire PPG signals.Based on UCOS-III real-time operating system and STemWin,the software design of the system includes application programming,graphical interface design,software filtering and extraction of PPG signal feature parameters extraction.At the same time,93 sets of experimental sample data were collected to provide reliable data for subsequent studies using this system.In the research of the hemoglobin detection model,the HHT adaptive filter algorithm was used to filter the original PPG signal,firstly.Then we use the differential method to achieve PPG feature parameters extraction.Then using RELIEFF and PCA algorithm to filter the extracted feature parameters.Finally,using the selected characteristic parameters as input,and the hemoglobin concentration detected by the MISSION hemoglobin analyzer as target value,the hemoglobin detection model was constructed using PLS.In this study,taking 62 sets of experimental sample data as the training set and 31 sets of experimental sample data as the check set,the correlation coefficient was 0.88,and the mean square error was 0.44 g/d L.To verify the performance of non-invasive hemoglobin monitoring equipment,in this work,16 volunteers used the noninvasive test equipment designed in this study and the minimally invasive MISSION hemoglobin analyzer to perform hemoglobin detection.16 sets of comparative data were obtained and analyzed to obtain a correlation coefficient of 0.88 with a mean square error of 1.58g/dL.The work illustrates that the detection model and device of hemoglobin can accurately monitor the hemoglobin concentration and have good stability.It is of great significance to the study of clinical non-invasive hemoglobin detection. |