Fingertip Pregnancy Pulse Signal Analysis Based On ResNet

Pulse diagnosis is a very distinctive diagnostic method in the field of traditional Chinese medicine.After years of development,traditional Chinese medicine pulse diagnosis has received more and more attention in the world for its non-invasive detection characteristics.However,so far,pulse diagnosis mainly relies on the practice and inheritance of doctors,which is not conducive to the demonstration of the reliability and large-scale promotion of pulse diagnosis.Therefore,how to digitize pulse and objectify pulse diagnosis has become an important research direction of modern Chinese medicine.In the field of Traditional Chinese medicine pulse diagnosis,"Ximai",which is pregnancy pulse,is a well-known pulse condition.Traditional Chinese medicine believes that during a woman’s pregnancy,the mother’s pulse will also change due to the growth of the fetus that needs nutrients from the mother.There are three mainstream methods of pregnancy test,Pregnancy test strips,B-ultrasound and Blood test.The pregnancy test strips cannot be reused and its accuracy is affected by some external factors.B-ultrasound and blood test need to go to the hospital for inspection.It is time-consuming and labor-intensive,among which blood test may cause some pain.Combining Traditional Chinese medicine pulse diagnosis and modern technology,we can obtain accurate pulse information from the human body.The idea of this article is to use modern technology to find the characteristics of pregnancy from the pulse of female fingertips.The specific steps are as follows:first collect fingertip pulse signals,then perform data preprocessing such as wavelet denoising,period segmentation,removal of abnormal values and baseline drift,and quality detection on the signal,finally use wavelet transformation and ResNet methods,women’s fingertip pulse signals can be accurately classified.This subject proposes a brand-new pregnancy test method.This method of screening pregnant women based on fingertip pulse waves can be used repeatedly and has a comfortable test experience,as well as high pregnancy test accuracy.On the basis of the above work,this subject has done three experiments.The first is a fingertip pulse signal classification experiment to determine whether a woman is pregnant.First collect the fingertip pulse data of 45 pregnant women and 45 non-pregnant women.After data preprocessing,10183 pulses of pregnant women and 12243 pulses of non-pregnant women are obtained.Among the 90 volunteers,66 people’s data was used as the training set and 24 people’s data was used as the test set.After calculation,the pulses of all 24 women in the test set were correctly classified,and the pregnancy test accuracy was calculated to be 100%.The second experiment is a classification experiment of early pregnancy women and non-pregnant women.The fingertip pulses of 16 early pregnant women and 15 non-pregnant women are used as the training set,and the fingertip pulses of 4 pregnant women and 4 non-pregnant women are used as the test set.In the collection,the physiological states of the 8 women are all correctly classified,so the accuracy of early pregnancy test is 100%.The third is the signal recognition accuracy experiment of pulse waves of different lengths.By inputting short-term pulses of 20 seconds,30 seconds,1 minute and 2 minutes,the calculated pregnancy test accuracy rates are 86.96%,90.48%,91.30%and 95.45%respectively.It shows that even if the input pulse time is only tens of seconds,the accuracy rate can be about 90%,and the longer the test time,the higher the accuracy.