Method for detecting influence of obesity on viscera health based on PPG signal of mobile terminal

Abstract

The invention discloses a method for detecting influence of obesity on viscera health based on PPG signals of a mobile terminal. Comprising the following steps: acquiring an original digital signal by adopting a photoplethysmography (PPG) signal; carrying out mean value normalization processing on the collected original digital signals; performing signal separation analysis on the normalized digital signal to generate an independent source signal; the independent source signals are screened, and a Butterworth low-pass filter is adopted to eliminate power frequency interference and myoelectricity high-frequency noise; a wavelet function is adopted, 9-layer wavelet decomposition is conducted on the pulse signals, and wavelet details of high-frequency noise larger than 20 Hz are determined according to a wavelet decomposition frequency spectrum. According to the detection method, the diagnosis and treatment accuracy is improved, the detection efficiency is higher, and the medical cost is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to a method for detecting the influence of obesity on the health of viscera, and more particularly to a method for detecting the influence of obesity on the health of viscera based on PPG signals of a mobile terminal. BACKGROUND

[0002] With the continuous change of people's lifestyle, the prevalence of diseases such as obesity is showing a rising trend. Obesity greatly increases the incidence of diseases such as hyperlipidemia, diabetes, and cardiovascular and cerebrovascular diseases.

[0003] The formation of obesity is closely related to unhealthy diet. Overeating of obese people leads to the accumulation of cream, phlegm, water and dampness in the body, resulting in dysfunction of viscera and body weight exceeding that of ordinary people. According to modern Chinese medicine theory, obesity is related to dysfunction of multiple viscera such as spleen and kidney, and its pathology is based on spleen deficiency, and phlegm, water and dampness as the standard, leading to dysfunction of blood circulation and function of each viscera.

[0004] Currently, to understand the health problems of obese people, it is necessary to rely on experienced doctors and a series of physical health examinations to judge the health risks of obese people. In traditional Chinese medicine, the generation of pulse is related to the beating of heart, the gain and loss of blood and the coordination of viscera. Experienced Chinese medicine doctors can determine the gain and loss of blood of viscera and infer the physical condition of patients by pulse-taking. Since pulse-taking is artificial, it is limited by the personal experience and subjective cognitive differences of Chinese medicine doctors, resulting in poor mastery of pulse-taking by young Chinese medicine doctors. In order to improve the accuracy of diagnosis and treatment, improve the detection efficiency and reduce the medical cost, there is an urgent need for a convenient and efficient detection method to promote medical development. SUMMARY

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a method for detecting the influence of obesity on the health of viscera based on PPG signals of a mobile terminal. The technical problem to be solved by the present application is that there is no algorithm for automatic evaluation of the influence of obesity on the health of viscera at present.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a method for detecting the influence of obesity on the health of viscera based on PPG signals of a mobile terminal, comprising the following steps: Step one: adopt photoelectric plethysmography (PPG) signal to collect original digital signals.

[0007] Step two: the original digital signals collected are subjected to mean normalization processing.

[0008] Step three: the mean normalized digital signals are subjected to signal separation analysis to generate independent source signals.

[0009] Step four: screening the independent source signal, using a Butterworth low-pass filter to eliminate power frequency interference and myoelectric noise high-frequency noise.

[0010] Step five: using wavelet function to carry out 9-layer wavelet decomposition on the pulse signal, and determining the high-frequency noise wavelet details greater than 20Hz according to the wavelet decomposition frequency spectrum.

[0011] Step six: filtering the high-frequency wavelet coefficients and low-frequency wavelet details, reconstructing the signal, and obtaining the denoised pulse signal.

[0012] Step seven: homogenizing the denoised pulse signal to obtain the average pulse waveform.

[0013] Step eight: Fourier transform harmonic analysis is performed on the collected homogenized waveform pulse signal to obtain a Fourier spectrum containing phase and amplitude information.

[0014] Step nine: the amplitude is obtained by the Abs function, and the signal within the frequency range of 20Hz is intercepted to obtain the normalized amplitude information.

[0015] Step ten: based on the Fourier transform results of the pulse of the healthy population and the gas-blood resonance principle, a basic database and the corresponding healthy population viscera data model are established.

[0016] Step eleven: according to the viscera data model of the healthy population, the health condition of the obese population is automatically evaluated.

[0017] The PPG signal is collected by the photoplethysmography method, and the original digital signal is collected by the mobile terminal, which includes but is not limited to a blood oxygen meter, a mobile phone, a bracelet, a smart wearable device, etc.

[0018] In order to obtain accurate pulse signals, the pulse signal time collected is not less than 15s, and more preferably more than 30s.

[0019] The mean normalized digital signal is analyzed by signal separation, and the signal separation method includes but is not limited to independent component analysis (ICA) method, fast independent component analysis (FastICA) method, convolutional neural network analysis method, Haar-like feature combined with AdaBoost classifier analysis method, etc.

[0020] The pulse signal is decomposed by the wavelet function, and the signal details are analyzed and studied. The signal is decomposed by 9 layers, and the high-frequency part and the low-frequency part of the signal are superimposed, which is represented by the following formula: S=c A 1+c D 1=c A2+c D 2+c D 1=...=c A 9+c D 9+c D 8+...+c D 2+c D 1 Wherein, the pulse signal is mainly distributed in 20Hz, the high frequency noise greater than 20Hz and the low frequency baseline drift noise are filtered out, and the final pulse signal of the application is obtained.

[0021] Wherein, the stable pulse signal of the application is mainly distributed in 20Hz, the high frequency noise greater than 20Hz and the low frequency baseline drift noise are filtered out, and the final pulse signal of the application is obtained.

[0022] The data model of the healthy population is collected, and the age of the population is 18-25 years old, 100 healthy samples of male are collected with the body weight index BMI between 18.5-24, and 100 healthy samples of female are collected with the body weight index BMI between 18.5-24.

[0023] Wherein, the healthy samples collected from the healthy population are invited to at least 10 expert judges to determine the health constitution by pulse diagnosis.

[0024] Wherein, the pulse signal collected is subjected to Fourier transform to obtain normalized amplitude data, and the amplitude size change of the obese population is calculated based on the amplitude basic data of the healthy population, and when the amplitude data change is greater than or equal to 10%, it is determined that the health condition is affected.

[0025] The basic data model is established by the air-blood resonance principle, and more than 10 experts are invited to verify the pulse diagnosis, and the harmonic frequencies of each level are calculated by Fourier transform, the first harmonic frequency is 0.6-1.8Hz, which is the air-blood vibration frequency of liver, the second harmonic frequency is 1.8-2.8Hz, which is the air-blood vibration frequency of kidney, the third harmonic frequency is 2.8-4.2Hz, which is the air-blood vibration frequency of spleen, the fourth harmonic frequency is 3.6-5.6Hz, which is the vibration frequency of lung, the fifth harmonic frequency is 4.6-7Hz, which is the vibration frequency of stomach, and the sixth harmonic frequency is 5.6-8Hz, which is the vibration frequency of gallbladder.

[0026] Wherein, the amplitude data change of the obese population is calculated based on the basic data of the healthy population, and the greater the amplitude data change value under different harmonic frequencies, the more serious the health status.

[0027] Wherein, the data of amplitude has the difference of positive value and negative value, the amplitude change is positive value, it is explained that the viscera needs more qi and blood supply to meet the requirement, the amplitude change is negative value, it is explained that the qi and blood that viscera runs when delivery is insufficient, from this, the influence situation of obesity to the health of viscera can be judged.

[0028] Technical effects and advantages of the present application: 1, the present application extracts characteristic value after processing original PPG signal, and establishes mathematical model based on qi and blood resonance theory between obesity and body viscera health condition, realizes the automatic evaluation of the influence of obesity on viscera health, alleviates the problem of medical resource shortage, reduces medical public expenditure, and improves the degree of automation of detection.

[0029] 2, the present application adopts optical sensing signal to collect human health information, and is convenient and simple to operate, and the application range of the collected mobile terminal is wide, including but not limited to blood oxygen meter, mobile phone, bracelet, intelligent wear and other equipment.

[0030] 3, the present application adopts qi and blood resonance principle to establish data model, and invites more than 10 experts to verify the reliability of pulse diagnosis result, calculates the harmonic frequency of each level through Fourier transform, the first harmonic frequency is 0.6-1.8Hz, which is the qi and blood vibration frequency of liver, the second harmonic frequency is 1.8-2.8Hz, which is the qi and blood vibration frequency of kidney, the third harmonic frequency is 2.8-4.2Hz, which is the qi and blood vibration frequency of spleen, the fourth harmonic frequency is 3.6-5.6Hz, which is the lung vibration frequency, the fifth harmonic frequency is 4.6-7Hz, which is the vibration frequency of stomach, and the sixth harmonic frequency is 5.6-8Hz, which is the vibration frequency of gallbladder.

[0031] 4, the present application collects PPG signal of obese population, based on the basic data of healthy population, carries out amplitude data change calculation, when the amplitude change value is larger, the health state is more serious. The detection method is simple to operate, the data is clear and clear, and is easy to popularize and apply. DETAILED DESCRIPTION

[0032] Figure 1 The whole flow chart of detection method Figure 2 Wavelet transform denoising result graph Figure 3 Fourier transform normalized amplitude spectrum Specific implementation method

[0033] The technical solutions in the present application will be described clearly and completely below with reference to the drawings in the present application, and additionally, the forms of each structure described in the following embodiments are only examples, and the method for detecting the influence of obesity on the health of viscera based on a PPG signal of a mobile terminal involved in the present application is not limited to the following embodiments. All other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present application.

[0034] Referring to Figure 1 The present application provides a method for detecting the influence of obesity on the health of viscera based on a PPG signal of a mobile terminal, comprising the following steps: Step one: Collect original digital signals by using a photoplethysmography (PPG) signal. The mobile terminal for collection includes but is not limited to a blood oxygen meter, a mobile phone, a bracelet, a smart wearable device, etc. The signal-to-noise ratio of the PPG signal collected by different mobile terminals will be different, but it does not affect the final test result. In order to obtain an accurate pulse signal, the pulse signal collection time is not less than 15s, and more preferably more than 30s; Step two: The original digital signals collected are subjected to mean normalization processing. The original signals are subjected to mean normalization processing to avoid data inconsistency caused by different collection methods; Step three: The normalized digital signals are subjected to signal separation analysis to generate independent source signals. Since the original PPG signals collected will have various noise interferences, it is necessary to extract the principal component signals to reduce the noise interferences; The normalized digital signals are subjected to signal separation analysis, and the signal separation methods include but are not limited to an independent component analysis (ICA) method, a fast independent component analysis (FastICA) method, a convolutional neural network analysis method, a Haar-like feature combined with an AdaBoost classifier analysis method, etc. Step four: The independent source signals are subjected to screening, and a Butterworth low-pass filter is used to eliminate power frequency interference and electromyographic noise high-frequency noise; Step five: A wavelet function is used to perform 9-layer wavelet decomposition on the pulse signals, and the wavelet details of high-frequency noise greater than 20Hz are determined according to the wavelet decomposition frequency spectrum; The wavelet transform decomposes the waveform to strengthen the research and analysis of signal details, and the signals are decomposed by 9 layers. The high-frequency part and the low-frequency part of the signals are superimposed, and are represented by the following formula: S=c A 1+c D 1=c A 2+c D 2+c D 1=...=c A 9+c D 9+cD 8+...+c D 2+c D 1 Step six: filter out high-frequency wavelet coefficients and low-frequency wavelet details, reconstruct the signal, and obtain the denoised pulse signal; Referring to Figure 2 After 9 layers of wavelet transform decomposition, high-frequency noise in the frequency range of the first layer and the second layer is filtered out, and baseline drift wavelet details in the frequency range of the eighth layer and the ninth layer are filtered out, and the filtered pulse signal can be obtained through wavelet reconstruction.

[0035] The stable pulse signal of the application is mainly distributed within 20Hz, and the high-frequency noise greater than 20Hz and the low-frequency baseline drift noise are filtered out to obtain the final pulse signal of the application; Step seven: uniformize the denoised pulse signal to obtain the average pulse waveform; Step eight: Fourier transform harmonic analysis is performed on the collected uniformized waveform pulse signal to obtain a Fourier spectrum containing phase and amplitude information; Step nine: referring to Figure 3 The amplitude is obtained by the Abs function, and the signal in the frequency range of 20Hz is intercepted to obtain the normalized amplitude information; Step ten: based on the Fourier transform results of the pulse of the healthy population and the gas-blood resonance principle, a basic database and a corresponding viscera data model are established; the viscera data model of the healthy population is collected, the age of the population is 18-25 years old, 100 healthy samples of male with body weight index BMI between 18.5-24 are collected, and 100 healthy samples of female with body weight index BMI between 18.5-24 are collected. The healthy samples collected from the healthy population are invited to at least 10 expert judges to determine their health constitution by pulse diagnosis; Step eleven: according to the viscera data model of the healthy population, the health condition of the obese population is automatically evaluated.

[0036] After Fourier transform of the collected pulse signal, normalized amplitude data is obtained, and the amplitude size change of the obese population is calculated based on the amplitude basic data of the healthy population.

[0037] 10%≤when the amplitude data changes<20%, it is determined that the health condition is slightly affected by obesity.

[0038] 20%≤when the amplitude data changes<30%, it is determined that the health condition is generally affected by obesity.

[0039] 30%≤when the amplitude data changes<40%, it is determined that the health condition is greatly affected by obesity.

[0040] 40%≤amplitude data change < 50%, then the health condition is determined to be greatly affected by obesity.

[0041] When the amplitude data change ≥ 50%, the health condition is determined to be extremely seriously affected by obesity.

[0042] When the amplitude data change < 10%, considering the fluctuation of blood movement, it is determined that obesity has little effect on health in this range.

[0043] According to the principle of blood resonance, a data model is established, and more than 10 experts are invited to verify the pulse diagnosis. The Fourier transform is used to calculate the harmonic frequency. The first harmonic frequency is 0.6-1.8 Hz, which is the blood vibration frequency of the liver. The second harmonic frequency is 1.8-2.8 Hz, which is the blood vibration frequency of the kidney. The third harmonic frequency is 2.8-4.2 Hz, which is the blood vibration frequency of the spleen. The fourth harmonic frequency is 3.6-5.6 Hz, which is the lung vibration frequency. The fifth harmonic frequency is 4.6-7 Hz, which is the stomach vibration frequency. The sixth harmonic frequency is 5.6-8 Hz, which is the bile vibration frequency.

[0044] Table 1 Data test results are confirmed by 10 expert judges for pulse diagnosis, and the detection results of the present application are basically consistent, which shows that the data model established based on the principle of blood resonance is feasible, simple and easy to popularize and apply.

[0045] Table 1 BMI body weight ≥ 28 obese population affects the health of viscera FFT spectrum information First harmonic (Hz) / amplitude change (liver) Second harmonic (Hz) / amplitude change (kidney) Third harmonic (Hz) / amplitude change (spleen) Fourth harmonic (Hz) / amplitude change (lungs) Fifth harmonic (Hz) / amplitude change (stomach) Sixth harmonic (Hz) / amplitude change (gall bladder) Sample 1 -10% -20% -30% +5% +2% +8% Sample 2 +5% -33% -25% -35% +8% -10% Sample 3 +13% -15% -35% -10% -10% +17% Sample 4 -9% -40% -20% +20% +20% -15% Sample 5 +30% -26% -10% -18% +12% +18% Sample 6 -10% -15% -40% +25% -32% -20%

Claims

1. A method for detecting the influence of obesity on the health of Zangfu organs based on mobile terminal PPG signals, characterized in that, It comprises the following steps: Step one: collect original digital signals by using the photoplethysmography (PPG) signal acquisition method; Step two: perform mean normalization processing on the collected original digital signals; Step three: perform signal separation analysis on the normalized digital signals to generate independent source signals; Step four: filter out power frequency interference and electromyographic noise high-frequency noise by using a Butterworth low-pass filter to screen the independent source signals; Step five: perform 9-layer wavelet decomposition on the pulse signals by using a wavelet function, and determine the high-frequency noise wavelet details greater than 20 Hz according to the wavelet decomposition frequency spectrum; Step six: filter out high-frequency wavelet coefficients and low-frequency wavelet details, reconstruct the signals, and obtain the denoised pulse signals; Step seven: perform homogenization on the denoised pulse signals to obtain the average pulse waveform; Step eight: perform Fourier transform harmonic analysis on the collected homogenized waveform pulse signals to obtain a Fourier spectrum containing phase and amplitude information; Step nine: obtain the normalized amplitude information by using the Abs function to intercept signals within a frequency range of 20 Hz; Step ten: based on the Fourier transform results of the pulses of healthy people and the gas-blood resonance principle, establish a basic database and a corresponding zang-fu data model; Step eleven: automatically assess the zang-fu health status of obese people according to the zang-fu data model of healthy people.

2. The detection method according to claim 1, characterized in that, The photoplethysmography (PPG) signal acquisition method is used to collect original digital signals. The mobile terminal includes but is not limited to a blood oxygen meter, a mobile phone, a bracelet, a smart wearable device, and the like.

3. The method of claim 1, wherein The mean normalized digital signals are subjected to signal separation analysis. The signal separation methods include but are not limited to the independent component analysis (ICA) method, the fast independent component analysis (FastICA) method, the convolutional neural network analysis method, the Haar-like feature combined with the AdaBoost classifier analysis method, and the like.

4. The method of claim 1, wherein In order to obtain accurate pulse signals, the pulse signal collection time is not less than 15s, and more preferably, the collection time is more than 30s.

5. The method of claim 1, wherein Wavelet transform decomposes the waveform, strengthens the research and analysis of signal details, and performs 9-layer decomposition on the signals. The signal high-frequency part and the low-frequency part are superimposed, and are represented by the following formula: S=c A 1+c D 1=c A 2+c D 2+c D 1=...=c A 9+c D 9+c D 8+...+c D 2+c D 1 After 9-layer wavelet decomposition, the high-frequency noise in the frequency range of the first layer and the second layer is filtered out, and the baseline drift wavelet details in the frequency range of the eighth layer and the ninth layer are filtered out. The filtered pulse signals can be obtained by wavelet reconstruction.

6. The method of claim 1, wherein The zang-fu data model of healthy people is collected. The age of the collected population is 18-25 years old. 100 healthy samples of male with a body mass index (BMI) between 18.5 and 24 are collected, and 100 healthy samples of female with a body mass index (BMI) between 18.5 and 24 are collected. The collected healthy samples of healthy people invite at least 10 expert judges to perform pulse diagnosis to determine their health constitution.

7. The method of claim 1, wherein, According to the principle of Qi and blood resonance, the data model was established, and more than 10 experts were invited to verify the pulse diagnosis. The Fourier transform was used to calculate the harmonic frequency. The first harmonic frequency of 0.6-1.8 Hz was the Qi and blood vibration frequency of liver, the second harmonic frequency of 1.8-2.8 Hz was the Qi and blood vibration frequency of kidney, the third harmonic frequency of 2.8-4.2 Hz was the Qi and blood vibration frequency of spleen, the fourth harmonic frequency of 3.6-5.6 Hz was the lung vibration frequency, the fifth harmonic frequency of 4.6-7 Hz was the stomach vibration frequency, and the sixth harmonic frequency of 5.6-8 Hz was the gallbladder vibration frequency.

Images