An abdominal component analyzer and its analytical method

Abstract

This invention discloses an abdominal composition analyzer and its analysis method. First, an abdominal composition analysis model is established using the maximum a posteriori estimation method. Then, the model is trained by detecting the abdominal bioelectrical impedance of a human sample. Finally, the abdominal bioelectrical impedance of the subject is fed into the trained model to obtain the abdominal composition content. Simultaneously, during the measurement of abdominal bioelectrical impedance, multi-frequency detection and self-calibration methods are employed to improve the accuracy of impedance data measurement. Experimental results show that the measurement error for the human abdomen can be controlled within 10%. Furthermore, compared to doctors and ordinary large-scale body composition analyzers, the abdominal fat analyzer is more targeted, more compact, and offers advantages such as being non-invasive, safe, convenient, and inexpensive, making it a feasible solution for home testing.

Description

Technical Field

[0001] This invention relates to the field of analytical instrument technology, specifically to an abdominal component analyzer and its analytical method. Background Technology

[0002] As people's living standards gradually improve, obesity has become a serious problem in modern society. Those with abdominal obesity, in particular, are prone to arteriosclerosis, general weakness, bloating, indigestion, poor mental health, and poor sleep quality. They are also susceptible to cardiovascular and cerebrovascular diseases such as hypertension, hyperlipidemia, diabetes, fatty liver, and coronary heart disease. In severe cases, these conditions can develop into myocardial infarction, cerebral infarction, stroke, hemiplegia, and cirrhosis. Therefore, it is necessary to use specialized equipment to detect and analyze the composition of the abdominal area to help people understand their own physical condition.

[0003] Because a balanced and standardized distribution of body composition improves human health, accurately and quickly measuring body composition can help individuals assess their health status. At the molecular level, the human body is composed of water, protein, fat, vitamins, minerals, and cellulose. Water accounts for approximately 60%, carbohydrates and fats for about 14%, and protein for about 17%. The body fluids contain many ions, giving the body electrical conductivity, which in turn gives it resistive properties. Human cells have a double membrane, with conductive body fluid on either side and a non-conductive medium between them—a structure similar to a capacitor in a circuit. Although the human body can be simplified into an impedance-capacitive model, exhibiting different impedance characteristics at different frequencies, theoretically providing information about physiological characteristics, the complexity of the human body makes it impossible to accurately measure the impedance of a single part. Therefore, accurately obtaining the impedance of the abdominal region is a key problem that needs to be solved first. Summary of the Invention

[0004] The present invention addresses the problem of detecting and analyzing abdominal components by providing an abdominal component analyzer and its analytical method.

[0005] To solve the above problems, the present invention is achieved through the following technical solution:

[0006] A method for analyzing abdominal components includes the following steps:

[0007] Step 1: Establish an abdominal component analysis model using the maximum a posteriori estimation method:

[0008]

[0009] In the formula, fat represents abdominal fat content, water represents abdominal water content, protein represents abdominal protein content, inSalt represents abdominal inorganic salt content, muscle represents abdominal muscle content, height represents height, weight represents body weight, BMI represents body mass index, impedance1, impedance2 and impedance3 represent the impedance values ​​of the abdomen measured at three different frequencies, w is the weight vector, and δ is the expected error vector of the model.

[0010] Step 2: For sample sets composed of different populations, measure the basic parameters, abdominal component content and abdominal impedance value of each sample to obtain the sample dataset corresponding to the sample set.

[0011] Step 3: Use the sample dataset obtained in Step 2 to train the abdominal component analysis model constructed in Step 1. Determine the weight vector and expected error vector of the abdominal component analysis model through the maximum a posteriori estimation method to obtain the trained abdominal component analysis model.

[0012] Step 4: For the subject, measure the subject's basic parameters and abdominal impedance value;

[0013] Step 5: Input the basic parameters of the subject and the abdominal impedance value obtained in Step 5 into the abdominal component analysis model trained in Step 3 to obtain the abdominal component content of the subject.

[0014] The above basic parameters include height, weight, and body mass index. Abdominal composition includes abdominal fat content, abdominal water content, abdominal protein content, abdominal inorganic salt content, and abdominal muscle content. Abdominal impedance values ​​include abdominal impedance values ​​measured at three different frequencies.

[0015] In steps 2 and 4 above, the method for measuring the impedance value of the abdomen is as follows:

[0016] Step a: Apply three sinusoidal excitation signals of different frequencies to the abdomen of the human body. The abdomen generates a corresponding electric field under the excitation of the sinusoidal excitation signals. Collect the electric field to obtain the measurement signal of the abdomen.

[0017] Step b: After amplifying, conditioning, and amplitude and phase detection of the measurement signal, the sampling impedance value of the abdomen is obtained by analog-to-digital conversion;

[0018] Step c: Perform digital low-pass filtering on the sampled impedance value of the abdomen, and use the minimum mean square error to determine its steady state, thereby obtaining the impedance value of the abdomen.

[0019] In step a above, the three frequency sinusoidal excitation signals are sinusoidal excitation signals in the frequency range of 1KHz to 1MHz.

[0020] An abdominal component analyzer for implementing the above-mentioned abdominal component analysis method comprises a constant current source generating circuit, a signal acquisition circuit, a signal amplification circuit, a signal conditioning circuit, an amplitude and phase detection circuit, an AD data acquisition circuit, a microprocessor, and a data processing platform. The control terminal of the microprocessor is connected to the constant current source generating circuit. The output terminal of the constant current source generating circuit is in contact with the human abdomen, and the input terminal of the signal acquisition circuit is in contact with the human abdomen, with a certain distance between the output terminal of the constant current source generating circuit and the input terminal of the signal acquisition circuit. The output terminal of the signal acquisition circuit is connected to the input terminal of the microprocessor after passing through the signal amplification circuit, the signal conditioning circuit, the amplitude and phase detection circuit, and the AD data acquisition circuit in sequence. The output terminal of the microprocessor is connected to the data processing platform.

[0021] The output of the above constant current source generating circuit consists of three sinusoidal excitation signals in the frequency range of 1KHz to 1MHz.

[0022] The aforementioned constant current source generating circuit consists of a DSS signal generator, an amplification and filtering circuit, and a voltage-controlled constant current source circuit. The DSS signal generator further comprises a phase accumulator, a waveform memory, a digital-to-analog converter (DAC), and a low-pass filter. The microprocessor's control terminal is connected to the input terminal of the phase accumulator, as well as the clock control terminals of the phase accumulator and the DAC. The output terminal of the phase accumulator is connected to the input terminal of the waveform memory, the output terminal of the waveform memory is connected to the input terminal of the DAC, and the output terminal of the DAC is connected to the input terminal of the low-pass filter. The output terminal of the low-pass filter is connected to the input terminal of the amplification and filtering circuit, the output terminal of the amplification and filtering circuit is connected to the input terminal of the voltage-controlled constant current source circuit, and the output terminal of the voltage-controlled constant current source circuit forms the output terminal of the constant current source generating circuit.

[0023] The signal conditioning circuit described above consists of a notch filter and a bandpass filter; the input terminal of the notch filter forms the input terminal of the signal conditioning circuit, the output terminal of the notch filter is connected to the input terminal of the bandpass filter, and the output terminal of the bandpass filter forms the output terminal of the signal conditioning circuit.

[0024] The aforementioned abdominal component analyzer further includes an impedance network for calibration; the impedance network is composed of two or more impedance matching branches connected in parallel, each impedance matching branch consisting of a precision resistor and a switch connected in series; the two ends of the impedance network are connected in parallel to the output of the constant current source generating circuit.

[0025] Compared with existing technologies, this invention first uses a maximum a posteriori estimation method to analyze abdominal composition, then trains the model by detecting the abdominal bioelectrical impedance of human samples, and finally feeds the abdominal bioelectrical impedance of the tested person into the trained model to obtain the abdominal composition content of the tested person. Simultaneously, during the measurement of abdominal bioelectrical impedance, multi-frequency detection and self-calibration methods are used to improve the accuracy of impedance data measurement. Experimental results show that the measurement error for the human abdomen can be controlled within 10%. Furthermore, compared with doctors and ordinary large-scale body composition analyzers, the abdominal fat analyzer is more targeted, more compact, and has the advantages of being non-invasive, safe, simple, and inexpensive, making it a feasible solution for home testing. Attached Figure Description

[0026] Figure 1 This is a single-cell impedance model.

[0027] Figure 2 for Figure 1 A simplified model.

[0028] Figure 3 This is a system block diagram of an abdominal component analyzer.

[0029] Figure 4 This is a schematic diagram of a constant current source generation circuit.

[0030] Figure 5 This is a schematic diagram of a DSS signal generator.

[0031] Figure 6 This is the schematic diagram of a voltage-controlled constant current source circuit.

[0032] Figure 7 This is a schematic diagram of a signal conditioning circuit.

[0033] Figure 8 This is a schematic diagram of the amplitude detection circuit.

[0034] Figure 9 This is a schematic diagram of a phase detection circuit.

[0035] Figure 10 This is a diagram showing the voltage relationship between the output of the amplitude detection circuit.

[0036] Figure 11 The output voltage relationship diagram for the phase detection circuit.

[0037] Figure 12 This is a schematic diagram of an impedance network.

[0038] Figure 13 This is the impedance response curve of the human abdomen.

[0039] Figure 14 This is a graph showing the result of digital low-pass filtering.

[0040] Figure 15 This is a schematic diagram of a data queue.

[0041] Figure 16 This is a flowchart for the calibration process.

[0042] Figure 17 This is a flowchart of the abdominal component analysis method. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific examples.

[0044] Abdominal component analyzers are designed to detect the impedance of the human abdomen to obtain information on abdominal fat, overall fat, water content, and other components. The principle is similar to measuring the resistivity of a conductor in electrical engineering. Since every conductor has a specific resistivity, for a material of known length, once its resistance is measured, the resistivity can be calculated, and the material composition can be determined. The volume of the conductor can be considered as the human body, and the length of the conductor can be represented by the length of human tissue. Resistivity is the resistance per unit volume of human tissue. By measuring the impedance of the human body, the content of relevant components can be determined. Fat in the human body is a weak conductor of electricity, while muscle and water are strong conductors. If there is a high fat content and a low muscle content, the bioresistivity is relatively high when current passes through; conversely, the bioresistivity is relatively low. Based on this information, mathematical models can be established according to different ages, genders, etc., to quantitatively analyze and obtain body components such as trunk fat content and water content.

[0045] Biological tissues are composed of cells, and the main components of cells are the cell membrane and cellular fluid. Different cells also have extracellular fluid and intercellular matrix. From an electrochemical perspective, the human body can be represented as an impedance model. Figure 1 As shown. At frequencies below 1MHz, the equivalent resistance of the cell membrane can be considered an open circuit, therefore it can be... Figure 1 The model is simplified, such as Figure 2 As shown. The simplified expression is as follows:

[0046]

[0047] Where w is the angular frequency, Cm is the cell membrane capacitance, Ri is the intracellular fluid resistance, and Re is the extracellular fluid resistance.

[0048] The impedance magnitude is:

[0049]

[0050] The phase angle is:

[0051]

[0052] The purpose of this invention is to measure the above two parameters.

[0053] In a healthy human body, water is distributed relatively constantly both inside and outside cells. However, in elderly patients, those lacking nutrition, and those with heart disease, the water balance in the body is disrupted. This is because the intracellular and extracellular fluids have different electron sensitivities; the extracellular fluid has a much higher electron sensitivity than the intracellular fluid.

[0054] The electrical properties of biological tissues exhibit different patterns across different frequency bands. High-frequency electronic signals measure the total impedance of the human body, reflecting the combined value of intracellular and extracellular fluids, while low-frequency electronic signals only reflect the resistance of the extracellular fluid. Because the components that pass through human tissues differ depending on the frequency of the excitation current, low-frequency signals cannot penetrate cells, measuring only the extracellular tissue; mid-frequency signals can penetrate some cells; and high-frequency signals can directly penetrate cells. In the alpha band, the electrical properties of biological tissues are primarily related to cell membrane properties. In the beta band, the properties of the cellular fluid (including intracellular and extracellular fluids) are reflected. Frequency greater than 1 MHz enters the gamma band, which reflects the state of water molecules. Clinically, most pathological conditions occur only in the alpha and beta bands. To ensure that the excitation can penetrate the cell membrane, a signal with a frequency greater than 1 kHz must be selected.

[0055] Most current lipid analyzers use a single 50kHz frequency signal. Due to the limited frequency and input parameters, they cannot accurately reflect the condition of various tissues in the human body. Multi-frequency bioelectrical impedance analysis (BIA) technology can measure both intracellular and extracellular fluid, thus effectively addressing this problem. Therefore, this invention uses three sinusoidal signals within the 1kHz–1MHz frequency range to measure the impedance of different tissue components in the human abdomen. This yields more targeted results, and self-calibration further improves measurement accuracy.

[0056] The present invention provides an abdominal component analyzer, such as... Figure 3As shown, the system consists of a constant current source generating circuit, an impedance acquisition circuit, a signal amplification circuit, a signal conditioning circuit, an amplitude and phase detection circuit, an AD data acquisition circuit, a microprocessor, and a data processing platform. The control terminal of the microprocessor is connected to the constant current source generating circuit. The output terminal of the constant current source generating circuit is in contact with the abdomen, and the input terminal of the impedance acquisition circuit is also in contact with the abdomen, with a certain distance between them. The output terminal of the impedance acquisition circuit is connected to the input terminal of the microprocessor after passing through the signal amplification circuit, signal conditioning circuit, amplitude and phase detection circuit, and AD data acquisition circuit. The output terminal of the microprocessor is connected to the data processing platform. The microprocessor controls the constant current source generating circuit to generate a constant current excitation signal, which is applied to the measured part of the human body (abdomen). This generates a corresponding electric field at the measured part. The measurement signal of the measured part is obtained by acquiring this electric field signal. After amplification and conditioning, this measurement signal undergoes amplitude and phase detection and is transmitted to the AD acquisition module, and then sent to the microprocessor. The microprocessor then measures the impedance value of the human abdomen. The impedance value of the human abdomen is sent to the data processing platform for data processing, and the corresponding human body composition content is calculated.

[0057] The constant current source generation circuit consists of a DSS signal generator, an amplification and filtering circuit, and a voltage-controlled constant current source circuit, such as... Figure 4 As shown. The DSS signal generator further consists of a phase accumulator, waveform memory, digital-to-analog converter, and low-pass filter, as follows: Figure 5As shown. The microprocessor's control terminal is connected to the input of the phase accumulator. The frequency control code output by the microprocessor is used to set the jump step size and control the frequency of the signal. The microprocessor's control terminal is connected to the clock control terminal of the phase accumulator and the digital-to-analog converter (DAC). The fclk signal output by the microprocessor is the reference clock of the DAC, and one point is output after each clock cycle. The output of the phase accumulator is connected to the input of the waveform memory, the output of the waveform memory is connected to the input of the DAC, and the output of the DAC is connected to the input of the low-pass filter. The output of the low-pass filter is connected to the input of the amplification and filtering circuit, the output of the amplification and filtering circuit is connected to the input of the voltage-controlled constant current source (VDC) circuit, and the output of the VDC circuit forms the output of the constant current source generation circuit. The phase accumulator consists of an N-bit adder and an N-bit accumulator register. It accumulates the phase value according to the frequency control code and inputs the accumulated value into the waveform memory. The waveform memory uses the value of the phase accumulator as the address to find the signal data corresponding to the phase value and outputs it to the DAC. A digital-to-analog converter (DAC) converts the digital output from the waveform memory into a corresponding analog signal. Due to quantization errors, aliasing occurs in the output waveform, necessitating a low-pass filter at the output to improve signal performance. The frequency output formula for a DDS signal generator is: fout = fclk * k / 2^n, where k is the frequency control word, and the output frequency resolution is fclk / 2^n. The signal output from the DDS generator may not be the desired amplitude, and bias voltage may also exist. Therefore, an amplification and filtering circuit is added after the DDS to remove bias and noise. Having obtained the desired frequency and amplitude, the next step is to ensure a constant current output. Finally, a voltage-controlled constant current circuit is added to guarantee the load-carrying capacity of the excitation signal. Figure 6 The circuit shown is a voltage-controlled constant current circuit.

[0058] The errors of the constant current excitation circuit designed based on this circuit under different load conditions are shown in Table 1 below:

[0059] Table 1. Errors of output signals under different loads

[0060]

[0061] As can be seen from the table above, the signal output by the voltage-controlled constant current circuit exhibits very small errors under different load conditions. This indicates that the signal output by the voltage-controlled constant current circuit meets our requirements. Although the error is already very small, showing less than 2% as indicated in the table, the trend of error change with increasing impedance shows a correlation. This error can be further corrected in the program. This also further ensures the accuracy of the measured impedance.

[0062] A microprocessor is used as the main control chip. The output of the constant current source generating circuit is in contact with the human abdomen to generate an alternating excitation signal that meets the requirements. The input of the impedance acquisition circuit is in contact with the human abdomen to acquire the voltage fed back from the human abdomen, thereby obtaining the impedance of the human abdomen.

[0063] Traditional methods for determining impedance based on voltage and current magnitudes can lead to discrepancies between the applied excitation signal and the calculated value due to signal attenuation within the circuit system. Table 1 also shows that the accuracy of the output current varies slightly depending on the load. Furthermore, the voltage magnitude acquired by the ADC and the ADC value do not necessarily correspond to zero-crossing points, potentially resulting in some offset. Therefore, using traditional methods based on acquired voltage and known current to determine impedance values ​​may introduce significant errors.

[0064] Since the relationship between impedance and voltage is linear, and the errors in Table 1 are also positively correlated, this invention uses a method of direct calculation based on the ADC value to determine the impedance. The calculation formula is as follows:

[0065] Z = K × ADC + ε

[0066] In the formula, K and ε are the slope and offset, respectively.

[0067] After the data is acquired, the measurement signal needs to be amplified. This invention uses a differential amplifier circuit to amplify the measurement. The differential amplifier circuit converts the sampled differential two-ended signal into a single-ended signal.

[0068] The amplified measurement signal needs to be conditioned. The conditioning flowchart is as follows: Figure 7 As shown, the signal conditioning circuit consists of a notch filter and a bandpass filter. The input of the notch filter forms the input of the signal conditioning circuit, and the output of the notch filter is connected to the input of the bandpass filter. The output of the bandpass filter forms the output of the signal conditioning circuit. The purpose of the notch filter is to avoid the influence of the 50Hz mains power. It mainly filters this frequency, with significant attenuation only around 50Hz. This ensures that only the mains frequency is filtered out without affecting signals of other frequencies; this is the main function of the notch filter. The bandpass filter can filter out high-frequency noise and some DC bias. For the low-frequency part, in addition to filtering out the DC component, it is also necessary to consider the subtle signal fluctuations caused by blood flow pulsation. Here, in order to obtain a more accurate and stable impedance value, this part of the signal variation needs to be eliminated. For the high-frequency part, the main component is noise, and the main thing to filter out is external interference noise.

[0069] The amplitude and phase detection section primarily detects the amplitude and phase of the signal. It includes amplitude detection circuitry and phase detection circuitry, such as... Figure 8 , Figure 9 As shown, a logarithmic detector is used for amplitude and phase detection. Two measurement signals, INPA and INPB, are input, and the ratio of the power of the two signals in decibels and the phase between them are output through VMAG and VPHS.

[0070] According to the principle of logarithmic detection, the formulas for calculating its amplitude and phase are as follows:

[0071] V MAG =(R F I SLP / 20)(P INA -P INB )+V CP

[0072] V PHS =-R F I Φ (|Φ(V INA )-Φ(V INB )|-90°)+V CP

[0073] In the above equation, PINA and PINB are the equivalent power in logarithmic units of VINA and VINB at a specified reference impedance. For the gain function, R is used. F I SLP The slope is 600mV / 10 GHz, divided by 20dB / 10 GHz equals 30mV / dB. With Vcp = 900mV as the midpoint, the voltage range from -30dB to +30dB is 0-1.8V. The slope of the phase function represents R... F I Φ This represents 10mV / degree, with 900mV as the midpoint, corresponding to 90 degrees; 0-180 degrees correspond to 1.8V-0V; the voltage range from 0 to -180 degrees is the same, but the slopes are opposite, where φ is the phase degree corresponding to each signal. For example... Figure 10 , Figure 11 As shown.

[0074] In this invention, the accuracy of impedance measurement has a significant impact on the results. Therefore, a self-calibration process is added to improve the accuracy of impedance measurement. Many factors can affect the accuracy of impedance measurement, such as the influence of the components themselves: due to manufacturing processes, even components of the same model from the same manufacturer will have slight differences. Another factor is the influence of frequency variations: because the human body's frequency response differs at different frequencies, changes in measurement frequency will inevitably cause changes in system parameters. Since the human body is a complex system, self-calibration is essential to minimize the differences in each measurement. To ensure the accuracy of impedance measurement, the system parameters (K, ε) need to be calibrated before the actual measurement.

[0075] To meet the error correction requirements of multiple precision resistors, an impedance network for correction needs to be added to the signal amplification circuit. This impedance network consists of two or more impedance matching branches connected in parallel, each branch consisting of one precision resistor and one switch connected in series. The two ends of the impedance network are connected in parallel to the output of the constant current source circuit, and the impedance network is connected to the output of the excitation signal. An analog switch is used to switch between measuring the correction network or the impedance of the human abdomen. For example... Figure 12 As shown.

[0076] After the aforementioned signal excitation, signal acquisition, and signal processing, the measured impedance value of the human abdomen can be read out by a microprocessor. However, given the complexity of the human body, directly reading the data still presents some problems. Considering the human body's impedance as a system, the human body also has its own frequency response. Therefore, the impedance value obtained at the instant the excitation is applied, due to the body's own frequency response, cannot directly reflect the true state of the human body. Only when the human body's frequency response reaches a steady state does the impedance value read out become meaningful.

[0077] like Figure 13 The figure shows the impedance response curve obtained by the microprocessor through the serial port and sent to the host computer, i.e., the data processing platform. The solid line represents the measured ADC value, the short dashed line represents the value after digital low-pass filtering, and the long dashed line represents the calculated impedance value. It is clear from the figure that the values ​​read after the system reaches stability are meaningful; otherwise, the values ​​read during the dynamic response period are not representative and cannot meet consistency requirements.

[0078] Now, the key is determining whether the response curve has reached a steady state. Here, we implement the principle of a low-pass filter using a digital algorithm.

[0079] The transfer function of a first-order passive low-pass filter in analog signals is known to be:

[0080]

[0081] This is the transfer function in the S-domain, and it must be converted to a discrete signal before it can be applied to a computer. Converting a continuous signal to a discrete signal can be achieved through the Z-transform. After the Z-transform, we obtain:

[0082]

[0083] Where T is the sampling period. This allows the above equation to be transformed into a difference equation:

[0084]

[0085] Where Y(n) is the output of the current filter, X(n) is the current sampled value, and Y(n-1) is the output value of the previous filter. It is easy to see from the above formula that T / (T+RC) and RC / (T+RC) are complementary, thus the digital filter algorithm can be simplified to:

[0086] Y(n) = a*X(n) + (1-a)*Y(n-1)

[0087] Where 'a' is the filter coefficient, adjusting this parameter to a suitable value will yield the following result: Figure 14 The results are shown in the figure. It can be observed from the figure that the dashed line represents the result after digital low-pass filtering. This makes it easier to determine when the signal enters a steady state compared to the solid line without digital low-pass filtering.

[0088] When the system reaches steady state, the sampled data remains essentially unchanged. Based on this idea, a linear linked list is created using data structure algorithms. The system determines whether the data is in steady state when the error successively approaches zero. Figure 15 The diagram shows the process of data being cyclically entered into the queue.

[0089] The formula for determining steady state can be defined by the minimum mean square error (LMS). For ease of calculation, the minimum absolute error (LAD) is used to define the steady-state criterion function. The condition for determining steady state is as follows:

[0090]

[0091] In the formula, n is the number of data in the queue, and ξ is the defined minimum steady-state error. The accuracy of steady-state determination can be adjusted by changing n and ξ.

[0092] The voltage levels on the human abdomen were measured using an impedance network and corrected using the least squares method. See also Figure 16 The specific correction method is as follows:

[0093] The voltage across the resistor is sampled by an ADC, and the resulting values ​​are ADC1, ADC2, ... . The slope relationship between the voltage and the resistance, and its offset, can be determined using maximum likelihood estimation.

[0094]

[0095] The challenge in ensuring the accuracy of human body resistance measurements lies in how to make the aforementioned variables adapt to changes in accordance with the complexity of human body impedance. During the measurement process, a queue is established to detect error signals in real time, and the errors in the queue are calculated to ensure that measurement differences can be detected in real time. If the detected measurement error exceeds a set threshold, the system parameters are recalculated and corrected to adapt to changes caused by the complexity of the human body, and the new corrected system parameters are pushed onto a stack to recalculate a more accurate human body impedance value.

[0096] Choose one such Figure 12 The impedance network was constructed, and the impedance values ​​of the impedance network at four different frequencies were measured. The error results before and after correction were compared, as shown in Table 2.

[0097] Table 2 Comparison of results before and after correction

[0098]

[0099] Comparing the error results before and after correction in the table above, it is clear that introducing a self-correction method significantly improves the accuracy of the measurement.

[0100] A model for abdominal component analysis using the MAP (Maximum A posteriori) estimation method is established. MAP estimation is a more advanced estimation method than ML estimation. Considering the posterior density π(w|d,x), the MAP estimate of the parameter vector w can be defined by the following equation:

[0101] w MAP =argmaxπ(w|d,x)

[0102] Where w is the weight vector, d is the expected output value representing one of the following parameters: abdominal fat, water content, trunk muscle, inSalt, or protein content, which is measured by standard medical instruments; x is the input vector, representing a person's height, weight, BMI (BMI = weight (kg) divided by the square of height (m), and impedance values ​​measured at different frequencies (impedance1, impedance2, impedance3).

[0103] The formula for calculating body composition is:

[0104]

[0105] In the formula, fat represents abdominal fat content, water represents abdominal water content, protein represents abdominal protein content, inSalt represents abdominal inorganic salt content, muscle represents abdominal muscle mass, height represents height, weight represents body weight, BMI represents body mass index, and impedance1, impedance2, and impedance3 represent the impedance values ​​measured at three different frequencies. For weight vectors, This represents the expected error vector of the model.

[0106] Using only impedance values ​​to estimate the content of various components in the human body cannot fully reflect the component content of the measured site. This is due to the complexity and holistic nature of the human body. A person's height, weight, and body mass index (BMI) are not only important indicators of health, but the electric field generated when stimulation is applied to human tissues is also influenced by the entire body. Therefore, the impedance obtained at the test site cannot accurately reflect the impedance value of that specific location; it is related to the entire body. Thus, height, weight, and BMI should be included as essential input variables when building a model. Using these variables as input variables can significantly improve measurement accuracy.

[0107] The formula for calculating the above body composition can be expanded as follows:

[0108] Abdominal fat = ω 11 *Height + ω 12 *Weight + ω 18 *BMI+ω 14 *Impedance 1+ω 18 *Impedance 2+ω 16 *Impedance 3+δ1

[0109] Abdominal water content = ω 21 *Height + ω 22 *Weight + ω 28 *BMI+ω 24 *Impedance 1+ω 28 *Impedance 2+ω 26 *Impedance 3+δ2

[0110] Abdominal protein = ω 31 *Height + ω 32 *Weight + ω 38 *BMI+ω 34 *Impedance 1+ω 38 *Impedance 2+ω 36 *Impedance 3+δ3

[0111] Abdominal inorganic salts = ω 41 *Height + ω 42*Weight + ω 48 *BMI+ω 44 *Impedance 1+ω 48 *Impedance 2+ω 46 *Impedance 3+δ4

[0112] Trunk muscles = ω 51 *Height + ω 52 *Weight + ω 58 *BMI+ω 54 *Impedance 1+ω 58 *Impedance 2+ω 56 *Impedance 3+δ5.

[0113] The general formula is as follows:

[0114]

[0115] Where δ represents the expected error of the model.

[0116] The challenge in the above model lies in selecting appropriate weights. Only with correct weights can the content information of various abdominal components be calculated more accurately. To minimize measurement errors, this invention uses supervised learning to determine the weights, with the training samples being data obtained from measurements using standard medical instruments. The model coefficients are determined using MAP estimation, an algorithm that incorporates an adjustable regularization parameter λ, which improves model accuracy. The specific parameter determination process is as follows:

[0117] Assume the training samples for parameter estimation conform to statistical independence and synchronous distribution, Gaussianity, and stability. The posterior density can be obtained from the Gaussian distribution function, and the MAP estimate can be expressed as:

[0118]

[0119] Where λ is the regularization parameter.

[0120] Define a quadratic function:

[0121]

[0122] Clearly, maximizing the parameters of w is equivalent to minimizing the quadratic function ξ(w). It's easy to see that the optimal estimate of w can be obtained by differentiating the quadratic function ξ(w) and setting the result to zero. MAP Based on this, the following MAP estimate of the parameter vector can be obtained:

[0123]

[0124] Where R is the autocorrelation matrix, r is the cross-correlation matrix, and I is the identity matrix.

[0125] After determining the specific model weights and expected error, the trained model can be obtained. Based on the abdominal impedance obtained from the detection, and according to the abdominal impedance acquisition method and abdominal model described above, the abdominal components can be obtained.

[0126] Due to the influence of temperature and humidity in the operating environment, the voltage values ​​of the same impedance collected by the system may vary under different conditions. Therefore, this invention includes a power-on self-test and system initialization to correct system parameters. Furthermore, as the equipment is used over time, errors in the circuit system will gradually accumulate. This invention also includes a correction option to allow users to correct data anomalies. The overall design flowchart is as follows: Figure 17 As shown, the specific steps are as follows:

[0127] Step 1: Establish an abdominal component analysis model using the maximum a posteriori estimation method:

[0128]

[0129] In the formula, fat represents abdominal fat content, water represents abdominal water content, protein represents abdominal protein content, inSalt represents abdominal inorganic salt content, muscle represents abdominal muscle content, height represents height, weight represents body weight, BMI represents body mass index, impedance1, impedance2, and impedance3 represent the abdominal impedance values ​​measured at three different frequencies, w is the weight vector, and δ is the expected error vector of the model.

[0130] Step 2: For sample sets composed of different populations, measure the basic parameters, abdominal component content and abdominal impedance value of each sample to obtain the sample dataset corresponding to the sample set.

[0131] Step 3: Use the sample dataset obtained in Step 2 to train the abdominal component analysis model constructed in Step 1. Determine the weight vector and expected error vector of the abdominal component analysis model through the maximum a posteriori estimation method to obtain the trained abdominal component analysis model.

[0132] Step 4: For the subject, measure the subject's basic parameters and abdominal impedance value.

[0133] Step 5: Input the basic parameters of the subject and the abdominal impedance value obtained in Step 5 into the abdominal component analysis model trained in Step 3 to obtain the abdominal component content of the subject.

[0134] The basic parameters mentioned above include height, weight, and body mass index. The abdominal component content mentioned above includes abdominal fat content, abdominal water content, abdominal protein content, abdominal mineral content, and abdominal muscle mass. The abdominal impedance values ​​mentioned above include abdominal impedance values ​​measured at three different frequencies.

[0135] The method for measuring the impedance value of the abdomen is as follows:

[0136] Step a: Apply three sinusoidal excitation signals of different frequencies to the abdomen of the human body. The abdomen generates a corresponding electric field under the excitation of the sinusoidal excitation signals. Collect the electric field to obtain the measurement signal of the abdomen.

[0137] Step b: After amplifying, conditioning, and amplitude and phase detection of the measurement signal, the sampling impedance value of the abdomen is obtained by analog-to-digital conversion;

[0138] Step c: Perform digital low-pass filtering on the sampled impedance value of the abdomen, and use the minimum mean square error to determine its steady state, thereby obtaining the impedance value of the abdomen.

[0139] It should be noted that although the embodiments described above are illustrative, they are not intended to limit the invention. Therefore, the invention is not limited to the specific embodiments described above. Any other embodiments obtained by those skilled in the art under the guidance of this invention without departing from its principles are considered to be within the protection scope of this invention.

Claims

1. A method for analyzing abdominal components, characterized in that, The steps include the following: Step 1: Establish an abdominal component analysis model using the maximum a posteriori estimation method: In the formula, Indicates abdominal fat content. Indicates abdominal water content. This indicates the protein content in the abdomen. This indicates the content of inorganic salts in the abdomen. Indicates abdominal muscle mass. Indicates height, represents weight, Indicates Body Mass Index (BMI). , and These represent the impedance values ​​of the abdomen measured at three different frequencies, where w is the weight vector. This represents the expected error vector of the model. Step 2: For sample sets composed of different populations, measure the basic parameters, abdominal component content and abdominal impedance value of each sample to obtain the sample dataset corresponding to the sample set. Step 3: Use the sample dataset obtained in Step 2 to train the abdominal component analysis model constructed in Step 1. Determine the weight vector and expected error vector of the abdominal component analysis model through the maximum a posteriori estimation method to obtain the trained abdominal component analysis model. Step 4: For the subject, measure the subject's basic parameters and abdominal impedance value; Before the formal measurement, the system parameters need to be determined. , Perform corrections, among which and These are the slope and offset, respectively: First, an impedance network for calibration is added. This impedance network consists of two or more impedance matching branches connected in parallel. Each impedance matching branch consists of a precision resistor and a switch connected in series. The two ends of the impedance network are connected in parallel to the output of the constant current source generating circuit. The impedance network is connected to the output of the excitation signal. The analog switch is used to switch between measuring the calibration network or the impedance of the human abdomen. Then, the voltage across the human abdomen is measured using an impedance network and corrected using the least squares method, i.e.: Using data structure algorithms, a linear queue linked list is created to determine whether the data is in a steady state when the error successively approaches zero. The condition for determining steady state is as follows: In the formula, To minimize absolute error, The number of data items in the queue. The first in the queue One data point, The minimum steady-state error is defined. The voltage across the resistor is measured, and the values ​​obtained by the ADC sampling are ADC1, ADC2, ... . The slope relationship between the voltage and the resistance, and its offset, can be obtained using maximum likelihood estimation. In the formula, The number of data points in the sampling queue. For the first The voltage of the resistor in the next sample. For the first The impedance value of the abdomen in the second sampling; During the measurement process, a queue is established to detect error signals in real time and calculate the errors in the queue to ensure that measurement differences can be detected in real time. If the detected measurement error is higher than the set threshold, the system parameters will be recalculated and corrected to adapt to the changes caused by the complexity of the human body. The new system parameters obtained after correction will be pushed onto the stack to recalculate a more accurate human body impedance value. Step 5: Input the basic parameters of the subject and the abdominal impedance value obtained in Step 4 into the abdominal component analysis model trained in Step 3 to obtain the abdominal component content of the subject. The above basic parameters include height, weight, and body mass index. Abdominal composition includes abdominal fat content, abdominal water content, abdominal protein content, abdominal inorganic salt content, and abdominal muscle content. Abdominal impedance values ​​include abdominal impedance values ​​measured at three different frequencies.

2. The method for abdominal component analysis according to claim 1, characterized in that, In steps 2 and 4, the method for measuring the abdominal impedance value is as follows: Step a: Apply three sinusoidal excitation signals of different frequencies to the abdomen of the human body. The abdomen generates a corresponding electric field under the excitation of the sinusoidal excitation signals. Collect the electric field to obtain the measurement signal of the abdomen. Step b: Amplify, condition, and detect the amplitude and phase of the measurement signal, and after analog-to-digital conversion, sample the sampling impedance value of the abdomen; Step c: Perform digital low-pass filtering on the sampled impedance value of the abdomen, and use the minimum mean square error to determine its steady state, thereby obtaining the impedance value of the abdomen.

3. The method for abdominal component analysis according to claim 2, characterized in that, In step a, the three frequency sinusoidal excitation signals are sinusoidal excitation signals in the frequency range of 1KHz to 1MHz.

4. An abdominal component analyzer that implements the abdominal component analysis method of claim 1, characterized in that, It consists of a constant current source generating circuit, a signal acquisition circuit, a signal amplification circuit, a signal conditioning circuit, an amplitude and phase detection circuit, an AD data acquisition circuit, a microprocessor, and a data processing platform. The control terminal of the microprocessor is connected to the constant current source generating circuit. The output terminal of the constant current source generating circuit is in contact with the human abdomen, and the input terminal of the signal acquisition circuit is in contact with the human abdomen, with a certain distance between the output terminal of the constant current source generating circuit and the input terminal of the signal acquisition circuit. The output terminal of the signal acquisition circuit is connected to the input terminal of the microprocessor after passing through the signal amplification circuit, the signal conditioning circuit, the amplitude and phase detection circuit, and the AD data acquisition circuit in sequence. The output terminal of the microprocessor is connected to the data processing platform.

5. An abdominal component analyzer according to claim 4, characterized in that, The constant current source generating circuit outputs three sinusoidal excitation signals in the frequency range of 1KHz to 1MHz.

6. An abdominal component analyzer according to claim 4, characterized in that, The constant current source generating circuit consists of a DSS signal generator, an amplification and filtering circuit, and a voltage-controlled constant current source circuit; the DSS signal generator is further composed of a phase accumulator, a waveform memory, a digital-to-analog converter, and a low-pass filter; The microprocessor's control terminal is connected to the input terminal of the phase accumulator, as well as the clock control terminals of the phase accumulator and the digital-to-analog converter; the output terminal of the phase accumulator is connected to the input terminal of the waveform memory, the output terminal of the waveform memory is connected to the input terminal of the digital-to-analog converter, and the output terminal of the digital-to-analog converter is connected to the input terminal of the low-pass filter. The output of the low-pass filter is connected to the input of the amplification and filtering circuit. The output of the amplification and filtering circuit is connected to the input of the voltage-controlled constant current source circuit. The output of the voltage-controlled constant current source circuit forms the output of the constant current source generating circuit.

7. An abdominal component analyzer according to claim 4, characterized in that, The signal conditioning circuit consists of a notch filter and a bandpass filter; the input terminal of the notch filter forms the input terminal of the signal conditioning circuit, the output terminal of the notch filter is connected to the input terminal of the bandpass filter, and the output terminal of the bandpass filter forms the output terminal of the signal conditioning circuit.

Images