A method for measuring hemoglobin concentration in vitro based on biosensor

Abstract

The present application relates to the technical field of in vitro diagnostic detection, and discloses a method for measuring hemoglobin concentration in vitro based on a biosensor. The method comprises real-time calibration of a sensor sensitive unit using a standard signal when a sample is introduced, and generating biosensor calibration parameters adapted to the current sample. According to the biosensor calibration, the original optical or electrochemical signal stream is typed, and the characteristic section and the background section are defined. A nonlinear cooperative measurement path is planned, which alternately samples the characteristic section and the background section within the detection time domain, and the sensor is driven to perform divide-and-conquer measurement according to the nonlinear cooperative measurement path, and a time-sequential measurement data set is obtained. Data fusion and calculation are performed on the measurement data set to obtain the hemoglobin concentration value. The method improves the accuracy and anti-interference ability of single detection through real-time individual calibration and adaptive optimization of the sampling path.

Description

Technical Field

[0001] This invention relates to the field of in vitro diagnostic testing technology, specifically a method for calculating hemoglobin concentration in vitro based on a biosensor. Background Technology

[0002] Current in vitro hemoglobin concentration detection methods mostly employ pre-calibrated fixed calibration curves. This method relies on standards to establish a universal model, which is then applied to all subsequent samples. However, biosensor units exhibit performance drift and individual variability, and different blood sample matrices are complex. Fixed calibration models cannot accurately reflect the true instantaneous response of individual samples, leading to systematic biases in the detection results.

[0003] Conventional signal acquisition typically employs a uniform and continuous time-series strategy to obtain a mixed signal curve of the entire reaction process. The effective response signal of hemoglobin is mixed with interference signals such as instrument background and sample background. This method indiscriminately acquires and processes information from all time periods, failing to distinguish the spatiotemporal characteristics of different components in the signal stream. This dilutes the effective information and limits the overall signal-to-noise ratio and resistance to matrix interference of the system.

[0004] Real-time, personalized calibration of individual samples and sensor operating status, as well as intelligent and efficient extraction of target response information from complex signal streams, are key to improving detection accuracy and reliability. Summary of the Invention

[0005] The purpose of this invention is to provide a method for in vitro measurement of hemoglobin concentration based on a biosensor, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a method for in vitro measurement of hemoglobin concentration based on a biosensor, the method comprising:

[0007] Perform dynamic calibration measurements of the biosensor to generate biosensor calibration parameters adapted to the current blood sample to be tested; the dynamic calibration measurements of the biosensor include real-time response calibration of a sensitive unit of the biosensor using a standard signal during the sample introduction phase;

[0008] Based on the biosensor calibration parameters, the raw optical or electrochemical signal streams acquired by the biosensor are subjected to multi-level signal classification to define the characteristic segments and background segments of the raw optical or electrochemical signal streams; the characteristic segments are identified as signal portions containing effective hemoglobin response information.

[0009] Based on the defined feature segment and background segment, a nonlinear cooperative measurement path is planned within the detection time domain of the biosensor; the cooperative measurement path specifies the alternating sampling order and dwell time of the feature segment and background segment at multiple discrete measurement moments;

[0010] Based on the aforementioned collaborative measurement path, the biosensor is driven to perform divide-and-conquer measurements on the blood sample to be tested in vitro to obtain a divide-and-conquer measurement data set; the divide-and-conquer measurement data set contains multiple signal subsets corresponding to the characteristic segment and the background segment at different measurement times;

[0011] Hierarchical fusion and solution are performed on the divide-and-conquer measurement data set to generate hemoglobin concentration values.

[0012] Preferably, the step of performing dynamic calibration measurements of the biosensor includes:

[0013] While the biosensor is in contact with the blood sample to be tested, a standard hemoglobin solution of known concentration is loaded into the sensitive unit of the biosensor to obtain the dynamic baseline curve of the biosensor in a mixed response state.

[0014] Feature points are extracted from the dynamic benchmark curve, including the rising inflection point, steady-state plateau, and falling inflection point of the dynamic benchmark curve.

[0015] The biosensor calibration parameters are calculated based on the extracted feature points. These parameters include signal gain per unit concentration, response time constant, and baseline drift compensation.

[0016] Preferably, the step of performing multi-level signal typing on the raw optical or electrochemical signal stream acquired by the biosensor based on the biosensor calibration parameters to define the characteristic segments and background segments of the raw optical or electrochemical signal stream includes:

[0017] The original optical or electrochemical signal stream is normalized and amplified using the unit concentration signal gain to obtain a preprocessed signal stream;

[0018] In the preprocessed signal stream, continuous time segments whose signal change rate exceeds a preset threshold are identified based on the response time constant, and these continuous time segments are initially marked as feature segments.

[0019] By combining the baseline drift compensation amount, baseline fitting and subtraction are performed on the signal portion outside the feature segment, and the subtracted stable signal portion is defined as the background segment.

[0020] Preferably, the step of planning a nonlinear cooperative measurement path within the detection time domain of the biosensor based on the defined feature segment and background segment includes:

[0021] Using the midpoint of the time in the background segment as the starting anchor point and the extreme point of the signal amplitude in the feature segment as the target anchor point, an initial measurement path connecting the starting anchor point and the target anchor point is constructed.

[0022] On the initial measurement path, an intermediate measurement node is inserted based on the signal gradient change in the characteristic section; the insertion density of the intermediate measurement node is proportional to the magnitude of the signal gradient.

[0023] All anchor points are connected to intermediate measurement nodes in a time sequence, and their respective signal acquisition dwell times are assigned to form a complete nonlinear collaborative measurement path.

[0024] Preferably, the step of driving the biosensor to perform separate measurements on the blood sample to be tested in vitro, based on the collaborative measurement path, to obtain a separate measurement data set, includes:

[0025] The signal acquisition unit of the control biosensor moves sequentially to each anchor point and intermediate measurement node according to the order specified in the collaborative measurement path;

[0026] At each anchor point and intermediate measurement node, the control signal acquisition unit acquires stable signals according to the dwell time specified by the cooperative measurement path, and binds the acquired stable signals with their corresponding node identifiers to generate a signal data packet;

[0027] After traversing all nodes on the collaborative measurement path, all generated signal data packets are summarized by timestamp to form a divide-and-conquer measurement data set.

[0028] Preferably, the hierarchical fusion and resolution includes performing a weighted difference operation on a subset of signals from the feature region and a subset of signals from the background region, specifically:

[0029] From the divide-and-conquer measurement data set, extract all signal subsets belonging to the characteristic segments respectively, calculate their weighted average value as the characteristic signal intensity;

[0030] From the divide-and-conquer measurement data set, extract all signal subsets belonging to the background segment, calculate their weighted average value, and use it as the background signal intensity;

[0031] The difference between the intensity of the feature signal and the intensity of the background signal is input into a preset concentration conversion model, which outputs the hemoglobin concentration value.

[0032] Preferably, the construction of the concentration conversion model includes:

[0033] Multiple standard blood samples with known hemoglobin concentrations were collected, and a biosensor was used to acquire signals from each standard blood sample to obtain the difference data between the corresponding characteristic signal intensity and the background signal intensity.

[0034] The difference data is correlated with known hemoglobin concentration values ​​to construct a training dataset;

[0035] The training dataset was fitted using a linear regression algorithm to obtain the slope and intercept parameters of the concentration conversion model.

[0036] To verify the accuracy of the concentration conversion model, the error between the model's predicted concentration and the actual concentration was calculated using an independent validation dataset. When the error exceeded a preset threshold, the slope and intercept parameters were readjusted.

[0037] The independent validation dataset consists of standard blood samples that were not used in the training process.

[0038] The readjusted slope and intercept parameters are stored in the concentration conversion model for real-time hemoglobin concentration calculation.

[0039] Preferably, the step of normalizing and amplifying the original optical or electrochemical signal stream using the unit concentration signal gain to obtain a preprocessed signal stream includes:

[0040] Read the raw optical or electrochemical signal streams acquired by the biosensor and obtain the baseline voltage value of the raw optical or electrochemical signal streams;

[0041] The normalized amplification factor is calculated based on the unit concentration signal gain, where the normalized amplification factor is the reciprocal of the unit concentration signal gain.

[0042] The baseline correction signal is obtained by subtracting the baseline voltage value from each data point of the original optical or electrochemical signal stream.

[0043] Multiply each data point of the baseline correction signal by the normalization amplification factor to obtain the amplitude normalized signal;

[0044] The amplitude-normalized signal is filtered by moving average to eliminate random fluctuations and output a smooth preprocessed signal stream.

[0045] Preferably, the initial measurement path connecting the starting anchor point and the target anchor point, using the midpoint of the time in the background segment as the starting anchor point and the extreme point of the signal amplitude in the feature segment as the target anchor point, includes:

[0046] Identify the start and end times of the background segment, calculate the midpoint of the time for the background segment, and set the midpoint as the starting anchor point.

[0047] Scan the signal amplitude within the feature segment, find the maximum or minimum point of the signal amplitude as the extreme point, and set the time point corresponding to the extreme point as the target anchor point.

[0048] A straight path is generated between the starting anchor point and the target anchor point in chronological order as the initial measurement path.

[0049] Preferably, at each anchor point and intermediate measurement node, the control signal acquisition unit acquires stable signals according to the dwell time specified by the cooperative measurement path, and binds the acquired stable signals with their corresponding node identifiers to generate a signal data packet, including:

[0050] Read the location information of the current node on the collaborative measurement path, including node type, time coordinates, and dwell time;

[0051] The control signal acquisition unit moves to the time coordinate corresponding to the current node and starts signal acquisition;

[0052] During the dwell time, multiple sets of signal values ​​are continuously collected, and the average value of these signal values ​​is calculated as a stable signal.

[0053] Generate a node identifier, which includes a node sequence number, a node type, and a timestamp;

[0054] The stable signal is combined with the node identifier to form a signal data packet, which includes identifier information in the header and signal value in the data section.

[0055] Compared with the prior art, the beneficial effects of the present invention are:

[0056] Real-time response calibration is performed concurrently during the sample introduction phase, using standard signals to instantly measure the sensitive unit. This operation moves the calibration process from before sample measurement to during the measurement process, ensuring that the generated biosensor calibration parameters reflect the true operating state of the biosensor when in contact with the specific sample matrix. This technology eliminates systematic response biases caused by individual performance differences of biosensor units, long-term use drift, and the effects of different sample matrices. The biosensor calibration parameters are linked to the test sample in real time, ensuring that each measurement is based on an individualized benchmark, improving the accuracy and reliability of single-test results and reducing intra-batch and inter-batch testing errors.

[0057] Based on signal classification results, a nonlinear collaborative measurement path is planned, specifying the exact order and dwell time for alternating sampling of characteristic and background segments at multiple discrete time points. This collaborative measurement path breaks away from the traditional uniform continuous sampling pattern, guiding the measurement system to perform focused, long-term acquisition in signal-rich characteristic segments, while conducting rapid or verification acquisition in background interference segments. This time-series, programmable measurement strategy achieves optimized allocation of measurement resources in the time dimension. It can actively focus on effective signals while suppressing the acquisition weight of background noise, thereby enhancing the effective components of the signal at the raw data level, improving the signal quality and signal-to-noise ratio used for final calculation, and making the concentration calculation results stable and reliable. Attached Figure Description

[0058] Figure 1 This is a flowchart illustrating the process of the method for in vitro measurement of hemoglobin concentration based on a biosensor as described in this invention.

[0059] Figure 2 A flowchart for dynamic calibration measurements of biosensors;

[0060] Figure 3 A flowchart for multi-level signal classification;

[0061] Figure 4 Comparison of preprocessing effects for biosensor signal streams;

[0062] Figure 5 The graph shows the linear fitting relationship of the hemoglobin concentration conversion model. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Please see Figure 1This invention provides a method for in vitro measurement of hemoglobin concentration based on a biosensor. The method includes: dynamic calibration measurement of the biosensor; simultaneously introducing a blood sample to be tested into the biosensor and introducing a standard hemoglobin solution of known concentration into the sensitive unit of the biosensor; observing the real-time response of the biosensor under the mixing of the standard hemoglobin solution and the blood sample to be tested; and generating a set of biosensor calibration parameters specifically adapted to the current measurement conditions. These biosensor calibrations dynamically reflect the response characteristics of the biosensor under this specific measurement environment. Based on the dynamically generated biosensor calibration parameters, the raw optical or electrochemical signal stream acquired by the biosensor in subsequent measurement cycles is processed and analyzed. The raw optical or electrochemical signal stream is a continuous data sequence that varies over time. Through a multi-level signal classification algorithm, two segments with different properties are accurately defined from this continuous data sequence: a feature segment containing effective hemoglobin concentration response information and a background segment mainly reflecting environmental background and baseline noise.

[0065] After clarifying the temporal distribution and signal characteristics of the feature and background segments, a nonlinear collaborative measurement path is further planned within the entire detection time window of the biosensor. This collaborative measurement path does not collect signals uniformly or randomly, but rather, based on the segmentation results of the previous step, it explicitly stipulates that at multiple discrete, non-uniformly distributed measurement moments, the signal acquisition unit should alternately access and acquire signals from the feature and background segments, and allocate a dwell time for each measurement moment or node, thereby combining enhanced acquisition of key signal regions with representative acquisition of non-key regions. After the collaborative measurement path is planned, the control unit of the biosensor drives the signal acquisition hardware according to the collaborative measurement path to perform divide-and-conquer measurement operations on the blood sample to be tested in vitro. That is, according to the order, position, and duration specified by the collaborative measurement path, signal segments are acquired at designated time points, ultimately obtaining a structured divide-and-conquer measurement data set. This divide-and-conquer measurement data set contains multiple signal subsets corresponding to the feature and background segments at different measurement moments. By performing hierarchical fusion and calculation algorithms on the divide-and-conquer measurement dataset, signal subsets from different segments and times are integrated and calculated to eliminate background interference, extract effective responses, and finally output accurate hemoglobin concentration values.

[0066] Example 1: See Figure 2In practical implementation, when the blood sample to be tested is physically introduced into the detection area of ​​the biosensor, such as a microchannel or reaction chamber, a standard hemoglobin solution of known precise concentration is simultaneously and micro-loaded onto a sensitive unit of the biosensor. This loading can be achieved through an integrated micropump and valve system, allowing the standard hemoglobin solution and the blood sample to be tested to contact and mix at the interface of the sensitive unit. In this implementation, the signal generated by the biosensor in the mixed response state is continuously recorded. The signal can be in the form of absorbance or fluorescence intensity changes detected by an optical sensor, or current or impedance changes detected by an electrochemical sensor. The entire process is recorded from the moment the solution is loaded until the signal reaches relative stability, thus forming a dynamic baseline curve. In this implementation, the dynamic baseline curve reflects the real-time response characteristics of the biosensor to hemoglobin under the combined conditions of current temperature, pH value, and sample matrix composition. The curve includes complete dynamic information such as the initial response, rapid rise, and approach to steady state.

[0067] In some embodiments, feature point extraction of the dynamic baseline curve is fundamental to subsequent calculations. Feature point extraction is performed using digital signal processing algorithms. First, the recorded dynamic baseline curve is smoothed and filtered to suppress noise. Then, the first derivative sequence of the dynamic baseline curve is calculated. The rising inflection point of the dynamic baseline curve is determined by identifying the first local maximum point in the first derivative sequence. This maximum point corresponds to the position where the slope of the dynamic baseline curve is the largest, marking the transition of the response from the initial delay stage to the rapid rising stage. The identification of the steady-state plateau is achieved by analyzing the signal fluctuations of the dynamic baseline curve within the later time window. When the amplitude of signal change over a continuous period is less than a preset absolute or relative threshold, the dynamic baseline curve is determined to have entered the steady-state plateau region. The feature points of the steady-state plateau can be represented as the average value of the signal in this stable region and its corresponding time interval. The identification of the falling inflection point is performed when the dynamic baseline curve exhibits a downward trend. It is determined by locating the local minimum point of the first derivative sequence in the falling phase. The falling inflection point may correspond to the transition point where the reaction reaches equilibrium or where mass diffusion is limited. In some embodiments, feature point extraction further includes identifying the start and end points of the dynamic baseline curve. The start point is the moment when the signal first significantly deviates from the initial baseline, and the end point is the moment when the preset signal acquisition period ends. The initial baseline is a stable signal reference value with no specific response acquired by the biosensor before it comes into contact with the blood sample to be tested and before the standard hemoglobin solution is loaded.

[0068] In practical implementation, based on the extracted feature points, a set of biosensor calibration parameters can be calculated for subsequent calibration and analysis of this measurement. The signal gain per unit concentration is obtained through a calculation formula. The numerator of the formula is the difference between the average signal value in the steady-state plateau region of the dynamic reference curve and the average signal value of the initial baseline. The denominator is the known concentration of the applied standard hemoglobin solution, and the result is the signal gain per unit concentration. The response time constant is obtained by fitting the rising segment of the dynamic reference curve. The rising segment is defined as the portion of the curve from the rising inflection point to the starting point of the steady-state plateau. An exponential function model is used to perform nonlinear least-squares fitting on the rising segment data, and the resulting exponential decay time constant is the response time constant. The response time constant reflects the response kinetics of the biosensor under current conditions. The baseline drift compensation is obtained by trend analysis of the pure baseline signal before the start of the dynamic reference curve or the signal in the non-characteristic response region throughout the entire measurement cycle. A linear fitting algorithm is used to calculate the drift slope and intercept of the baseline signal. The baseline drift compensation can be expressed as the signal drift per unit time.

[0069] Optionally, the loading concentration of the standard hemoglobin solution needs careful selection during the acquisition of the dynamic baseline curve. The loading concentration should fall within the linear detection range of the biosensor and should not be too high to completely mask the signal of the sample to be tested. Optionally, the standard hemoglobin solution can be loaded via pulsed injection or continuous low-flow-rate injection accompanying sample introduction, with the aim of clearly generating a characteristic dynamic baseline curve for analysis based on the mixed response. In specific implementations, the sensitive unit can be an independent reference sensing unit or the same sensing unit shared with the main detection unit. In the case of a shared sensing unit, the loading of the standard hemoglobin solution needs to be closely coordinated with the loading of the blood sample to be tested in time to ensure that the mixed response can be accurately captured. In specific implementations, the signal acquisition system needs to have a sufficiently high sampling rate and resolution to accurately record the detailed changes of the dynamic baseline curve. It is understood that the feature point extraction algorithm needs to have a certain degree of noise resistance, and the window size and threshold parameters of the smoothing filter need to be preset or adaptively adjusted according to the noise characteristics and response speed of the biosensor. In practice, the process of calculating the unit concentration signal gain, response time constant, and baseline drift compensation can be automatically completed by calibration software running on an embedded processor or external computer. The calculation results are stored in a designated memory area for direct use in subsequent signal classification steps.

[0070] Example 2: See Figure 3In specific implementation, the system first reads the raw optical or electrochemical signal stream acquired by the biosensor. This raw optical or electrochemical signal stream is a data sequence recorded sequentially at a fixed sampling frequency. Each data point contains a timestamp and a signal amplitude. The signal amplitude may be a voltage value (in volts) in the optical sensor and a current value (in amperes) in the electrochemical sensor. In specific implementation, the baseline voltage value of the raw optical or electrochemical signal stream is obtained. This baseline voltage value is obtained by analyzing a pre-defined data segment at the beginning of the raw optical or electrochemical signal stream. This data segment corresponds to the stage before sample loading or the initial loading phase before the sensor generates a specific response. The arithmetic mean of the signal amplitudes of all data points in this data segment is calculated, and this arithmetic mean is determined as the baseline voltage value. In some embodiments, the calculation of the normalized amplification factor requires the calculation of the unit concentration signal gain obtained from dynamic calibration measurements. The unit concentration signal gain is a physically meaningful parameter representing the change in sensor signal caused by a unit concentration of hemoglobin solution. Its calculation formula is:

[0071]

[0072] in: Represents signal gain per unit concentration, in mV / (g / dL) or μA / (g / L). This represents the calculated normalized amplification factor, in units of (g / dL) / mV or (g / L) / μA. In practice, baseline correction is performed on the original optical or electrochemical signal stream. This is done by iterating through each data point of the original optical or electrochemical signal stream and subtracting the previously calculated baseline voltage value from the signal amplitude of that data point. This operation generates a new data sequence, the baseline-corrected signal, which eliminates the inherent offset of the biosensor and some low-frequency interference. Amplitude normalization is then performed by multiplying the value of each data point of the baseline-corrected signal by the normalized amplification factor. This ensures that the amplitude scale of the output signal matches the theoretical sensitivity to changes in hemoglobin concentration, generating an amplitude-normalized signal. Moving average filtering is applied to the amplitude-normalized signal to eliminate random fluctuations; the window width of the moving average filter... Based on signal sampling frequency Settings, for example This ensures that the window covers a reasonable short time interval, and the filtered data point values ​​are replaced by the average of all data points within the window, ultimately outputting a smooth preprocessed signal stream.

[0073] In practical implementation, characteristic segments are identified based on the response time constant. Obtained from dynamic calibration measurements, this is used to characterize the rate of response of the biosensor. A preset threshold for calculating the rate of signal change is used. Preset threshold Can be compared with response time constant To establish an inverse relationship, one feasible definition is: ,in: This is a scaling factor set according to the biosensor type and noise level, with units consistent with the signal amplitude. The signal change rate is calculated in the preprocessed signal stream, obtained by performing discrete difference operations on the preprocessed signal stream, i.e., calculating the ratio of the difference in amplitude between adjacent data points to the time interval. The entire preprocessed signal stream is scanned to identify instances where the absolute value of the signal change rate continuously exceeds a preset threshold. Time segments indicating drastic changes in the biosensor response are recorded, with the start and end indices of each such consecutive time segment initially labeled as a characteristic region. In some embodiments, to prevent false positives caused by noise spikes, a minimum duration threshold can be set, allowing the signal to change only when the rate of change exceeds a preset threshold. Only when the duration of a segment exceeds the minimum duration threshold is the corresponding time segment marked as a feature segment.

[0074] Optionally, the background segment can be defined by combining the baseline drift compensation amount, whereby the baseline drift compensation amount is... Obtained from dynamic calibration measurements, this represents the drift rate of the initial baseline over time. In practice, baseline fitting is performed on the signal portion outside the characteristic segment. This portion includes the initial stage of the original signal, the response gap, and the stage after the response ends. A polynomial fitting method is used to fit the signal data in these areas, typically choosing a first or second order fitting order, and the baseline drift compensation amount is then applied. As a constraint on the fitting process, for example, the slope of the fitted curve at the starting point is close to... In practice, the fitted baseline curve is subtracted from the original signal data outside the characteristic segment to obtain the baseline-subtracted signal. Within this baseline-subtracted signal, continuous portions with stable signal values ​​and small fluctuations are identified as background segments. The criterion for determining background segments is that the standard deviation of the signal amplitude is below a global threshold. In practice, the initially labeled characteristic segments also need to be corrected using the baseline fitted from adjacent background segments. This involves subtracting the baseline fitted value at the corresponding time point from the signal value of the characteristic segment. After baseline correction, the characteristic segment is finally confirmed as a characteristic segment. Optionally, the time boundary between the characteristic segment and the background segment can be precisely defined by finding the zero-crossing point of the first derivative of the signal or points where the signal energy changes significantly.

[0075] It is understandable that the multi-level signal classification process relies on the unit concentration signal gain, response time constant, and baseline drift compensation provided by dynamic calibration measurements. These biosensor calibration parameters enable the classification process to adapt to changes in each specific measurement environment. Normalization amplification processing ensures the comparability of signal amplitudes from different batches of biosensors or under different measurement conditions, providing a unified scale for subsequent threshold judgment and segmentation. In practice, the entire multi-level signal classification algorithm can be executed in real time on an embedded microprocessor. Steps such as generating the preprocessed signal stream, labeling feature segments, and fitting and subtracting background segments are completed sequentially through a pre-defined program. The classification results are stored in the form of a data structure, explicitly recording the start and end time indices of feature segments and background segments in the original optical or electrochemical signal stream, along with their corresponding corrected signal values. In practice, the window width of the moving average filter needs careful selection. An excessively wide window may smooth out signal details, while an excessively narrow window may fail to effectively suppress noise. The window width selection needs to match the response time of the biosensor and the signal sampling frequency. The specific determination of the window width needs to consider the response time constant of the biosensor and the signal sampling frequency set by the system. Typically, the window width can be set to cover the number of sampling points corresponding to a typical response time constant of the biosensor, ensuring that the filtering operation smooths out high-frequency noise without excessively masking the true dynamic characteristics of the signal. For example, if the response time constant is long or the sampling frequency is high, the window width can be increased accordingly to improve the smoothing effect; conversely, the window width needs to be reduced to preserve signal details. The order of the polynomial fitting baseline also needs to be carefully considered. An excessively high order may fit weak, real responses into the baseline and result in incorrect subtraction, while an excessively low order may fail to accurately describe complex baseline drift trends. It is usually selected based on the length and complexity of the background signal segment. The order of the polynomial fitting baseline is selected based on the length and fluctuation characteristics of the background signal segment. For background segments with short signal lengths or relatively smooth fluctuations, first-order linear fitting can effectively describe the baseline drift trend. However, if the background segment is long and exhibits significant nonlinear drift characteristics, second-order or higher-order polynomials are required for fitting to more accurately capture the complex changes in the baseline and avoid underfitting or overfitting. In practical implementation, the final definition of the characteristic segment and the background segment is a direct input for subsequent planning of the collaborative measurement path.

[0076] See Figure 4This figure is a core visualization result of the multi-level signal classification process. The horizontal axis represents measurement time, and the vertical axis represents signal amplitude. The figure contains three key curves: the grayscale curve represents the original optical or electrochemical signal stream, exhibiting significant random noise and baseline fluctuations; the black curve represents the preprocessed signal stream, the result after baseline correction, amplification, and moving average filtering, with significantly reduced fluctuations and a smoother trend; the dashed line represents the baseline value, corresponding to the stable signal mean at the beginning of the original optical or electrochemical signal stream. This figure intuitively demonstrates the role of signal preprocessing, improving the signal-to-noise ratio and providing clear and reliable preprocessed data for subsequent definition of characteristic and background segments.

[0077] Example 3: In specific implementation, it is necessary to first identify the start and end times of the background segment. The start and end times are time indices directly read from the result data structure obtained from the multi-level signal classification steps. The start and end times are added together and divided by two to calculate the midpoint of the background segment. This midpoint is set as the starting anchor point of the collaborative measurement path, representing a typical background signal reference position. In specific implementation, the signal amplitude within the feature segment is scanned. The signal amplitude within the feature segment refers to the signal value after baseline correction. The maximum or minimum point of the signal amplitude is found through a traversal comparison algorithm. This depends on the response characteristics of the biosensor. For example, for absorbance-type sensors, an increase in hemoglobin concentration leads to an increase in absorbance, so the maximum point is sought; for fluorescence quenching-type sensors, the minimum point may be sought. The time point corresponding to the found extreme point of signal amplitude is set as the target anchor point of the collaborative measurement path. In some embodiments, a straight path is generated between the starting anchor point and the target anchor point in chronological order as the initial measurement path. This straight path is a simple linear connection in the time dimension from the starting anchor point to the target anchor point, without involving path planning in the signal amplitude space. The time span of the initial measurement path is calculated, which is the difference between the target anchor point time and the starting anchor point time. A basic sampling interval on the initial measurement path is determined based on the maximum signal sampling frequency supported by the biosensor hardware. The basic sampling interval is no greater than an integer multiple of the reciprocal of the sampling frequency to ensure that the sampling points on the path are temporally reachable. The coordinate information of the starting anchor point and the target anchor point is marked on the initial measurement path. The coordinate information includes time coordinates and signal amplitude coordinates. The time coordinates are obtained directly from the time point, and the signal amplitude coordinates are obtained from the values ​​of the corresponding time point in the preprocessed signal stream.

[0078] In practice, intermediate measurement nodes are inserted into the initial measurement path based on the signal gradient changes in the characteristic segments. The signal gradient changes in the characteristic segments are obtained by calculating the derivative of the signal amplitude with respect to time within those segments; the magnitude of the derivative reflects the drasticness of the signal change. On the initial measurement path, regions with large signal gradients indicate rapid signal changes, requiring higher sampling densities to capture signal morphology details. Therefore, the insertion density of intermediate measurement nodes is directly proportional to the magnitude of the signal gradient. The relationship between insertion density and signal gradient magnitude can be expressed as:

[0079]

[0080] in: This represents the signal gradient at time t (unit: mV / s). This represents the insertion density of intermediate measurement nodes per unit time around time t (unit: nodes / s). This is a scaling factor (unit: inserts / mV) for adjusting the insertion density. In practice, along the time axis of the initial measurement path, the absolute value of the signal gradient at each potential location (e.g., with a basic sampling interval as the step size) is calculated. When the absolute value of the gradient exceeds a preset threshold... Within a continuous region, according to the calculated density Intermediate measurement nodes are inserted, and their specific time coordinates can be evenly or randomly distributed within a high-density region. In some embodiments, the insertion of intermediate measurement nodes also needs to avoid being too dense to exceed the hardware sampling capacity; therefore, a minimum time interval is set. As a constraint, ensure that the time interval between two adjacent measurement nodes is not less than .

[0081] Optionally, all anchor points are sequentially connected to intermediate measurement nodes and a dwell time is assigned. Anchor points include the starting anchor point, the target anchor point, and all inserted intermediate measurement nodes. All nodes are sorted in chronological order from earliest to latest, forming a complete node sequence. A signal acquisition dwell time is assigned to each node in the node sequence. The dwell time allocation is based on the node's location and type. For the starting anchor point, which serves as a background reference, a relatively short dwell time can be assigned since the background signal is usually relatively stable. For the target anchor point, since it is located near the signal extreme point, the signal may plateau or inflection point. In order to obtain stable measurement values, a relatively long dwell time needs to be allocated. For intermediate measurement nodes, the dwell time is... The allocation can be dynamically adjusted based on the signal gradient at its location. In areas with large gradients, the signal changes rapidly, and the residence time of the allocation is extended. The dwell time can be relatively short to quickly capture transient states, and the allocated dwell time can be adjusted in regions where the gradient is gentle but the signal value is important. The dwell time can be appropriately extended to obtain more stable readings. Optionally, the specific value of the dwell time can be determined in advance through experiments based on the signal-noise level and the required measurement accuracy, and stored in a lookup table. In practice, the complete cooperative measurement path is ultimately described as an ordered list, where each item records information about a measurement node, typically including the node number, node type, node time coordinate, expected signal amplitude coordinate at the node, and the dwell time allocated to that node.

[0082] It is understandable that the planning of the collaborative measurement path is based on an intelligent sampling strategy using prior signal classification results. The nonlinearity of the path is reflected in the non-uniform distribution of measurement nodes on the time axis and the differentiated allocation of dwell time. It is also understandable that using the midpoint of the background segment as the starting anchor point allows for the acquisition of a more representative background signal, avoiding interference from transitional signals that may exist at the edges of the background segment. In practical implementation, identifying the extreme points of signal amplitude within the characteristic segment requires ensuring that the point is indeed a global or local extreme, which can be confirmed by comparing the signal amplitude of that point with several points before and after it. In practical implementation, although the initial measurement path is described as a straight line, this path is only used to plan the time coordinate sequence of the measurement nodes and does not mean that the signal acquisition probe or beam needs to move in a straight line in the signal amplitude space. The algorithm for inserting intermediate measurement nodes needs to consider computational complexity and real-time performance. A simplified implementation is to insert the intermediate measurement nodes when the signal gradient exceeds a threshold. In the designated area, nodes are inserted at a higher frequency, fixed as an integer multiple of the basic sampling interval. In practical implementation, after the collaborative measurement path is planned, its data structure is transmitted to the control unit of the biosensor. The control unit parses the data and drives the signal acquisition hardware to strictly perform measurements according to the time sequence and dwell time specified by the collaborative measurement path. This collaborative measurement path ensures that measurement resources are preferentially concentrated in segments with rich signal variations and critical areas, while also performing necessary sampling in background segments, thus achieving a synergy between measurement efficiency and information integrity.

[0083] Example 4: A biosensor is driven by a collaborative measurement path to perform divide-and-conquer measurements on a blood sample in vitro. In this implementation, the biosensor's control unit reads the path description data structure generated during the collaborative measurement path planning step. This data structure contains an ordered list of nodes, each storing node type, time coordinates, signal amplitude coordinates, and dwell time information. In this implementation, the biosensor's signal acquisition unit moves sequentially to each anchor point and intermediate measurement node according to the node order specified in the collaborative measurement path. For optical biosensors, the signal acquisition unit may include a tunable light source and a photodetector. The movement operation involves adjusting the illumination time window of the light source and the integration time window of the detector to match the node's time coordinates. For electrochemical biosensors, the signal acquisition unit may include a potentiostat and a working electrode. The movement operation involves controlling the potentiostat to apply the corresponding potential sequence according to the node's time coordinates and recording the response current. In some embodiments, the node order of the collaborative measurement path is strictly arranged in ascending order of time coordinates. The operation timing of the signal acquisition unit is synchronized by a high-precision clock to ensure that the signal acquisition program starts at the specified time point for each node. Reading the location information of the current node on the collaborative measurement path is the first step in performing data acquisition. The location information includes whether the node type is the starting anchor point, the target anchor point, or an intermediate measurement node. The node's time coordinate indicates the absolute or relative time at which signal acquisition begins. The dwell time indicates the duration for which signal acquisition should continue. Controlling the signal acquisition unit to move to the time coordinate corresponding to the current node involves configuring a timer or waiting for the system clock to reach a specified time. Subsequently, the signal acquisition hardware is started to begin recording the signal.

[0084] In practice, signal acquisition and stabilization signal generation are performed at each anchor point and intermediate measurement node. After signal acquisition is initiated, the signal acquisition unit continuously acquires multiple sets of raw signal values ​​at a predetermined sampling frequency within a specified dwell time. The number of sets of raw signal values ​​depends on the dwell time and the sampling frequency. The stabilization signal is calculated by averaging the multiple sets of raw signal values ​​acquired within the dwell time. The calculation formula is as follows:

[0085]

[0086] in: This indicates the number of samples collected during the stay period. One original signal value (unit: mV). This represents the total number of raw signal values ​​collected during the dwell time. This represents the calculated stable signal value (unit: mV). A node identifier is generated and bound to the current node. The node identifier is a data structure or string containing a node sequence number to identify the node's position in the path, a node type to distinguish between the starting anchor point, the target anchor point, and an intermediate measurement node, and a timestamp to record the precise start time of signal acquisition. The calculated stable signal value and the generated node identifier are combined into a signal data packet. The signal data packet typically includes a header and a data section. The header stores all the information of the node identifier, and the data section stores the stable signal value. And other possible metadata such as collection status codes.

[0087] Optionally, the movement and synchronization of the signal acquisition unit at the node can be implemented through an interrupt service routine. When the system clock reaches the node's time coordinate, a hardware interrupt is triggered. The interrupt service routine is responsible for starting signal acquisition and setting a timer to stop acquisition after the dwell time expires. Optionally, during the generation of data packets, in addition to the stable signal value and node identifier, the standard deviation or variance of the original signal value during the acquisition process can be added to evaluate the signal quality at that node. Refer to Table 1, a table of node information for a collaborative measurement path.

[0088] Table 1: Information Table of Collaborative Measurement Path Nodes

[0089]

[0090] In practice, the completion of the divide-and-conquer measurement is marked by the traversal of all nodes on the collaborative measurement path. The control unit checks whether all entries in the node list have been processed, i.e., whether each node has generated a corresponding signal data packet and stored it in the buffer. The process of constructing the divide-and-conquer measurement data set includes sorting and encapsulating all data packets in the buffer. The sorting is based on the node sequence number or timestamp field in the data packet. The encapsulated divide-and-conquer measurement data set is then passed as a complete data structure to the subsequent hierarchical fusion and solution module. It can be understood that the divide-and-conquer measurement process strictly follows the pre-planned collaborative measurement path, resulting in a non-uniform temporal distribution of measurement activities, focusing on covering key change points in characteristic sections. Collecting multiple sets of signal values ​​at nodes and calculating the average as a stable signal can effectively suppress random noise and improve the reliability of single measurement data. In practice, by introducing node identifiers, each signal data packet is given clear spatiotemporal attributes, ensuring that subsequent processing can correctly distinguish the signal source.

[0091] Example 5: Performing hierarchical fusion and decomposition on the divide-and-conquer measurement dataset to generate hemoglobin concentration values. In specific implementation, the first step of hierarchical fusion and decomposition is to perform a weighted difference operation on the signal subsets from the feature segment and the signal subsets from the background segment. Based on the node type field of the node identifier, extract all signal subsets from the divide-and-conquer measurement dataset whose node type is marked as "intermediate measurement node" or "target anchor point" and whose time coordinates are within the time range of the feature segment. These signal subsets belong to the feature segment. Calculate the weighted average of the feature segment as the feature signal strength. The calculation of the weighted average considers the dwell time corresponding to each signal subset. Signal subsets with longer dwell times usually have higher signal-to-noise ratios and are therefore given greater weight. The calculation formula is:

[0092]

[0093] in: This represents the total number of signal subsets belonging to the characteristic region. Indicates the first Stable signal values ​​(in mV) of a subset of signals in a characteristic segment. Indicates the first The weights of the signal subsets of each characteristic segment are dimensionless; This can be set as the preset dwell time for the corresponding node of the signal subset. In specific implementations, all signal subsets marked as "starting anchor points" or whose time coordinates are located in the background segment are extracted from the divide-and-conquer measurement data set in the same way. These signal subsets belong to the background segment, and the weighted average value of the background segment is calculated as the background signal intensity. The calculation formula is:

[0094]

[0095] in: This indicates the total number of signal subsets belonging to the background segment. Indicates the first Stable signal values ​​(in mV) of a subset of background signal segments. Indicates the first The weights of a subset of background signal segments are dimensionless. The characteristic signal intensity is... (Unit: mV) Subtract background signal strength (Unit: mV), to obtain a net response difference. (Unit: mV), that is This net response difference The input is fed into a preset concentration conversion model, which then adjusts the net response difference based on its internal parameters. Perform calculations and output the final hemoglobin concentration value. .

[0096] In some embodiments, the construction of the concentration conversion model is a separate, preliminary calibration process. This process begins with collecting multiple standard blood samples with known hemoglobin concentrations. The number of standard blood samples should be sufficient to cover the expected measurement range, for example, uniformly distributed from low to high concentrations. The known hemoglobin concentration values ​​are determined and recorded using recognized reference methods. Using the exact same biosensor model, measurement procedure, and signal processing method as the subsequent actual measurements, signal acquisition and processing are performed on each standard blood sample. This involves executing the complete steps from dynamic calibration measurement to the generation of a separate measurement dataset, ultimately obtaining the difference between the characteristic signal intensity and the background signal intensity for each standard blood sample. ,in It is an index of standard blood samples. It lists the known hemoglobin concentration values ​​for each standard blood sample. The difference data obtained from its calculation To link them together and form a data pair The data pairs from all standard blood samples are collected to construct the training dataset for model training. A linear regression algorithm is then used to fit the training dataset; the linear regression algorithm aims to find a straight line. The optimal slope parameter is calculated by fitting a line to minimize the sum of squared perpendicular distances from the line to all data points. and intercept parameter slope parameter and intercept parameter The core mapping relationship of the concentration conversion model was jointly defined. In practical implementation, to verify the accuracy of the concentration conversion model, another set of standard blood samples not used in the training is needed as an independent validation dataset. The difference data of each sample in the validation set is then processed using the constructed concentration conversion model. To make a prediction, the predicted concentration value is obtained. Calculate the predicted concentration value Compared with actual known concentration values The error between the training data and the target data can be evaluated using metrics such as mean absolute error, mean relative error, or root mean square error. When the error exceeds a preset allowable threshold, it indicates that the model is not fitting well and needs to be readjusted. This may include re-examining the quality of the training data, increasing the number of training samples, or re-performing linear regression to adjust the slope parameter. and intercept parameter .

[0097] Optionally, in the weighted difference operation, the weights and The setting, besides being based on dwell time, can also incorporate signal stability indicators during signal subset acquisition, such as the standard deviation of signal values. Signal subsets with smaller standard deviations can be assigned higher weights. Optionally, the concentration conversion model is not limited to a simple linear model; under certain response characteristics, a linearized nonlinear model can also be used. However, linear models are widely used due to their simplicity and clear physical meaning of parameters. In some embodiments, the construction of the training dataset may need to consider compensation for environmental factors such as temperature and sample matrix. Multiple sets of data can be collected under different conditions to construct a multiple linear regression model that includes environmental variables as input. It can be understood that weighted difference operations effectively eliminate systematic errors caused by sensor baseline drift and background interference, highlighting signal changes directly related to hemoglobin concentration. It can be understood that the construction of the concentration conversion model will reflect the relative response signal of the biosensor. Compared with absolute hemoglobin concentration values Connect them.

[0098] See Figure 5 This graph is the core visualization result of the concentration conversion model construction process. The horizontal axis represents the difference between the feature and background signals, and the vertical axis represents the hemoglobin concentration. Gray dots in the graph represent standard sample data, i.e., the correlation data between the net response difference and the actual concentration of a standard hemoglobin solution with a known concentration. The black line represents the linear fitting model, which is the mapping relationship obtained by fitting the standard sample data using a linear regression algorithm. This graph intuitively illustrates the conversion logic from signal difference to hemoglobin concentration: the standard sample data points are distributed along a linear trend, and the fitted line closely matches the data points, indicating a significant linear correlation between the net response difference and hemoglobin concentration. This model is the core output. In subsequent actual detection, simply substituting the net response difference of the sample to be tested into this model allows for rapid calculation of the corresponding hemoglobin concentration, providing a standardized mapping basis for the concentration quantification method.

[0099] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for in vitro measurement of hemoglobin concentration based on a biosensor, characterized in that, The method includes: Perform dynamic calibration measurements of the biosensor to generate biosensor calibration parameters adapted to the current blood sample to be tested; the dynamic calibration measurements of the biosensor include real-time response calibration of a sensitive unit of the biosensor using a standard signal during the sample introduction phase; Based on the biosensor calibration parameters, the raw optical or electrochemical signal streams acquired by the biosensor are subjected to multi-level signal classification to define the characteristic segments and background segments of the raw optical or electrochemical signals; the characteristic segments are identified as signal portions containing effective hemoglobin response information. Based on the defined feature segment and background segment, a nonlinear cooperative measurement path is planned within the detection time domain of the biosensor; the cooperative measurement path specifies the alternating sampling order and dwell time of the feature segment and background segment at multiple discrete measurement moments; The process involves planning a nonlinear collaborative measurement path within the detection time domain of the biosensor, based on defined feature segments and background segments. This includes: constructing an initial measurement path connecting the start and target anchor points, using the midpoint of the background segment as the starting anchor point and the extreme point of the signal amplitude within the feature segment as the target anchor point; inserting an intermediate measurement node along the initial measurement path based on the signal gradient change in the feature segment; the insertion density of the intermediate measurement node is proportional to the magnitude of the signal gradient; and connecting all anchor points and intermediate measurement nodes sequentially, allocating their respective signal acquisition dwell times to form a complete nonlinear collaborative measurement path. Based on the aforementioned collaborative measurement path, the biosensor is driven to perform divide-and-conquer measurements on the blood sample to be tested in vitro to obtain a divide-and-conquer measurement data set; the divide-and-conquer measurement data set contains multiple signal subsets corresponding to the characteristic segment and the background segment at different measurement times; Hierarchical fusion and solution are performed on the divide-and-conquer measurement data set to generate hemoglobin concentration values.

2. The method for in vitro measurement of hemoglobin concentration based on a biosensor according to claim 1, characterized in that, The process of performing dynamic calibration measurements of the biosensor to generate biosensor calibration parameters adapted to the current blood sample to be tested includes: While the biosensor is in contact with the blood sample to be tested, a standard hemoglobin solution of known concentration is loaded into the sensitive unit of the biosensor to obtain the dynamic baseline curve of the biosensor in a mixed response state. Feature points are extracted from the dynamic benchmark curve, including the rising inflection point, steady-state plateau, and falling inflection point of the dynamic benchmark curve. The biosensor calibration parameters are calculated based on the extracted feature points. These parameters include signal gain per unit concentration, response time constant, and baseline drift compensation.

3. The method for in vitro measurement of hemoglobin concentration based on a biosensor according to claim 2, characterized in that, Based on the biosensor calibration parameters, the raw optical or electrochemical signal stream acquired by the biosensor is subjected to multi-level signal typing to define the characteristic segments and background segments of the raw optical or electrochemical signal stream, including: The original optical or electrochemical signal stream is normalized and amplified using the unit concentration signal gain to obtain a preprocessed signal stream; In the preprocessed signal stream, continuous time segments whose signal change rate exceeds a preset threshold are identified based on the response time constant, and these continuous time segments are initially marked as feature segments. By combining the baseline drift compensation amount, baseline fitting and subtraction are performed on the signal portion outside the feature segment, and the subtracted stable signal portion is defined as the background segment.

4. The method for in vitro measurement of hemoglobin concentration based on a biosensor according to claim 3, characterized in that, The step of driving the biosensor to perform divide-and-conquer measurements on the blood sample to be tested in vitro, based on the aforementioned collaborative measurement path, to obtain a divide-and-conquer measurement data set, includes: The signal acquisition unit of the control biosensor moves sequentially to each anchor point and intermediate measurement node according to the order specified in the collaborative measurement path; At each anchor point and intermediate measurement node, the control signal acquisition unit acquires stable signals according to the dwell time specified by the cooperative measurement path, and binds the acquired stable signals with their corresponding node identifiers to generate a signal data packet; After traversing all nodes on the collaborative measurement path, all generated signal data packets are summarized by timestamp to form a divide-and-conquer measurement data set.

5. The method for in vitro measurement of hemoglobin concentration based on a biosensor according to claim 1, characterized in that, The hierarchical fusion and decomposition includes performing a weighted difference operation on the signal subset from the feature segment and the signal subset from the background segment, specifically: From the divide-and-conquer measurement data set, extract all signal subsets belonging to the characteristic segments respectively, calculate their weighted average value as the characteristic signal intensity; From the divide-and-conquer measurement data set, extract all signal subsets belonging to the background segment, calculate their weighted average value, and use it as the background signal intensity; The difference between the intensity of the feature signal and the intensity of the background signal is input into a preset concentration conversion model, which outputs the hemoglobin concentration value.

6. The method for in vitro measurement of hemoglobin concentration based on a biosensor according to claim 5, characterized in that, The construction of the concentration conversion model includes: Multiple standard blood samples with known hemoglobin concentrations were collected, and a biosensor was used to acquire signals from each standard blood sample to obtain the difference data between the corresponding characteristic signal intensity and the background signal intensity. The difference data is correlated with known hemoglobin concentration values ​​to construct a training dataset; The training dataset was fitted using a linear regression algorithm to obtain the slope and intercept parameters of the concentration conversion model. To verify the accuracy of the concentration conversion model, the error between the model's predicted concentration and the actual concentration was calculated using an independent validation dataset. When the error exceeded a preset threshold, the slope and intercept parameters were readjusted. The independent validation dataset consists of standard blood samples that were not used in the training process. The readjusted slope and intercept parameters are stored in the concentration conversion model for real-time hemoglobin concentration calculation.

7. The method for in vitro measurement of hemoglobin concentration based on a biosensor according to claim 3, characterized in that, The process of normalizing and amplifying the original optical or electrochemical signal stream using unit concentration signal gain to obtain a preprocessed signal stream includes: Read the raw optical or electrochemical signal streams acquired by the biosensor and obtain the baseline voltage value of the raw optical or electrochemical signal streams; The normalized amplification factor is calculated based on the unit concentration signal gain, where the normalized amplification factor is the reciprocal of the unit concentration signal gain. The baseline correction signal is obtained by subtracting the baseline voltage value from each data point of the original optical or electrochemical signal stream. Multiply each data point of the baseline correction signal by the normalization amplification factor to obtain the amplitude normalized signal; The amplitude-normalized signal is filtered by moving average to eliminate random fluctuations and output a smooth preprocessed signal stream.

8. The method for in vitro measurement of hemoglobin concentration based on a biosensor according to claim 3, characterized in that, The initial measurement path, which uses the midpoint of the time interval in the background segment as the starting anchor point and the extreme points of the signal amplitude within the feature segment as the target anchor points, and connects the starting anchor point and the target anchor point, includes: Identify the start and end times of the background segment, calculate the midpoint of the time for the background segment, and set the midpoint as the starting anchor point. Scan the signal amplitude within the feature segment, find the maximum or minimum point of the signal amplitude as the extreme point, and set the time point corresponding to the extreme point as the target anchor point. A straight path is generated between the starting anchor point and the target anchor point in chronological order as the initial measurement path.

9. The method for in vitro measurement of hemoglobin concentration based on a biosensor according to claim 4, characterized in that, At each anchor point and intermediate measurement node, the control signal acquisition unit acquires stable signals according to the dwell time specified by the cooperative measurement path, and binds the acquired stable signals with their corresponding node identifiers to generate a signal data packet, including: Read the location information of the current node on the collaborative measurement path, including node type, time coordinates, and dwell time; The control signal acquisition unit moves to the time coordinate corresponding to the current node and starts signal acquisition; During the dwell time, multiple sets of signal values ​​are continuously collected, and the average value of these signal values ​​is calculated as a stable signal. Generate a node identifier, which includes a node sequence number, a node type, and a timestamp; The stable signal is combined with the node identifier to form a signal data packet, which includes identifier information in the header and signal value in the data section.

Images