Respiratory segment positioning method and device under voice instruction guidance, equipment and medium

Abstract

The invention relates to the field of respiratory monitoring, in particular to a respiratory segment positioning method and device under voice instruction guidance, equipment and a medium. A radar non-contact detection technology is adopted, a test signal is generated by capturing thoracic and abdominal respiratory movement, contact type air blowing operation is not needed, dependence on contact type acquisition equipment is abandoned, and the problem that the operation threshold is high is solved; secondly, testing actions are guided through voice instructions instead of guidance of professional medical staff, and the requirement for operation speciality is lowered through standardized instructions; then, the radar is flexible to install and adapts to various postures, and the limitation of professional medical scenes is broken through; and finally, automatically extracting the breathing segment and calculating the breathing index of the measured object without complicated manual intervention, thereby simplifying the process and reducing the equipment and use cost.

Description

Technical Field

[0001] This invention relates to the field of respiratory monitoring, specifically to a method, device, equipment, and medium for locating respiratory segments under voice command guidance. Background Technology

[0002] In the diagnosis and assessment of chronic obstructive pulmonary disease (COPD), pulmonary function testing (PFT) is the clinically recognized gold standard. Its core involves using a pulmonary function analyzer to collect expiratory flow rate and volume data from the subject's mouth, generating respiratory mechanics curves, and then calculating key indicators such as forced vital capacity (FVC), forced expiratory volume in one second (FEV1), and FEV1 / FVC ratio. Clinically, a FEV1 / FVC ratio below 70% usually indicates persistent airflow limitation and can serve as an important diagnostic criterion for COPD. However, existing pulmonary function analyzer-based testing methods have significant limitations in practical applications: On the one hand, the testing process requires the test subject to hold the mouthpiece in their mouth and perform forced breathing, which is difficult for the elderly and other groups with weaker physical functions, and may lead to test failure or data distortion.

[0003] On the other hand, the testing scenarios are highly limited, and can only be carried out in hospitals or large medical institutions, which cannot meet the rapid testing needs in scenarios such as community screening and daily family monitoring; at the same time, the testing process requires guidance from professional medical staff, making it difficult to achieve contactless, low-cost, and convenient widespread application. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method, apparatus, device and medium for respiratory segment localization guided by voice commands, in order to solve the problem that current lung function testing devices use a contact-based blowing method for data collection, and are dependent on the testing scenario and professional operation, which makes it impossible to achieve contactless, low-cost and convenient lung function testing.

[0005] In a first aspect, embodiments of the present invention provide a method for locating respiratory segments under voice command guidance, the method comprising: The test signal is obtained by using radar to detect the test actions performed by the test object according to the voice command; Based on the test signal, the respiratory rate waveform and respiratory signal of the chest and abdomen of the tested object during the test process are extracted; The respiratory signal is analyzed to obtain the segment index representing the chest and abdomen after the subject inhales until the lungs are full and then exhales. The target respiratory segment is extracted from the respiratory rate waveform according to the segment index, and the respiratory index corresponding to the subject is calculated based on the target respiratory segment.

[0006] Furthermore, the analysis of the respiratory signal yields an index of segments characterizing the chest and abdomen during the subject's inhalation until the lungs are full and then exhalation, including: The respiratory signal is averaged according to the distance gate dimension to obtain the displacement signal, and the displacement signal is then averaged according to the time dimension to obtain the smoothed respiratory signal; Detect multiple key points of different types in the smooth breathing signal; Candidate keypoint combinations are obtained by pairing multiple keypoints of different types. According to preset conditions, target signal segments representing the chest and abdomen are selected from the respiratory signals corresponding to the candidate key point combinations to characterize the exhalation of the lungs after the subject inhales until the lungs are full, and the segment index corresponding to the target signal segments is extracted.

[0007] Furthermore, the process of pairing multiple keypoints of different types to obtain candidate keypoint combinations includes: Classify multiple different types of key points to obtain a set of key points corresponding to each type; For each type of key point set, three adjacent key points are selected sequentially and paired to obtain basic candidate units; Based on the corresponding adjacent three-point pairing results of all types of key points, candidate key point combinations containing the complete respiratory cycle are constructed.

[0008] Furthermore, the step of selecting segments representing the chest and abdomen during exhalation after the subject inhales until the lungs are full, from the respiratory signals corresponding to the candidate key point combinations according to preset conditions, includes: Obtain the signal features of the respiratory signal segment corresponding to each group of candidate key point combinations; Compare the aforementioned signal characteristics with the preset signal characteristics corresponding to the exhalation action after inhalation until the lungs are full; If the signal feature matches the preset signal feature, then the respiratory signal segment corresponding to the corresponding candidate key point combination is taken as the target signal segment.

[0009] Furthermore, the step of extracting the target respiratory segment from the respiratory rate waveform based on the segment index includes: Determine the time range corresponding to the segment index in the respiratory signal; The target respiratory segment is extracted from the respiratory rate waveform using the time range corresponding to the segment index.

[0010] Furthermore, after extracting the target respiratory segment from the respiratory rate waveform according to the segment index, the method further includes: The expiratory segment is extracted based on the target respiratory segment, and the first integral value of the first respiratory rate signal within the expiratory segment and the baseline is calculated. Calculate the second respiratory rate signal within one second of the start of the expiratory phase, and calculate the second integral value of the second respiratory rate signal and the baseline; The one-second rate is calculated based on the first integral value and the second integral value.

[0011] Furthermore, the method also includes: Based on the chest and abdominal region signals corresponding to different distance gates in the test signal, calculate the energy percentage of the respiratory velocity waveform in each region. The current body position of the tested object is determined based on the energy proportion of the respiratory rate waveform in each region. Based on the current body position, extract the steady segment signal before inspiration from the target respiratory segment, and calculate the mean of the steady segment as the initial baseline under the current body position; The initial baseline is corrected using the empirical value of the baseline offset corresponding to the current body position to obtain the baseline.

[0012] Secondly, embodiments of the present invention provide a voice-command-guided respiratory segment localization device, the device comprising: The testing module is used to detect the test actions performed by the test object according to voice commands using radar, and to obtain test signals; The extraction module is used to extract the respiratory rate waveform and respiratory signal of the chest and abdomen of the tested object during the test based on the test signal; The analysis module is used to analyze the respiratory signal to obtain the segment index representing the chest and abdomen after the subject inhales until the lungs are full and then exhales, and to extract the target respiratory segment from the respiratory rate waveform according to the segment index, and to calculate the respiratory index corresponding to the subject based on the target respiratory segment.

[0013] Thirdly, embodiments of the present invention provide a computer device, including: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the method described in the first aspect or any corresponding embodiment thereof.

[0014] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer instructions that cause a computer to perform the method described in the first aspect or any of its corresponding embodiments.

[0015] This application employs non-contact radar detection technology, generating test signals by capturing chest and abdominal respiratory movements. It eliminates the need for contact-based air-blowing operations, thus avoiding reliance on contact-based data acquisition equipment and resolving the issue of high operational barriers. Secondly, it uses voice commands to guide testing actions, replacing professional medical personnel guidance and reducing the professional requirements of operation through standardized instructions. Thirdly, the radar is flexible in installation and adaptable to various postures, overcoming the limitations of professional medical scenarios. Finally, it automatically extracts respiratory segments and calculates the respiratory indicators of the tested subject without complex manual intervention, simplifying the process while reducing equipment and operating costs. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a method for locating respiratory segments guided by voice commands according to some embodiments of the present invention. Figure 2 This is a schematic diagram of a radar non-contact breathing test according to some embodiments of the present invention; Figure 3 This is a schematic diagram of the respiratory rate waveform and respiratory Doppler signal obtained after preprocessing the test signal according to some embodiments of the present invention; Figure 4 This is a schematic diagram of an effective deep inhalation and rapid exhalation target breathing segment according to some embodiments of the present invention; Figure 5 This is a flowchart illustrating a method for locating another respiratory segment according to some embodiments of the present invention; Figure 6 This is a schematic diagram illustrating the calculation of lung function indicators based on the expiratory phase according to some embodiments of the present invention; Figure 7 This is a structural block diagram of a respiratory segment localization method apparatus according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0019] According to embodiments of the present invention, a method, apparatus, device, and medium for respiratory segment localization guided by voice commands are provided. It should be noted that the steps shown in the flowcharts in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowcharts, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0020] This embodiment provides a method for locating respiratory segments guided by voice commands. Figure 1 This is a flowchart of a voice command-guided respiratory segment localization method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps: Step S101: Use radar to detect the test actions performed by the test object according to the voice command, and obtain the test signal.

[0021] In this embodiment, during the test preparation phase, the radar installation location and the subject's posture need to be determined based on the subject's physical condition and comfort requirements. Priority is given to test scenarios that allow the subject to relax physically and mentally. If the subject has limited mobility, a lying position can be used, with the radar installed directly above the subject's chest, for example, 50cm from the bed surface. It should be fixed with a bracket to ensure the detection direction is directly facing the chest and abdomen area, ensuring the radar beam completely covers the chest and abdominal movement area. If the subject is in good condition, a sitting or standing position can be chosen, with the radar installed at a suitable height directly in front of the chest, avoiding interference from clothing wrinkles or surrounding obstacles. After the test starts, the voice system will guide the entire process according to a preset procedure. First, relaxation prompts will be played, such as "Please adjust your sitting posture, relax your whole body, and maintain natural breathing for 30 seconds," allowing the subject's breathing to stabilize. At this time, the radar begins warm-up and enters signal acquisition standby mode.

[0022] Once the subject's breathing stabilizes, a voice command clearly states the test action requirements, such as "Please inhale deeply until your lungs are fully full, then resume natural breathing." Simultaneously, the radar begins signal acquisition, capturing minute displacements in the chest and abdomen caused by respiration. The entire test includes multiple sets of repetitive deep-inhale-rapid-exhale movements, typically with three core test actions. Between each set, the voice system inserts a recovery prompt, such as "Please breathe naturally 5 times, prepare for the next set," ensuring the subject does not experience shortness of breath due to continuous testing. During acquisition, the radar converts the echo signal into a test signal, which is then transmitted to a computer via a data transmission module. This transmission module can be implemented via, but is not limited to, Bluetooth communication, WiFi connection, or direct data cable connection. Anti-interference coding technology is used during transmission to prevent signal loss or distortion, ultimately forming a complete test signal set including natural breathing, deep-inhale-rapid-exhale, and action interval segments, providing the original basis for subsequent data processing.

[0023] For example, such as Figure 2 As shown in the figure, a scenario of non-contact breathing test using radar is illustrated: the subject lies flat on the bed, and the radar is installed 50cm above the bed surface and directly above the subject's chest, monitoring the subject's chest movement through the emitted signals; at the same time, a voice device issues standardized voice commands such as "inhale...exhale..." to guide the subject in the test; the radar echo signal is transmitted to the computer in real time via data transmission (such as Bluetooth), and the computer will automatically extract the deep inhale and rapid exhale segments and calculate lung function indicators.

[0024] Step S102: Extract the respiratory rate waveform and respiratory signal of the chest and abdomen of the test subject during the test based on the test signal.

[0025] In the embodiments of this application, the test signal initially received is mixed with environmental noise (such as air flow, equipment operation noise), interference from other parts of the human body (such as limb micro-movement, heartbeat vibration) and electromagnetic interference. Therefore, the first step is to filter out the noise.

[0026] After signal purification, the processed signal is decomposed by time-frequency transformation to convert the time-domain signal into a two-dimensional time-frequency spectrum, from which the frequency components corresponding to respiratory movements are identified. Normal respiratory frequency is usually between 0.1-0.3Hz, and the frequency of deep inhalation and rapid exhalation will briefly increase. Based on this, the respiratory-related signal range can be initially locked.

[0027] Next, the displacement information corresponding to the respiratory motion is extracted by the phase demodulation algorithm. Combined with the Doppler effect principle of radar, the displacement change is converted into the motion velocity data of the chest and abdomen. After data smoothing and standardization, a continuous respiratory velocity waveform is generated. The peak and trough of the waveform correspond to the maximum velocity point and the rest point of the respiratory motion, respectively.

[0028] Simultaneously, amplitude features are extracted from the signal, quantifying the amplitude changes caused by respiratory movements. After removing abnormal fluctuations, a respiratory signal is formed, whose period and amplitude are directly related to respiratory depth and frequency. During the extraction process, the timestamps corresponding to the signal must be simultaneously labeled to ensure that the respiratory velocity waveform and the respiratory signal are perfectly aligned in the time dimension.

[0029] For example, Figure 3 The diagram shows the respiratory rate waveform and respiratory Doppler signal obtained after preprocessing the test signal. Figure 3 The top image (respiratory velocity waveform) shows the velocity changes in the subject's chest and abdomen caused by breathing, with time (seconds) on the horizontal axis and amplitude on the vertical axis. The fluctuations in the waveform reflect the speed differences of chest and abdominal movements during breathing. The peaks, troughs, and other characteristic points can be used to identify the stages of the breathing action. For example, a deep inhalation followed by a rapid exhalation will show more obvious waveform fluctuations.

[0030] Figure 3 The image below (respiratory Doppler thermogram) shows the following: the horizontal axis represents time (seconds), and the vertical axis represents the distance gate (which can be understood as different spatial regions monitored by radar). The intensity of the respiratory Doppler signal at different distance gates and times is represented by the color intensity (color scale from -0.5 to 0.5). The color changes reflect the movement characteristics of the chest and abdomen at different spatial locations and time points, which helps to analyze the spatial distribution and temporal patterns of respiratory movements from multiple dimensions.

[0031] Step S103: Analyze the respiratory signal to obtain the segment index representing the chest and abdomen during the subject's inhalation to lung fullness and exhalation, extract the target respiratory segment from the respiratory rate waveform according to the segment index, and calculate the corresponding respiratory index of the subject based on the target respiratory segment.

[0032] In this embodiment of the application, analyzing the respiratory signal to obtain a segment index characterizing the chest and abdomen during the subject's inhalation until the lungs are full and then exhaling includes the following steps A1-A4: Step A1: Average the respiratory signal according to the distance gate dimension to obtain the displacement signal, and then average the displacement signal according to the time dimension to obtain the smoothed respiratory signal.

[0033] Specifically, the respiratory signal is generated by radar multi-range gate acquisition. Each range gate corresponds to a different spatial position in the chest and abdomen region of the tested object. Due to subtle differences in the amplitude of chest and abdominal movements at different positions, and the presence of interference noise between range gates in the signal, directly using the raw signal would lead to feature extraction bias. Therefore, a range gate dimensional averaging process is first performed: a preset signal fusion algorithm is invoked, traversing the respiratory signal data acquired by all valid range gates, and the signal amplitudes of different range gates at the same time stamp are arithmetically averaged. By fusing multi-spatial-dimensional data, local noise interference is canceled out, while preserving the core features of the overall chest and abdominal movement. Finally, a continuous displacement signal is output, and the amplitude change of this signal directly corresponds to the displacement fluctuations of the chest and abdomen caused by respiration.

[0034] After distance-gated averaging, further elimination of temporal noise and random fluctuations in the displacement signal is required, necessitating time-dimensional moving average processing. First, the moving window parameters are set, taking into account the physiological characteristics of respiratory signals (a normal respiratory cycle is approximately 3-5 seconds, and a deep inhalation / rapid exhalation cycle is approximately 3 seconds). The window length is set to 0.2 seconds, and the window sliding step size is set to 0.05 seconds, ensuring that critical details of respiratory movements are not lost while smoothing noise. During processing, using the time axis as a reference, the moving window is sequentially moved along the displacement signal. A weighted average is calculated for all signal points within each window, with the highest weight at the center of the window, gradually decreasing towards both ends. This weighting method reduces phase shift during signal smoothing. Once the window slides to cover the entire displacement signal, the generated smooth breathing signal will exhibit continuous and gentle waveform characteristics. The spikes and jumps caused by breathing airflow disturbances and small fluctuations in radar signals are effectively filtered out. The rising edge, falling edge, and peak position of the chest and abdominal movements are clearer, providing high signal-to-noise ratio data support for accurate identification of various key points in subsequent steps. The entire processing is automatically completed by the signal processing module, and the processing results are stored in real time and synchronized to the subsequent analysis unit.

[0035] Step A2: Detect multiple key points of different types in the smooth breathing signal.

[0036] First, the types of key points to be detected are clearly defined. Combining the waveform characteristics and physiological significance of the respiratory signal, five types of key points are identified, including zero-crossing points, peak points, trough points, and upper and lower 30% quantiles. Zero-crossing points correspond to the transition moments of respiratory actions (inhalation to exhalation or exhalation to inhalation), peak and trough points correspond to the extreme positions of lung filling and emptying, respectively, and the upper and lower 30% quantiles are used to help define the beginning and end stages of respiratory actions. Before detection, the smoothed respiratory signal needs to be preprocessed. Mean normalization maps the signal amplitude uniformly to the 0-1 interval, eliminating detection bias caused by differences in signal amplitude among different subjects. Simultaneously, a first-order difference algorithm is used to calculate the rate of change of the signal, providing rate characteristics for key point identification.

[0037] Differentiated detection algorithms are used to address the characteristics of different types of key points. For zero crossings, the critical point where the signal amplitude changes from positive to negative or from negative to positive is located by traversing the signal sequence. This is then verified by combining the sign change of the first-order differential signal. If the differential signal shows a sign jump simultaneously, the point is confirmed as a valid zero crossing, thus avoiding misjudgment caused by small signal fluctuations. For peak and valley points, the sliding window extreme value detection method is used, with a window of 0.3 seconds to traverse the signal. When the signal amplitude at the center position of the window is greater than (or less than) all other points in the window, and the first-order difference signal of that point changes from positive to negative (or from negative to positive), and the second-order difference signal is less than (or greater than) zero, it is determined to be a peak point (or valley point). For the upper and lower 30% quantiles, first calculate the maximum and minimum amplitude values ​​of the entire signal sequence to determine the amplitude dynamic range. Then, calculate the amplitude thresholds corresponding to 30% and 70% of this range respectively. When traversing the signal, locate the points where the threshold is first reached and the points where the threshold is first deviated from. Combine the signal change rate to confirm the quantile position.

[0038] All detected key points must have their precise timestamps and amplitude information recorded to form a clearly categorized key point sequence. At the same time, outliers are eliminated through neighborhood verification—if the time interval between a key point and an adjacent point of the same type is less than 0.1 seconds, it is determined to be a false key point and deleted. The final output key point sequence will fully reflect the waveform characteristics of the respiratory signal.

[0039] Step A3: Pair multiple key points of different types to obtain candidate key point combinations.

[0040] Specifically, candidate keypoint combinations are obtained by pairing multiple keypoints of different types, including: classifying multiple keypoints of different types to obtain a set of keypoints corresponding to each type; selecting three adjacent keypoints in sequence for each set of keypoints corresponding to each type to obtain basic candidate units; and constructing candidate keypoint combinations containing the complete respiratory cycle based on the pairing results of the three adjacent points corresponding to all types of keypoints.

[0041] Multiple keypoints of different types detected (such as zero-crossing points, peak points, and upper 30% quantiles) are categorized and organized according to their type. Based on the keypoint's characteristic attributes (such as zero-crossing points representing the moment of respiratory transition and peak points representing the lung filling limit), corresponding keypoint sets are created for each set. The original timestamp, amplitude, and signal location information of each keypoint within a set are retained to ensure the feature integrity of each type of keypoint after classification. Next, for each type of keypoint set, the keypoints within the set are sorted according to their timestamps. Then, three adjacent keypoints are selected sequentially as a group for pairing to form basic candidate units. For example, in the peak point set, if the time-ordered sequence is P1, P2, P3, P4…, then (P1, P2…) are constructed sequentially. Basic candidate units such as (P2, P3, P4) are constructed, each corresponding to the characteristic changes of the key point of the corresponding type within a continuous time interval. Finally, the pairing results of adjacent three points corresponding to all types of key points are summarized. Combined with the physiological laws of the respiratory cycle (such as a complete respiratory cycle must include key stages such as the beginning of inhalation, the process of inhalation, the end of inhalation, the process of exhalation, and the end of exhalation), the basic candidate units of different types are cross-type associated and integrated. Invalid combinations that are not discontinuous in time and cannot form a complete respiratory stage are eliminated. Finally, candidate key point combinations that can completely cover the entire process of "inhalation-exhalation" or "exhalation-inhalation" are constructed to ensure that each combination can correspond to a signal segment with complete respiratory characteristics, providing comprehensive candidate samples for subsequent screening of deep inhalation and rapid exhalation segments.

[0042] Step A4: Select target signal segments representing the chest and abdomen during inhalation and exhalation after the lungs are filled from the respiratory signals corresponding to the candidate key point combinations according to preset conditions, and extract the segment index corresponding to the target signal segments.

[0043] Specifically, according to preset conditions, segments representing the chest and abdomen during inhalation to lung fullness and exhalation are selected from the respiratory signals corresponding to candidate key point combinations. This includes: acquiring the signal features of the respiratory signal segments corresponding to each candidate key point combination; comparing the signal features with preset signal features corresponding to the inhalation to lung fullness and exhalation action; if the signal features match the preset signal features, then the respiratory signal segment corresponding to the corresponding candidate key point combination is taken as the target signal segment.

[0044] First, for each candidate key point combination, based on the timestamp information of each key point within the combination, the corresponding signal segment is accurately extracted from the complete respiratory signal. Then, the core signal features of the segment are extracted, including time features (duration of the inspiratory phase, duration of the expiratory phase, and total duration of the entire segment), amplitude features (amplitude increase during inspiratory process, difference between peak and trough amplitude, and percentage of peak relative to the overall maximum value of the signal), and rate features (average rate of increase of signal amplitude during inspiratory phase, average rate of decrease of signal amplitude during expiratory phase, and ratio of respiratory rate). These feature parameters are then organized into a standardized feature vector.

[0045] Next, the pre-stored preset signal characteristic thresholds corresponding to the action of "inhaling until the lungs are full and then exhaling" are retrieved. These thresholds are based on a large amount of clinical sample data. For example, the duration of the inhalation phase should be 0.8-1.5 seconds, the duration of the exhalation phase should be 1.2-3.5 seconds, the difference between the peak and trough amplitudes should be more than 1.8 times the difference between the average amplitudes of the signals, and the ratio of the exhalation rate to the inhalation rate should be greater than 1.2.

[0046] Finally, the extracted feature vectors of each candidate segment are compared with the preset feature thresholds one by one. If all feature parameters of a segment fall within the preset threshold range and the key features (such as peak amplitude ratio and respiratory rate ratio) meet the core matching conditions, the signal features of the segment are determined to match the target action features, and it is identified as the target signal segment representing "inhalation to lung filling and then exhalation". At the same time, the candidate key point combination information and the signal start and end positions corresponding to the segment are recorded.

[0047] In this embodiment of the application, extracting the target respiratory segment from the respiratory velocity waveform according to the segment index includes the following steps B1-B2: Step B1: Determine the time range corresponding to the fragment index in the respiratory signal.

[0048] Specifically, first, the basic parameters for respiratory signal acquisition are retrieved to determine the signal sampling frequency (e.g., 100Hz represents 100 signal points per second) and the acquisition start timestamp. Based on this, a conversion formula for "segment index - time" is established: start time of the time range = acquisition start time + (start index ÷ sampling frequency), end time = acquisition start time + (end index ÷ sampling frequency). Then, the segment index corresponding to the target signal segment is extracted (e.g., z1 is the start index, z2 is the end index). Combining this with the above formula, the specific range of this index interval on the time axis is calculated. For example, if the sampling frequency is 100Hz, the acquisition start time is 0 seconds, the start index z1 = 500, and the end index z2 = 1200, then the corresponding time range is 5 seconds to 12 seconds.

[0049] Simultaneously, verification is required in conjunction with the timestamp records of the respiratory signal. The calculated time range is compared with the timestamps of key points (such as peak points and zero-crossing points) within the index interval of the signal. If the error is within ±0.01 seconds, the time range is confirmed to be valid. If the error exceeds the threshold, the sampling frequency and index value are rechecked to ensure that the correspondence between the segment index and the respiratory signal time range is accurate, providing a precise time positioning basis for the subsequent extraction of respiratory velocity waveform segments.

[0050] Step B2: Extract the target respiratory segment from the respiratory rate waveform using the time range corresponding to the segment index.

[0051] Specifically, firstly, the respiratory rate waveform data is sorted by timestamp to construct a waveform time series that is completely synchronized with the respiratory signal timeline, ensuring a one-to-one correspondence between the two in the time dimension. Next, based on the time range obtained in step B1 (e.g., 5 seconds to 12 seconds), the start and end positions of this interval are located in the respiratory rate waveform time series, and the corresponding waveform data point indices are marked—if the waveform sampling frequency is consistent with the respiratory signal, the segment indices z1 and z2 can be directly used for positioning; if the sampling frequencies are different, the corresponding indices of the waveform data are obtained by time ratio conversion.

[0052] Subsequently, based on the located index interval, the data segment is extracted from the complete respiratory rate waveform, while preserving core feature information such as waveform amplitude and rate of change. Lossless data truncation must be used during extraction to avoid waveform distortion caused by interpolation or compression. After extraction, the target respiratory segment needs to be verified for completeness, checking whether it contains the complete waveform features corresponding to the respiratory action (such as the rising edge and peak point of the inspiratory phase and the falling edge of the expiratory phase). The segment is then compared with the corresponding respiratory signal segment in time synchronization to ensure a complete match in the action phase division. Finally, the target respiratory segment containing time stamps and feature parameters is output.

[0053] For example, Figure 4 The image shows the target respiratory segments representing effective deep inhalation and rapid exhalation extracted after conditional filtering. It includes three deep inhalation and rapid exhalation segments: cycle 1, cycle 2, and cycle 3. Each cycle is labeled with the duration of inhalation and exhalation (e.g., cycle 1 inhalation is 6.3s and exhalation is 1.1s). After extracting each respiratory segment through the index corresponding to the deep inhalation and rapid exhalation segment, the inspiratory and expiratory segments can be distinguished, and the respiratory rate waveform characteristics of each cycle are presented.

[0054] In the embodiments of this application, after extracting the target respiratory segments, the respiratory rate (unit: breaths / minute) can be calculated by counting the number of target segments and the total duration, i.e., the total number of segments divided by the total duration and converted to minutes; the respiratory depth can also be calculated by extracting the maximum and minimum respiratory velocity values ​​in each target segment, i.e., the absolute value of the difference between the two; at the same time, the integral area of ​​the velocity waveform within the segment can be further analyzed to obtain the estimated ventilation volume for each breath, or the respiratory rhythm stability can be evaluated by calculating the coefficient of variation of the waveform to obtain the respiratory index of the subject.

[0055] This application employs non-contact radar detection technology, generating test signals by capturing chest and abdominal respiratory movements. It eliminates the need for contact-based air-blowing operations, thus avoiding reliance on contact-based data acquisition equipment and resolving the issue of high operational barriers. Secondly, it uses voice commands to guide testing actions, replacing professional medical personnel guidance and reducing the professional requirements of operation through standardized instructions. Thirdly, the radar is flexible in installation and adaptable to various postures, overcoming the limitations of professional medical scenarios. Finally, it automatically extracts respiratory segments and calculates the respiratory indicators of the tested subject without complex manual intervention, simplifying the process while reducing equipment and operating costs.

[0056] In this embodiment of the application, after extracting the target respiratory segment from the respiratory velocity waveform according to the segment index, as follows: Figure 5 As shown, the method also includes: Step S201: Extract the expiratory segment based on the target respiratory segment, and calculate the first integral value of the first respiratory rate signal within the expiratory segment and the baseline.

[0057] First, from the identified target breathing segment of the "deep inhale, rapid exhale" mode, the expiratory segment is divided based on the characteristics of the breathing velocity waveform. Combining respiratory physiology and waveform characteristics, the interval where the breathing velocity waveform rapidly declines from the peak point and crosses the baseline y=0 until the next trough point is the expiratory segment. The start and end positions of this interval are accurately located and extracted using segment indexing. Then, using the baseline y=0 as a reference, a numerical integration algorithm (such as the trapezoidal integral method) is used to integrate the breathing velocity signal within the expiratory segment: the time axis of the expiratory segment is divided into several small time intervals, and the area of ​​the trapezoid formed by the breathing velocity signal and the baseline y=0 in each interval is calculated. All trapezoidal areas are summed to obtain the preliminary integration result. Since the integration result may be negative due to the signal direction, its absolute value needs to be taken. This absolute value is the first integral value representing the forced vital capacity (FVC). Sufficient numerical precision must be maintained during the calculation process to ensure the accuracy of the indicator.

[0058] Step S202: Calculate the second respiratory rate signal within one second of the start of the expiratory phase, and calculate the second integral value of the second respiratory rate signal and the baseline.

[0059] In practice, the precise start timestamp of the expiratory phase (i.e., the time corresponding to the peak of the waveform) is first determined. A 1-second segment of the signal is then extracted from this starting point to obtain the second respiratory rate signal. If the total duration of the expiratory phase is less than 1 second, the entire expiratory phase signal is used as the second respiratory rate signal. The actual extraction time is recorded for later explanation. Next, using the baseline y=0 as a reference, the signal within this 1-second interval is integrated: the 1-second time axis is subdivided into multiple small intervals. The area enclosed by the signal and the baseline within each interval is calculated and accumulated. The absolute value of the integral result for this interval is taken; this absolute value represents the second integral value representing the forced expiratory volume in the first second (FEV1). After calculation, the signal duration corresponding to this integral must be labeled. If the duration is less than 1 second, this must be clearly stated in the data record to ensure the integrity and traceability of the indicators.

[0060] Step S203: Calculate the one-second rate based on the first integral value and the second integral value.

[0061] Based on the first integral value (FVC) and the second integral value (FEV1), validity is verified. If FVC is 0, the integral result is deemed invalid, and the previous steps must be repeated to extract the expiratory segment and recalculate the integral. If the signal duration corresponding to FEV1 is less than 1 second, the calculation result must be marked with the information "FEV1 calculated based on a signal of less than 1 second". After verification, the one-second rate is calculated according to the definition: one-second rate = second integral value (FEV1) ÷ first integral value (FVC), retaining two decimal places to meet the accuracy requirements of clinical diagnosis. After calculation, the one-second rate is compared with the reference threshold for COPD clinical diagnosis (usually 70%), and the specific value and corresponding diagnostic reference opinion are output to provide data support for subsequent medical diagnosis. At the same time, the specific values ​​of FVC, FEV1, and one-second rate, as well as the calculation time, are simultaneously stored in the database.

[0062] For example, Figure 6 This demonstrates the process of calculating lung function indicators based on the expiratory phase. Figure 6 The image above shows the respiratory rate waveform, with the baseline y=0 as a reference. The deep inhalation and rapid exhalation segments marked are used to locate the expiratory phase. Figure 6 The image below is presented intuitively through integral calculations, including forced expiratory volume in one second (FEV1) and forced vital capacity (FVC). These two parameters can be used to calculate the one-second rate (FEV1 / FVC), which can assist in the diagnosis of COPD.

[0063] In this embodiment of the application, the method further includes: calculating the energy proportion of the respiratory velocity waveform of each region based on the chest and abdominal region signals corresponding to different distance gates in the test signal; determining the current body position of the subject based on the energy proportion of the respiratory velocity waveform of each region; extracting the steady segment signal before inspiration from the target respiratory segment according to the current body position, and calculating the mean of the steady segment as the initial baseline under the current body position; and correcting the initial baseline using the baseline offset empirical value corresponding to the current body position to obtain the baseline.

[0064] Specifically, firstly, for the chest and abdominal region signals corresponding to different distance gates in the test signal, energy values ​​of the respiratory velocity waveform in each region are calculated using energy calculation methods such as Fourier transform. Then, the energy proportion is obtained by the ratio of the energy value of each region to the total energy. Next, based on the distribution characteristics of the energy proportion of each region (such as the difference in energy proportion of different chest and abdominal regions under lying, sitting, and standing positions), the current body position of the test subject is determined. Then, based on the determined current body position, the steady breathing phase before the inspiratory action is located from the target respiratory segment, the signal of this steady segment is extracted, and its mean is calculated using algorithms such as moving average, which serves as the initial baseline under the current body position. Finally, the baseline offset empirical value corresponding to the current body position is retrieved (this empirical value is statistically derived from a large amount of test data in the same body position), and the initial baseline is calculated with the empirical value (such as by addition or adjustment) to complete the correction of the initial baseline, thus obtaining the final baseline used for calculating lung function indicators.

[0065] This application embodiment uses signals from different distance gate regions to calculate the energy ratio to determine body position, which can accurately adapt to different states of the tested subject, such as lying down and sitting, and avoid baseline deviation caused by body position differences. Secondly, it extracts the signal of the steady segment before inspiration and uses the mean as the initial baseline to ensure that the baseline fits the current respiratory baseline state and reduces respiratory fluctuation interference. Finally, it combines the baseline offset experience value corresponding to the body position to correct the initial baseline, further offsetting the systematic error related to body position, making the baseline more in line with the actual physiological scenario, and effectively improving the accuracy of subsequent lung function index calculation.

[0066] This embodiment also provides a voice-guided respiratory segment localization device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0067] This embodiment provides a voice-command-guided respiratory segment localization device, such as... Figure 7 As shown, it includes: Test module 701 is used to detect the test actions performed by the test object according to the voice command using radar, and obtain test signals; The extraction module 702 is used to extract the respiratory rate waveform and respiratory signal of the chest and abdomen of the test subject during the test based on the test signal; Analysis module 703 is used to analyze the respiratory signal to obtain the segment index representing the chest and abdomen during the subject's inhalation to lung filling and then exhalation, and to extract the target respiratory segment from the respiratory rate waveform based on the segment index.

[0068] In this embodiment, the analysis module 703 is used to average the respiratory signal according to the distance gate dimension to obtain a displacement signal, and to slide average the displacement signal according to the time dimension to obtain a smoothed respiratory signal; detect multiple different types of key points in the smoothed respiratory signal; pair the multiple different types of key points to obtain candidate key point combinations; and select target signal segments representing the chest and abdomen during the subject's inhalation to lung filling and exhalation according to preset conditions from the respiratory signals corresponding to the candidate key point combinations, and extract the segment index corresponding to the target signal segments.

[0069] In this embodiment of the application, the analysis module 703 is specifically used to classify multiple different types of key points to obtain a set of key points corresponding to each type; for each set of key points corresponding to each type, three adjacent key points are selected in sequence for combination and pairing to obtain basic candidate units; based on the pairing results of the three adjacent points corresponding to all types of key points, a candidate key point combination containing the complete respiratory cycle is constructed.

[0070] In this embodiment of the application, the analysis module 703 is specifically used to obtain the signal features of the respiratory signal segment corresponding to each group of candidate key point combinations; compare the signal features with the preset signal features corresponding to the exhalation action after inhalation until the lungs are full; if the signal features match the preset signal features, then the respiratory signal segment corresponding to the corresponding candidate key point combination is taken as the target signal segment.

[0071] In this embodiment, the analysis module 703 is specifically used to determine the time range corresponding to the segment index in the respiratory signal; and to extract the corresponding target respiratory segment from the respiratory velocity waveform using the time range corresponding to the segment index.

[0072] In this embodiment of the application, the device further includes: a first calculation module, configured to extract the expiratory segment based on the target respiratory segment and calculate a first integral value of the first respiratory velocity signal within the expiratory segment and the baseline; calculate a second respiratory velocity signal within one second of the start of the expiratory segment and calculate a second integral value of the second respiratory velocity signal and the baseline; and calculate a one-second rate based on the first integral value and the second integral value.

[0073] In this embodiment of the application, the device further includes: a second calculation module, used to calculate the energy proportion of the respiratory velocity waveform of each region based on the chest and abdominal region signals corresponding to different distance gates in the test signal; determine the current body position of the test subject based on the energy proportion of the respiratory velocity waveform of each region; extract the steady segment signal before inspiration from the target respiratory segment according to the current body position, and calculate the mean of the steady segment as the initial baseline under the current body position; and correct the initial baseline using the baseline offset empirical value corresponding to the current body position to obtain the baseline.

[0074] Please see Figure 8 , Figure 8 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 8 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system).

[0075] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0076] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0077] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device as shown by a landing page for an app. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, which can be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0078] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0079] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0080] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0081] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for locating respiratory segments under voice command guidance, characterized in that, The method includes: The test signal is obtained by using radar to detect the test actions performed by the test object according to the voice command; Based on the test signal, the respiratory rate waveform and respiratory signal of the chest and abdomen of the tested object during the test process are extracted; The respiratory signal is analyzed to obtain the segment index representing the chest and abdomen after the subject inhales until the lungs are full and then exhales. The target respiratory segment is extracted from the respiratory rate waveform according to the segment index, and the respiratory index corresponding to the subject is calculated based on the target respiratory segment.

2. The method according to claim 1, characterized in that, The analysis of the respiratory signal yields an index of segments characterizing the chest and abdomen during the subject's inhalation until the lungs are full and then exhalation, including: The respiratory signal is averaged according to the distance gate dimension to obtain the displacement signal, and the displacement signal is then averaged according to the time dimension to obtain the smoothed respiratory signal; Detect multiple key points of different types in the smooth breathing signal; Candidate keypoint combinations are obtained by pairing multiple keypoints of different types. According to preset conditions, target signal segments representing the chest and abdomen are selected from the respiratory signals corresponding to the candidate key point combinations to characterize the exhalation of the lungs after the subject inhales until the lungs are full, and the segment index corresponding to the target signal segments is extracted.

3. The method according to claim 2, characterized in that, The process of pairing multiple key points of different types to obtain candidate key point combinations includes: Classify multiple different types of key points to obtain a set of key points corresponding to each type; For each type of key point set, three adjacent key points are selected sequentially and paired to obtain basic candidate units; Based on the corresponding adjacent three-point pairing results of all types of key points, candidate key point combinations containing the complete respiratory cycle are constructed.

4. The method according to claim 2, characterized in that, The step of selecting segments representing the chest and abdomen during exhalation after the subject inhales until the lungs are full, from the respiratory signals corresponding to the candidate key point combinations according to preset conditions, includes: Obtain the signal features of the respiratory signal segment corresponding to each group of candidate key point combinations; Compare the aforementioned signal characteristics with the preset signal characteristics corresponding to the exhalation action after inhalation until the lungs are full; If the signal feature matches the preset signal feature, then the respiratory signal segment corresponding to the corresponding candidate key point combination is taken as the target signal segment.

5. The method according to claim 2, characterized in that, Extracting the target respiratory segment from the respiratory rate waveform based on the segment index includes: Determine the time range corresponding to the segment index in the respiratory signal; The target respiratory segment is extracted from the respiratory rate waveform using the time range corresponding to the segment index.

6. The method according to claim 1, characterized in that, After extracting the target respiratory segment from the respiratory velocity waveform according to the segment index, the method further includes: The expiratory segment is extracted based on the target respiratory segment, and the first integral value of the first respiratory rate signal within the expiratory segment and the baseline is calculated. Calculate the second respiratory rate signal within one second of the start of the expiratory phase, and calculate the second integral value of the second respiratory rate signal and the baseline; The one-second rate is calculated based on the first integral value and the second integral value.

7. The method according to claim 6, characterized in that, The method further includes: Based on the chest and abdominal region signals corresponding to different distance gates in the test signal, calculate the energy percentage of the respiratory velocity waveform in each region. The current body position of the tested object is determined based on the energy proportion of the respiratory rate waveform in each region. Based on the current body position, extract the steady segment signal before inspiration from the target respiratory segment, and calculate the mean of the steady segment as the initial baseline under the current body position; The initial baseline is corrected using the empirical value of the baseline offset corresponding to the current body position to obtain the baseline.

8. A voice-command-guided respiratory segment localization device, characterized in that, The device includes: The testing module is used to detect the test actions performed by the test object according to voice commands using radar, and to obtain test signals; The extraction module is used to extract the respiratory rate waveform and respiratory signal of the chest and abdomen of the tested object during the test based on the test signal; The analysis module is used to analyze the respiratory signal to obtain the segment index representing the chest and abdomen after the subject inhales until the lungs are full and then exhales, and to extract the target respiratory segment from the respiratory rate waveform according to the segment index, and to calculate the respiratory index corresponding to the subject based on the target respiratory segment.

9. A computer device, characterized in that, include: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to perform the method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to perform the method of any one of claims 1 to 7.

Images