A PPG-based atrial fibrillation premature beat identification auxiliary system

Abstract

The present invention discloses a PPG-based atrial fibrillation and premature beat identification assistance system, comprising a light source, a sensor, and a computing center. The light source irradiates probe light onto human tissue, the sensor acquires PP data from the tissue and transmits it to the computing center, the computing center filters the PP data and draws a Poincare map, and the computing center performs logical operations on the PP data according to preset rules, outputting and displaying the output. The present invention provides a PPG-based atrial fibrillation and premature beat identification assistance system, based on the PPG principle. It utilizes a light source and a sensor to acquire light signals passing through human tissue. After preliminary processing and logical operations by the computing center, the system extracts graphical feature data from the Poincare map, outputs it, and displays it for medical personnel to identify and judge symptoms.

Description

Technical Field

[0001] The present invention relates to the technical field of PPG identification systems, and more specifically, to a PPG-based atrial fibrillation premature beat identification auxiliary system. Background Art

[0002] Heart movement can reveal valuable information about the human body, including health, lifestyle, emotional state, and the early onset of heart disease. The most important metric for monitoring heart movement is heart rate. Traditional medical equipment monitors heart rate and activity by measuring electrophysiological signals and an electrocardiogram (ECG). This typically involves electrodes connected to the chest cavity, triggering electrical activity and thereby obtaining signals about cardiac tissue movement. With technological advancements, photoplethysmography (PPG) has gradually become a popular method for monitoring cardiac activity. PPG is an optical technology that can obtain information about cardiac function without measuring bioelectrical signals. Instead, it monitors optical signals passing through human tissue to obtain information about that tissue's activity.

[0003] Compared to ECG technology, PPG offers significant advantages in data collection. ECG often uses chest leads, which is extremely inconvenient and technically demanding. Furthermore, it can only measure the surface of the chest cavity, which is quite restrictive. PPG, on the other hand, can collect data from any location on the body and can be used with wearable devices like smart wristbands and smartwatches, offering fewer limitations and greater ease of use. In terms of data performance, ECG peaks are due to ventricular contraction, while PPG peaks are caused by vasoconstriction, resulting in a strong correlation between the two.

[0004] PPG is suitable for heart rate monitoring because it is easy to collect data based on PPG and the gap between the collected data and ECG data is small. Therefore, there is a need for a PPG-based auxiliary system that can help medical staff identify atrial fibrillation and premature beats, and provide PP data to assist medical staff in making identification and judgment. Summary of the Invention

[0005] The present invention provides a PPG-based atrial fibrillation premature beat identification auxiliary system. Based on the PPG principle, it uses light sources and sensors to obtain light signals passing through human tissue, and extracts the graphic feature data of the Poincare map after preliminary processing and logical operations by a computing center. The data is output and displayed for medical personnel to identify and judge symptoms.

[0006] The technical solution of the present invention is as follows:

[0007] A PPG-based atrial fibrillation premature beat identification auxiliary system includes a light source, a sensor and a computing center. The light source irradiates detection light onto human tissue, the sensor obtains PP data from the human tissue and sends it to the computing center, the computing center filters the PP data and draws a Poincare map, and the computing center performs logical operations on the PP data according to preset rules and outputs it for display.

[0008] PP data, also known as photoplethysmography (PP) data, is collected during a single cardiac cycle. Specifically, PP data refers to the distance between the peak points of each pulse wave after acquiring a PPG signal. This distance is theoretically equivalent to the interval between two consecutive heartbeats.

[0009] The light source transmits probe light to human tissue through contact or non-contact means, causing the probe light to be transmitted or reflected by the tissue. The sensor receives the transmitted or reflected light from the body. The computing center performs logical operations on the PP data. The computing center receives the PP data from the sensor, performs logical operations based on preset rules, and outputs the results for medical personnel to make a decision.

[0010] In the above-mentioned PPG-based atrial fibrillation premature beat identification auxiliary system, before the computing center performs logical operations according to the preset rules, the PP data is quality screened and abnormal data is eliminated, so that the computing center obtains multiple sets of continuous, stable, and waveform-stable PP data.

[0011] This involves preliminary processing of the PP data acquired by the sensor, ensuring that the data used by the computing center for logical operations is composed of multiple sets of continuous, stable, and relatively average PP data. According to existing technologies, waveforms plotted from PP data acquired through PPG collection are simpler than those obtained through ECG (electrophysiological signals and electrocardiograms), with fewer small peaks and smoother waveforms. Peaks, caused by vasoconstriction, are clearly distinguishable. Therefore, quality screening and exclusion of abnormal data during this process are performed to obtain continuous, stable, and regular PP data, minimizing abnormal fluctuations in PP data caused by various interferences.

[0012] Furthermore, the received PP data is processed as follows:

[0013] Processing method 1: filtering, a filter is set between the sensor and the computing center, so that all PP data sent by the sensor are sent to the computing center through the filter;

[0014] Processing method 2: Determine whether the PP data is continuous and stable;

[0015] Processing method 3: Filter PP data with high waveform quality;

[0016] Processing method 4: Mark the PP data of the first PP interval after the waveform changes from irregular to regular and remove it;

[0017] Processing method 5: Compare the PP data of each group with the same phase data, and eliminate or repair the PP data with high deviation values;

[0018] A PPG-based atrial fibrillation premature beat identification auxiliary system uses any one or a combination of any of the above-mentioned processing methods 1 to 5 to process PP data.

[0019] The above-mentioned processing methods 2 to 5 are all completed in the computing center. The PP data received by the computing center are all saved as cache data. The PP data that are not filtered out or eliminated are marked left and right, and can still be restored or extracted in the cache data. Elimination only means that it is not included in the scope of logical operations.

[0020] Furthermore, in processing method 1, the PP data processed by the filter is further processed for marginal effects to eliminate PP data with unstable signals.

[0021] Processing method 1 is used to eliminate frequency anomalies and reduce data anomalies caused by external interference, short-term equipment problems, etc., to obtain a smooth signal after removing noise, which is conducive to feature recognition and screening.

[0022] Processing method 2 is used to deal with dynamic problems such as floating and fluctuations that are often caused by equipment. This type of PP data is marked and then eliminated.

[0023] Processing method 3 is to analyze the waveform image of the PP data. If there are irregular waveforms such as breakpoints, noise, inconsistent waveform shapes, etc. in the waveform, the corresponding PP data will be eliminated.

[0024] Processing method 4 removes the PP data in the buffer period of the change process, or marks the beginning of regular PP data.

[0025] Processing Method 5 is used to make the PP data used in logic operations more stable and average, better reflecting the normal state of human tissue. Because the PP data processed by the system is cached, signals with marked poor waveforms can be compared with signals with good waveforms for phase alignment to determine if waveform repair is possible. This ensures that every generated PP data point is usable in subsequent logic operations.

[0026] The data eliminated after the above processing method does not enter the logic operation process. The logic operation process is only for accurate and stable signals that have not been eliminated. The system processes these accurate signals and performs calculations.

[0027] The above-mentioned PPG-based atrial fibrillation premature beat identification auxiliary system sets the area with dense scattered points as a calculation area according to the degree of aggregation of scattered points in the Poincare map, obtains the coordinates of the four poles of the calculation area, determines the number of calculation areas in the Poincare map, and performs logical operations on Poincare maps with different numbers of calculation areas according to the preset rules.

[0028] The Poincaré plot reflects the distribution of the PP interval time series in phase space, encompassing both linear and nonlinear trends in heart rate variability, enabling rapid assessments. The Poincaré plot reflects the relationship between two adjacent PP intervals. Typically, the x-axis represents the value of the current PP interval, while the y-axis represents the value of the next PP interval, recording the ratio of the two. Generally, the length of the plot represents the overall heart rate variability, while the width represents the difference between adjacent PP intervals, representing the instantaneous heart rate change. Data on the line y = x in the plot represent the range of heart rate variation during the measurement period, while data above the line y = -x represent heart rate variability.

[0029] The computing center performs logical operations on the graphic data in the Poincaré map and outputs the output for medical staff to judge and identify. Its essence is to output the graphic features of the scattered points in the Poincaré map, including the number of scattered point dense areas and the corresponding shape features of the scattered point dense areas, and output the Poincaré map at the same time. In this way, medical staff can intuitively obtain more accurate data and understand the patient's physical condition more accurately, rather than judging or calculating values by themselves, thereby improving the efficiency of medical staff. It also makes it convenient for medical staff to conduct research and summarize rules so as to implement medical methods that are more suitable for patients and have better efficacy.

[0030] Furthermore, the computing center sets a set density. A certain area in the Poincare map has a scatter point density greater than or equal to the set density and is not adjacent to other areas with scatter point densities greater than or equal to the set density. This area is set as a computing area.

[0031] or,

[0032] Calculate the scatter point density of each unit area in the Poincare map, count all unit areas that meet the set density, merge all adjacent unit areas that meet the set density and record them as one calculation area.

[0033] Since setting the density requires that the dense area of scattered points, that is, the main body of scattered points in the Poincare map, be selected into the calculation area, the value of the set density should not be too large. It should include dense points as much as possible and eliminate relatively discrete scattered points.

[0034] Furthermore, the poles of the calculation area include a first pole located at the upper left of the calculation area, a second pole located at the lower left of the calculation area, a third pole located at the upper right of the calculation area, and a fourth pole located at the lower right of the calculation area. All scattered points in the calculation area are located inside the graph where the four poles are connected end to end.

[0035] Furthermore, when the number of calculation areas in the Poincare map is one, the computing hub calculates and outputs the following data:

[0036] a. First spacing: the spacing between the first pole and the second pole;

[0037] b. Second spacing: the spacing between the third pole and the fourth pole;

[0038] c. Fifth point: the intersection of the line connecting the first and second points and the line y=x;

[0039] d. Sixth point: the intersection of the line connecting the third and fourth points and the line y=x;

[0040] e. The third spacing: the spacing between the fifth pole and the sixth pole;

[0041] f. First angle: the angle between the line connecting the third and fifth poles and the line y=x;

[0042] g. Second angle: the angle between the line connecting the fourth and fifth poles and the line y=x;

[0043] h. The third angle: the sum of the first angle and the second angle.

[0044] The heart rate of patients with atrial fibrillation is characterized by irregular, high-frequency alternation of a long PP interval and a short PP interval. The Poincare plot often appears fan-shaped, that is, the scattered points are densely distributed, one end is wider, and the other end is thinner and aligned with the zero point. The amplitude of the heart rate change increases when the heart rate is slow.

[0045] Furthermore, when the number of calculation areas in the Poincare map is three, the calculation area close to the straight line y=x is the first calculation area, and the other two calculation areas are the second calculation area and the third calculation area. The calculation center calculates and outputs the following data:

[0046] a. Calculate the slope of the second calculation area and the third calculation area;

[0047] b. Determine whether the second calculation area and the third calculation area are similar in area;

[0048] c. Calculate the areas of the first, second, and third calculation regions;

[0049] d. Determine whether the second calculation area and the third calculation area are symmetrical about the line y=x.

[0050] The heart rate of paroxysmal atrial premature beats or paroxysmal ventricular premature beats is usually a long PP interval followed by a short PP interval. Its Poincare map has a large area of scattered points densely populated area similar to a "torpedo shape" along the straight line y=x, which is roughly symmetrical along the straight line y=x. On both sides of this scattered points dense area, there is a smaller area of scattered points dense area with the straight line y=x as the symmetry axis, and the areas and shapes of the two are similar.

[0051] Furthermore, when the number of computational regions in the Poincare map is four, the computational hub calculates and outputs the following data:

[0052] a. Calculate the areas of the first, second, third, and fourth calculation regions;

[0053] b. Calculate the length of each calculation area;

[0054] c. Calculate the width of each calculation area;

[0055] d. Calculate the slope of each calculation area.

[0056] The heart rate of frequent atrial premature beats and frequent ventricular premature beats is a long PP interval followed by a short PP interval. Its Poincare map is usually four smaller scattered point dense areas. The density of each scattered point dense area is significantly lower than the scattered point density of the above two cases. There is still a scattered point dense area along the straight line y=x, and the other three are less regular. The areas of the four scattered point dense areas are not much different.

[0057] Furthermore, in the above calculation process, the calculation center performs shape fitting processing on each calculation area, and the calculation center calculates the data of the fitting graphics and outputs it for display.

[0058] Because each calculation region is essentially a combination of scattered points, the calculation process is relatively complex. Fitting can be used to create a fitted graph to simplify the calculation process. There are no restrictions on the settings for fitting graphs; you can import a fitting model to build your own, or fit common shapes such as ellipses.

[0059] The present invention according to the above scheme has the following beneficial effects:

[0060] 1. Convenient data collection. Based on the characteristics of PPG, this system offers excellent convenience in data collection. Compared to ECG, it is not restricted by location and does not impose restrictions on patients' daily lives. Furthermore, in states such as sleep, where the signal is more stable and accurate, PPG can capture more realistic human information, whereas ECG may affect sleep quality and image data quality. Furthermore, the simple data collection method can also simplify data collection equipment, similar to devices such as smart bracelets and smart wristbands. Combined with cloud services, these smart devices can be integrated into people's daily lives, more effectively expanding their use cases and promoting public health and personal health management.

[0061] 2. Outputting intuitive graphic data can effectively assist medical staff in judging the patient's condition. Whether it is directly output or output after fitting, it can effectively reflect the patient's physical condition. All data have been quality screened and fitted. Compared with judgment based on initial data or simple graphics, the judgment basis provided to medical staff is less subjective and more objective. At the same time, it can also reduce the requirements of medical staff. Later improvements or improvements can be turned to how to improve the accuracy of data collection, improve data screening and fitting processing model induction, etc., that is, make this system a platform for output recognition and judgment, improve the calculation model within the platform, and even combine intelligent programs to perform intelligent preliminary diagnosis according to preset conditions and output graphic data, so as to provide more medical staff with more accurate identification and judgment of patients' conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0063] Figure 1 This is the Poincare map corresponding to the PP data when the number of calculation regions is one.

[0064] Figure 2 The Poincare map corresponding to the PP data when the number of calculation regions is three.

[0065] Figure 3 The Poincare map corresponding to the PP data is calculated when the number of regions is four. DETAILED DESCRIPTION

[0066] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0067] A PPG-based atrial fibrillation premature beat identification auxiliary system includes a light source, a sensor and a computing center. The light source irradiates detection light to human tissue, the sensor obtains PP data from the human tissue and sends it to the computing center, the computing center screens the PP data and draws a Poincare map, and the computing center performs logical operations on the PP data according to preset rules.

[0068] In this embodiment, the light source uses infrared light as its detection light. Compared to other wavelengths of light, infrared light has better muscle penetration and can detect deeper muscle tissue, making it easier for the sensor to capture it. Furthermore, infrared light can be emitted by LEDs, and the entire system can be powered by a low-voltage polymer lithium battery. Combined with a DC-DC boost circuit, this provides stable output and ensures stable detection light from the light source.

[0069] There are two types of sensors: contact and non-contact. Contact sensors are often loaded on wristband devices such as smart watches. These devices are also equipped with a light source, which is attached to the human skin. The sensor then receives the light transmitted or reflected by the human body. Non-contact sensors include high-definition cameras. The high-definition cameras shoot at parts of the human body with dense blood vessels and thin skin, such as the face, ears, and fingers, to capture changes in light intensity in these parts of the human body, thereby obtaining corresponding PP data.

[0070] 1. Obtain accurate PP data.

[0071] PP data is obtained by detecting in living tissues through photoelectric means. The principle is that different tissues of the human body absorb different amounts of detection light, resulting in regular changes in the light intensity of the detection light after transmission or reflection. In addition, the tissue will also undergo regular changes in the transmitted or reflected detection light under different physiological activities. Therefore, by irradiating certain tissues of the human body with detection light, and then receiving the changed detection light or monitoring the image of the human tissue during the process through the sensor, a light signal (light signal data) with a certain regularity can be obtained. The sensor converts the received light signal data and converts it into electrical signal data. The electrical signal data is filtered by a filter to remove cluttered signals with significantly different frequencies, forming a filtered electrical signal and then sent to the receiver. The receiver sends the filtered electrical signal to the computing center for logical calculation.

[0072] 2. Perform preliminary processing on PP data.

[0073] Within a certain period of time, the detection light is continuously irradiated on the human tissue, and the contact or non-contact sensor also continuously receives multiple optical signal data, and converts it into an electrical signal and sends it to the receiver, and the receiver sends it to the computing center. In the process of forming the optical signal, converting the optical signal into an electrical signal, transmitting the electrical signal, etc., there are various interferences, resulting in multiple quality problems in the PP data finally received by the computing center. For example, the optical signal received by the non-contact sensor is greatly affected by the ambient light, there is self-interference before the sensor stabilizes, and there is temporary interference in the signal transmission. Therefore, it is necessary to take measures to perform preliminary processing on the initial data to achieve preliminary screening of the data.

[0074] A filter is set between the sensor and the receiver to filter the electrical signal data. The light source remains unchanged, and the electrical signal data is in a band structure. The electrical signal after the filter still has boundary effects, so the computing center needs to select continuous and stable data segments for logical operations on the filtered data after the filter. Since the received filtered electrical signal is a continuous waveform signal, and a stable data segment waveform does not move at any point on its waveform line, there is no second intersection with the vertical line, and there are more than two waveforms of the same shape, it is necessary to determine whether the obtained filtered data meets this requirement. During the data acquisition process, the detection light of the light source is stable and unchanged. Human tissue should have a certain regularity when it is relatively static or in a regular state of movement. Therefore, the waveform generated by the computing center should be the same or similar, and even two or more exactly the same repeated patterns can be seen. To meet the above regularity requirements, the signal within the stable band is filtered out from the filtered electrical signal.

[0075] The waveform quality of the filtered electrical signal is judged and screened. For example, if there are breakpoints in the waveform, noise in the waveform, inconsistent waveform shape, etc., the PP data corresponding to the waveform points with such waveform quality problems are eliminated or marked and do not enter the logical operation process.

[0076] Multiple sets of samples are taken and compared with the filtered electrical signals of the same phase, screening for PP data with similar phases. After a period of continuous light output, the computing center obtains a set of PP data. After a certain interval, the light output is continued for the same period of time, and the computing center obtains another set of PP data. Repeat these steps to obtain multiple sets of PP data. Within these PP data sets, each data point with the same phase is compared. By eliminating or marking data that is too high or too low, or by using data with better waveforms to repair data with poor waveforms, more stable PP data is obtained.

[0077] The first PP interval after a long period of waveform quality problems does not enter the logical operation process. When the waveform is in an oscillating state, the waveform obviously does not have a stable factor and the waveform cannot show regularity. The corresponding PP data during this period is a problem of poor waveform quality, not only the logical operation process. When the waveform is in a stable state, after the waveform begins to show regularity, the calculation center looks at all the PP data. The PP data of the first PP interval when the waveform is in a regular state still does not enter the logical operation process, and marks this position. The first waveform in the waveform correction regression may have a large deviation in its corresponding PP data. In fact, this state should be in a changing state, and the data quality is questionable. To be on the safe side, the PP data of the first PP interval after a long period of unstable waveform quality does not enter the logical operation.

[0078] In this embodiment, the initial processing flow of PP data is as follows:

[0079] Step S1: caching PP data. This means that all data processing is performed after all data collection processes are completed.

[0080] Step S2: filtering, using a filter to remove the waveform of the noise signal and reduce the interference of irrelevant signals.

[0081] Step S3: Identify and store characteristic points, such as peak points, fluctuation points, and other waveform data with obvious characteristics.

[0082] Step S4: Waveform quality judgment.

[0083] Step S5. In-phase comparison, compare and judge with the feature points of the last processed instantaneous data, eliminate data with low waveform quality or completely inconsistent waveforms, or use the feature points of the last processed instantaneous data to repair after eliminating unqualified data.

[0084] Step S6: Save the current PP data.

[0085] Step S7: Save the current PP data feature points as reference values for the next processing of instantaneous data.

[0086] 2. Perform secondary processing on PP data.

[0087] After removing or marking PP data with quality issues, the computational center now obtains multiple sets of high-quality PP data. Based on these PP data, the computational center creates a Poincare map. Using the current PP interval as the X-axis coordinate and the next RR interval as the Y-axis coordinate, the Poincare map is generated, representing the heart rate of the tissue being examined.

[0088] After completing the Poincaré map, first, the heat algorithm is used to remove abnormal discrete points on the Poincaré map and remove obviously unstable data points.

[0089] Next, based on the degree of clustering of scattered points, areas with densely distributed scattered points are designated as calculation areas, and the number of calculation areas is determined. Visually, the degree of clustering of scattered points within the Poincaré map clearly indicates that the map can be divided into several calculation areas. For example, the Poincaré map for healthy heart rate has a clear cluster of scattered points, allowing accurate determination that the Poincaré map has only one calculation area. The calculation center then compares the number of scattered points per unit area based on a pre-set number of scattered points. If the scattered point density of an area meets the set density and is not adjacent to other areas with the set density, that is, if there are no surrounding areas with the set density, then the area is considered a calculation area. Alternatively, after subdividing the Poincaré map, all unit areas that meet the set density are counted. Adjacent unit areas with a scattered point density greater than the set density are considered one calculation area, while non-adjacent unit areas are considered another. This process continues outward, forming a single calculation area, thereby determining the total number of calculation areas in the Poincaré map. After determining the number of calculation areas, the logical operation process varies depending on the number of calculation areas.

[0090] Finally, determine the basic value of the logical operation. After dividing the calculation area, for each calculation area, determine the upper left point - the first pole, the lower left point - the second pole, the upper right point - the third pole, and the lower right point - the fourth pole of the calculation area, which are respectively the upper left boundary point, the lower left boundary point, the upper right boundary point, and the lower right boundary point in the calculation area. That is, all scattered points in the calculation area are located inside the graph where these four points are connected end to end.

[0091] 3. Perform logical operations on PP data.

[0092] (1) When the number of calculation areas is one:

[0093] like Figure 1 As shown, at this time, the scattered points of the Poincare map form an image that is approximately fan-shaped, and the figure is basically symmetrically located on both sides of the straight line y=x.

[0094] The computing center calculates:

[0095] a. The distance between the first pole and the second pole - the first distance;

[0096] b. The distance between the third pole and the fourth pole - the second distance;

[0097] c. The intersection of the line connecting the first and second poles and the line y=x - the fifth pole;

[0098] d. The intersection of the line connecting the third and fourth poles and the line y=x is the sixth pole;

[0099] e. The distance between the fifth pole and the sixth pole - the third distance;

[0100] f. The angle between the line connecting the third and fifth poles and the line y=x - the first angle;

[0101] g. The angle between the line connecting the fourth and fifth poles and the line y=x - the second angle;

[0102] h. The sum of the first angle and the second angle - the third angle.

[0103] The calculation center outputs the above calculated values so that medical personnel can make logical judgments on atrial fibrillation identification based on the above calculated values.

[0104] (2) When the number of calculation areas is three:

[0105] like Figure 2 As shown, the scattered points of the Poincare map at this time constitute an image containing three areas where scattered points gather. One with a larger area is set as the first calculation area, and two with smaller areas are set as the second calculation area and the third calculation area respectively. The first calculation area is roughly symmetrically distributed along the straight line y=x, and the second calculation area and the third calculation area are respectively set on both sides of the first calculation area with the straight line y=x as the symmetry axis.

[0106] At this time, the calculation center:

[0107] a. Calculate the slopes of the second and third calculation areas.

[0108] Regarding the slope of the calculation area, that is, the average value of the slopes of the lines connecting the four poles of each calculation area and the origin is obtained, the ratio of the line formed by the first pole and the origin to the x-axis, the ratio of the line formed by the second pole and the origin to the x-axis, the ratio of the line formed by the third pole and the origin to the x-axis, and the ratio of the line formed by the fourth pole and the origin to the x-axis, and the average value of the four slopes is obtained as the slope of the calculation area.

[0109] In another embodiment, the slope of the major axis of the fitted ellipse of each calculation area is used as the slope of the calculation area. Through fitting, a minimum ellipse is made to cover as much of the calculation area as possible, and the ratio of the major axis of the ellipse to the x-axis is calculated as the slope of the major axis, that is, the slope of the calculation area.

[0110] b. Determine whether the second calculation area and the third calculation area are similar in area.

[0111] Similarly, due to the actual situation such as interference in data collection and fluctuation of physiological activities of human tissue, it is not realistic to have the areas of the second calculation area and the third calculation area be completely equal. Therefore, after obtaining the four extreme points, the enclosed areas of the four extreme points are calculated respectively and compared. When the ratio or difference of the two areas is close to the set area similarity value, it can be considered that the areas of the second calculation area and the third calculation area are similar.

[0112] c. Calculate the areas of the first calculation area, the second calculation area, and the third calculation area.

[0113] d. Determine whether the second calculation area and the third calculation area are symmetrical about the line y=x.

[0114] Due to interference in data collection and fluctuations in human physiological activity, it's unlikely that the second and third calculation regions will be completely symmetrical. Therefore, a virtual calculation region symmetrical to the first or second calculation region is established to determine the degree of overlap between the second or first calculation region and the virtual calculation region. When the overlap reaches a certain set value, the second and third calculation regions are considered symmetrical about the line y = x. Symmetry between the second and third calculation regions can also be determined using the slope calculated in step a.

[0115] The calculation center outputs the above calculated values so that medical personnel can make logical judgments on the identification of paroxysmal atrial premature beats or paroxysmal ventricular premature beats based on the above calculated values.

[0116] (3) When the number of calculation areas is four:

[0117] like Figure 3 As shown, there are four calculation areas on the Poincare map, namely the first calculation area, the second calculation area, the third calculation area and the fourth calculation area. Compared with the above two cases, the four calculation areas in this embodiment have a lower density of scattered points and a relatively smaller area, but the first calculation area still exists on the straight line y=x. The first calculation area is roughly axisymmetric with respect to the straight line y=x, the second calculation area exists above the straight line y=x, and the third and fourth calculation areas exist below the straight line y=x.

[0118] The computing center calculates:

[0119] a. Calculate the areas of the first, second, third, and fourth calculation regions. The area of each calculation region is considered to be the area enclosed by the four extreme points of each calculation region.

[0120] b. Calculate the length of each calculation area. Perform ellipse fitting on each calculation area so that a minimum ellipse covers as much of the calculation area as possible. Calculate the length of the major axis of the ellipse as the length of the calculation area.

[0121] c. Calculate the width of each calculation area. Perform ellipse fitting on each calculation area so that a minimum ellipse covers as much of the calculation area as possible. Calculate the length of the minor axis of the ellipse as the width of the calculation area.

[0122] d. Calculate the slope of each calculation area.

[0123] Calculate the average of the slopes of the lines connecting the four poles of each calculation area with the origin, the ratio of the line formed by the first pole and the origin to the x-axis, the ratio of the line formed by the second pole and the origin to the x-axis, the ratio of the line formed by the third pole and the origin to the x-axis, and the ratio of the line formed by the fourth pole and the origin to the x-axis, and obtain the average of the four slopes as the slope of the calculation area.

[0124] In another embodiment, the slope of the major axis of the fitted ellipse of each calculation area is used as the slope of the calculation area. Through fitting, a minimum ellipse is made to cover as much of the calculation area as possible, and the ratio of the major axis of the ellipse to the x-axis is calculated as the slope of the major axis, that is, the slope of the calculation area.

[0125] The calculation center outputs the above-mentioned calculated values so that medical personnel can make logical judgments on the identification of frequent atrial premature beats and frequent ventricular premature beats based on the above-mentioned calculated values.

[0126] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A PPG-based atrial fibrillation premature beat identification auxiliary system, characterized in that: The system comprises a light source, a sensor, and a computing hub. The light source irradiates detection light onto human tissue. The sensor acquires PP data from the human tissue and sends it to the computing hub. The computing hub screens the PP data and draws a Poincare map. The computing hub performs logical operations on the PP data according to preset rules and outputs the data for display. According to the degree of clustering of scattered points in the Poincare map, the area with dense scattered point distribution is set as a calculation area, the coordinates of the four poles of the calculation area are obtained, the number of calculation areas in the Poincare map is determined, and logical operations are performed on Poincare maps with different numbers of calculation areas according to the preset rules: The calculation center sets a set density. A certain area in the Poincare map has a scatter point density greater than or equal to the set density and is not adjacent to other areas with scatter point density greater than or equal to the set density. This area is set as a calculation area. or, Calculate the scattered point density of each unit area in the Poincare map, count all unit areas that meet the set density, merge all adjacent unit areas that meet the set density and record them as one calculation area; The poles of the calculation region include the first pole located at the upper left of the calculation region, the second pole located at the lower left of the calculation region, the third pole located at the upper right of the calculation region, and the fourth pole located at the lower right of the calculation region. All scattered points in the calculation region are located inside the graph where the four poles are connected end to end. When the number of calculation areas in the Poincare map is one, the computing hub calculates and outputs the following data: a. First spacing: the spacing between the first pole and the second pole; b. Second spacing: the spacing between the third pole and the fourth pole; c. Fifth point: the intersection of the line connecting the first and second points and the line y=x; d. Sixth point: the intersection of the line connecting the third and fourth points and the line y=x; e. The third spacing: the spacing between the fifth pole and the sixth pole; f. First angle: the angle between the line connecting the third and fifth poles and the line y=x; g. Second angle: the angle between the line connecting the fourth and fifth poles and the line y=x; h. The third angle: the sum of the first angle and the second angle; When the number of calculation areas in the Poincare map is three, the calculation area close to the straight line y=x is the first calculation area, and the other two calculation areas are the second calculation area and the third calculation area. The calculation center calculates and outputs the following data: a. Calculate the slope of the second calculation area and the third calculation area; b. Determine whether the second calculation area and the third calculation area are similar in area; c. Calculate the areas of the first, second, and third calculation regions; d. Determine whether the second calculation area and the third calculation area are symmetrical about the line y=x; When the number of computational regions in the Poincare map is four, the computational hub calculates and outputs the following data: a. Calculate the areas of the first, second, third, and fourth calculation regions; b. Calculate the length of each calculation area; c. Calculate the width of each calculation area; d. Calculate the slope of each calculation area.

2. A PPG-based atrial fibrillation premature beat identification auxiliary system according to claim 1, characterized in that: Before the computing center performs logical operations according to the preset rules, the PP data is quality screened and abnormal data is eliminated, so that the computing center obtains multiple sets of continuous, stable, and waveform-stable PP data.

3. A PPG-based atrial fibrillation premature beat identification auxiliary system according to claim 2, characterized in that: How to process received PP data: Processing method 1: filtering, a filter is set between the sensor and the computing center, so that all PP data sent by the sensor are sent to the computing center through the filter; Processing method 2: Determine whether the PP data is continuous and stable; Processing method 3: Filter PP data with high waveform quality; Processing method 4: Mark the PP data of the first PP interval after the waveform changes from irregular to regular and remove it; Processing method 5: Compare the PP data of each group with the same phase data, and eliminate or repair the PP data with high deviation values; The PPG-based atrial fibrillation premature beat identification auxiliary system uses any one or a combination of any of the above-mentioned processing methods 1 to 5 to process the PP data.

4. The PPG-based atrial fibrillation premature beat identification auxiliary system according to claim 1, characterized in that: The computing center performs shape fitting processing on each computing area, and the computing center calculates data of the fitting graphics and outputs it for display.

Images