CN106037720A - Application method of hybrid continuous information analysis technology in medicine - Google Patents

Abstract

The invention discloses a medical application method of a hybrid continuous information analysis technology. The electrocardiogram signal acquisition module in the electrocardiogram hybrid continuous analysis system is used to obtain the electrocardiogram signal from the electrocardiogram device, and then the electrocardiogram signal is pre-predicted by the electrocardiogram detection module. Processing, the processed signal extracts the signal characteristics through the electrocardiogram detection method and makes each basic heart rate event for analysis by the compound event processing heartbeat recognition module. The compound event processing heartbeat recognition module converts the basic heart rate The event is compared with the information in the patient history database module to identify the abnormal heart rate event, and when there is a serious abnormal heart rate event, the abnormal heartbeat alarm and pre-diagnosis information are transmitted to the pre-diagnosis output module, and the pre-diagnosis output module sends the patient's physician and/or Or family members send alerts and distress messages.

Description

Application method of hybrid continuous information analysis technology in medicine

technical field

The invention relates to an application method of a hybrid continuous information analysis technology, in particular to a medical application method of a hybrid continuous information analysis technology.

Background technique

The electrocardiogram (electrocardiograms, ECG) is the bioelectrical change accompanied by the heart beat in each cardiac cycle. The graph of various forms of potential changes is drawn from the body surface by the electrocardiography device, which gives the heart of each person. It can help analyze the abnormal heartbeat in the ECG signal. Periodic P, QRS, and T wave sequences can be observed in the ECG. In this sequence, the QRS wave group has the largest amplitude, and its detection helps to calculate its surrounding P, T waves and other characteristics of the heartbeat of.

Cardiovascular disease has always been one of the main causes of harm to human health and death, and the identification and analysis of ECG is of great clinical significance. At present, the clinical application of ECG in my country is mainly divided into three forms: one is that the patient receives a short-term test of the electrocardiogram, such as physical examination and ordinary heart detection, usually the scan of the patient's heartbeat does not exceed one minute, and the patient's ECG signal will be It is printed directly on paper, analyzed by the doctor, and then informs the patient of any abnormalities in the ECG. The second is that for hospitalized patients with life-threatening or heart disease, an ECG monitor will be provided to detect the patient's heartbeat, blood pressure, blood oxygen level and other data. The purpose of this method is to detect when the patient is in a life-threatening situation, the ECG monitor will give an alarm, such as the patient tremor, jump stop, etc., most of the patient's ECG information is not recorded. This makes it difficult for the doctor to obtain a good grasp of the patient's ECG situation for a period of time. The exception is for patients in the intensive care unit (ICU). The third is to conduct 24-hour ECG monitoring for patients after cardiac surgery or patients with severe heart disease, commonly known as "backpack". This is to connect a portable ECG recorder with leads connected to the human body, allowing the patient to move on their own to record the patient's ECG data in the past 24 hours. These data are then taken out from the recorder and imported into a computer for analysis.

For the above three forms, there are three common problems to face: First, the patient’s medium and long-term ECG data are not archived, which means that the patient’s recovery or condition cannot be analyzed through the analysis of historical ECG degree of deterioration. Secondly, for the analysis of the electrocardiogram, after the patient gets the electrocardiogram recorded on the paper, he has to find a professional doctor to analyze it, which is not real-time, and there is nothing he can do to save a sudden heart attack. Third, only critically ill patients can obtain professional medical help. For a large number of elderly people who generally suffer from hypertension and coronary heart disease, they have no way to understand their daily physical conditions. And due to individual differences in the human body, each person's normal heartbeat is also different. We can find that the normal heartbeat of the human body is 60-100 beats per minute through Internet searches, so there is a problem with this value. For example, if a person's normal heartbeat is 60 beats per minute, if his heartbeat suddenly becomes 100 beats per minute , although the value is still in the range of normal heartbeat, it is already in tachycardia for him. In addition, some people's normal heartbeat itself is high or low, some people's normal heartbeat is above 110 beats per minute, and some people's normal heartbeat is below 60 beats per minute. In judging, mistakes are bound to happen.

Therefore, a system is designed to detect the patient's heartbeat in a timely manner in combination with the patient's own heartbeat. When the patient has an abnormal heartbeat, the abnormal heartbeat can be displayed, and the performance of the abnormal heartbeat can be compared with the patient's medical records and the patient's past heartbeat history. It is an urgent problem to be solved to compare the data, give an alarm for serious abnormal heartbeats, give auxiliary pre-diagnosis for each patient, and help doctors better understand the patient's condition.

Contents of the invention

The purpose of the present invention is to provide a medical application method of mixed continuous analysis technology. This method can combine the patient's own heartbeat to detect the patient's heartbeat in a timely manner. When the patient has an abnormal heartbeat, the abnormal heartbeat can be displayed, and the performance of the abnormal heartbeat can be compared with the patient's medical records and the patient's past heartbeat historical data. Comparing, alarming for severe abnormal heartbeat, giving auxiliary pre-diagnosis for each patient, helping doctors to better understand the patient's condition.

The technical scheme of the present invention: a method for applying the hybrid continuous information analysis technology in medicine. The electrocardiogram signal acquisition module in the ECG hybrid continuous analysis system obtains the electrocardiogram signal from the electrocardiogram device, and then the electrocardiogram signal is detected by the electrocardiogram detection module. Carry out preprocessing, the processed signal extracts the signal characteristics through the electrocardiogram detection method and makes each basic heart rate event for analysis by the compound event processing heartbeat recognition module. The basic heart rate event is compared with the information in the patient history database module to identify the abnormal heart rate event, and when there is a serious abnormal heart rate event, the abnormal heartbeat alarm and pre-diagnosis information are transmitted to the pre-diagnosis output module, and the pre-diagnosis output module sends the patient's physician and/or family members to send alerts and distress messages.

In the medical application method of the aforementioned hybrid continuous information analysis technology, the electrocardiogram signal acquisition module acquires signals from an electrocardiogram monitor or a portable electrocardiograph, and the signals are transmitted to a mobile phone or a PC through Bluetooth.

In the medical application method of the aforementioned hybrid continuous information analysis technology, the electrocardiogram detection module identifies each heartbeat from the voltage signal sequence transmitted by the electrocardiogram signal acquisition module, and extracts the heartbeat heart rate, RR wave interval, P Wave, QRS wave group and T wave characteristic parameters, and make these characteristic parameters into basic heart rate events and transmit them to the compound event processing heartbeat identification module for matching use of abnormal heart rate events; at the same time, store the ECG data in the historical database module .

In the medical application method of the aforementioned hybrid continuous information analysis technology, the electrocardiogram detection module includes a signal preprocessing module and an electrocardiogram detection module; the electrocardiogram detection module includes a QRS wave group detection module and a P, T wave detection module; Among them, the signal preprocessing module firstly uses wavelet transform to remove the power frequency interference from the voltage signal transmitted by the electrocardiogram signal acquisition module and adjusts the baseline with polynomial fitting to solve the problem of baseline drift; the QRS wave group detection module uses a dynamic threshold detection method to detect The data transmitted in real time is operated by point-by-point variance, point-by-point square, and point-by-point square of signal amplitude, and the integral of the moving window is performed at the R peak; the P and T wave detection modules are used to analyze the window period of a complete heartbeat from Find the position and information of the P and T waves corresponding to the heartbeat from the signals before and after the QRS complex.

In the medical application method of the aforementioned hybrid continuous information analysis technology, the dynamic threshold detection method includes the following steps:

A. Establish a window with a size of 1000 sample points. With the new signal input, the old sample points are moved out of the window;

B. The preprocessed signal is subjected to sliding average processing, and the sliding window of 11 sample points is averaged;

C. Use the dynamic threshold THR1 equal to the sum of the average value of the sample points in the sliding window and the sample average value to filter out the signal part below the threshold;

D. Use the dynamic threshold THR2 equal to the difference between the average value of the sample points in the sliding window and the sample average value to filter out the part of the signal above the threshold;

E. Compare the interval between the non-zero intervals generated in step C and step D. When the interval between non-zero intervals is less than 50 sample points, set the next adjacent non-zero interval to zero, and combine the results of step C and step D ;

F, the part after the processing of step E is all in the QRS wave group, find the most value point in the wave peak as the point of the R peak;

G. After the processing in step D, if the non-zero interval does not fall back at the edge of the sliding window, it is considered that the QRS complex has not reached the highest point, that is, there is an error in the latest R peak value found in step F, and the latest R peak value analyzed in step G an R peak marker;

H, if the non-zero interval processed in step D has fallen back at the edge of the window, then perform step F again to compare the new result with the R peak position found in the sliding window to determine the latest R peak position;

J, the two sizes of the interval edges that are not zero after the calculation step E are processed are the intervals of the R wave as the sample points where the sign of the second derivative is changed in the 20 sample intervals;

K. Through the two RR wave intervals before and after, calculate the interval of the complete heartbeat in the middle, and analyze the windows before and after the QRS complex, and find out the interval that meets the sub-threshold THR3 as the interval of P and T waves for analysis;

L, save the QRS wave group, P wave, T wave interval time, peak value, start and end sample point position information in each heartbeat interval and send it to the compound event processing heartbeat recognition module.

In the medical application method of the aforementioned hybrid continuous information analysis technology, the compound event processing heartbeat recognition module monitors each basic heart rate event output by the electrocardiogram detection module through the hybrid continuous analysis module, and processes the heartbeat recognition through the compound event The EPL statement in the module identifies whether each basic heart rate event meets the conditions of a normal heartbeat. For an abnormal heart rate event, it is judged which abnormal heart rate event it is by matching with the abnormal heart rate meter in the patient history database module, and according to the abnormal heart rate event Position in the event tree or event graph to determine whether this abnormal heart rate event can form other more complex abnormal heart rate events, and determine whether to remove the corresponding pattern matching statement, and finally send the resulting new complex event to the complex The event processing heartbeat identification module; and all the basic heart rate events of each heartbeat are stored in the historical database module as historical data.

In the medical application method of the aforementioned mixed continuous information analysis technology, the historical database module includes a patient's basic information table, a stored basic heart rate table, an abnormal heart rate table, an abnormal heart rate sub-table, and a heart rate variability analysis value table; Each table is divided into two storage parts, day and night, according to time.

In the medical application method of the aforementioned hybrid continuous information analysis technology, the hybrid continuous analysis module records the mid- and long-term indicators of the patient's heart rate, and uses these indicators to adjust the screening of cardiac physiological events caused by individual differences, and Applying it to complex heart rate event processing enables the matching mode used in complex heart rate event processing to be dynamically changed and adjusted.

In the medical application method of the aforementioned mixed continuous information analysis technology, the realization of the mixed continuous analysis module includes three parts: mixed basic events, mixed complex events and comprehensive pre-diagnosis; The analysis and statistics of each person's normal heart rate of the P wave and T wave relative to the distribution of the QRS complex, and mining or analysis of all the past single heart rate to obtain the historical characteristic value of the heart rhythm, both using mixed continuous analysis The method is compared to analyze the basic heart rate type attribute; the mixed complex event extracts the different abnormal heart rate events that have occurred in the patient history database module during the monitoring process of the abnormal heart rate event by the compound event processing heartbeat recognition module eigenvalues, and use these eigenvalues for real-time detection of abnormal heart rate events in patients; Symptom-related information is sent to healthcare professionals.

Beneficial effects of the present invention: Compared with the prior art, the application method of the present invention combines the patient's own heartbeat to detect the patient's heartbeat in a timely manner. When the patient has an abnormal heartbeat, it can display the abnormal heartbeat and display the abnormal heartbeat The performance of the heartbeat is compared with the patient's medical records and the patient's past heartbeat historical data, and an alarm is given for serious abnormal heartbeats, and auxiliary pre-diagnosis for each patient is given to help doctors better understand the patient's condition. The invention combines people's actual needs, designs an electrocardiographic hybrid continuous analysis system and related processing methods, and proposes an electrocardiographic detection method and a complex event processing module. An electrocardiogram detection method is proposed to detect the basic heart rate of the electrocardiogram and convert it into a basic event flow. The Esper stream event processing engine is used to build a continuous analysis system to perform pattern matching on the basic event flow generated by the electrocardiogram detection method. Store the real-time ECG data and abnormal heart rate data generated by the system into the historical database. At the same time, the EPL statement for common abnormal heart rhythms is designed, and the relationship between a large number of complex events is connected in series in the form of a complex event tree for dynamic pattern matching, which reduces the pressure on the system. Finally, the historical ECG data and characteristics are extracted from the database and provided to the event flow analysis engine for more accurate and personalized pattern matching. After matching the preset abnormal heart rhythm event, it will give a timely warning, and quickly provide relevant information such as the patient's relevant history records and medication status, so as to improve the treatment rate of medical staff. In summary, it mainly includes the following advantages:

1. For the special needs of ECG analysis, a real-time ECG detection method is proposed, which can better provide real-time event support for ECG monitoring and analysis.

2. Design the ECG hybrid continuous analysis system based on complex event processing, and introduce complex event processing analysis technology into ECG monitoring and analysis.

3. A hybrid continuous analysis architecture is designed. The traditional continuous analysis technology can only set the threshold in advance. If the threshold is exceeded, an alarm will be issued. Compared with each person's physique and heartbeat characteristics are different, the hybrid continuous analysis technology can be compared to a certain extent. Ok fix that.

4. The ECG hybrid continuous analysis system adopts complex event processing technology to change data-driven to event-driven mode, which saves computing power and saves a lot of raw data storage space and only stores relatively small event data.

Description of drawings

Accompanying drawing 1 is the structural representation of electrocardiogram mixing continuous analysis system of the present invention;

Attached Figure 2 is a hierarchical schematic diagram of complex events;

Accompanying drawing 3 is the flowchart of composite event processing heartbeat recognition module;

Accompanying drawing 4 is the mixed monitoring flowchart of heartbeat rate;

Accompanying drawing 5 is the event tree schematic diagram of ventricular tachycardia event;

Accompanying drawing 6 is the schematic diagram of ventricular bigeminy;

Accompanying drawing 7 is the schematic diagram of ventricular trigeminy;

Accompanying drawing 8 is the error rate schematic diagram of MIT ECG database;

Accompanying drawing 9 is the schematic diagram of the identification of abnormal cardiac rhythm by complex event technology;

Accompanying drawing 10 is the schematic diagram of the abnormal heart rate recognition rate of hybrid continuous analysis technique;

detailed description

The present invention will be further described below in conjunction with the accompanying drawings and embodiments, but not as a basis for limiting the present invention.

Embodiment of the present invention: The electrocardiogram hybrid continuous analysis system researched and designed by the present invention analyzes and saves a large amount of biological sign data generated by the patient at every moment through the CEP (composite event processing) technology. The research and design of the ECG hybrid continuous analysis system can solve the problem that a large amount of biological sign data is directly discarded, and the analysis of these data can be used to save lives and improve physical conditions.

The electrocardiogram (electrocardiograms, ECG) is the bioelectrical change accompanied by the heart beat in each cardiac cycle. The graph of various forms of potential changes is drawn from the body surface by the electrocardiography device, which gives the heart of each person. It can help analyze the abnormal heartbeat in the ECG signal. Periodic P, QRS, and T wave sequences can be observed in the ECG. In this sequence, the QRS wave group has the largest amplitude, and its detection helps to calculate its surrounding P, T waves and other characteristics of the heartbeat of. The main purpose of the present invention is the research and attempt on the combination of electrocardiogram analysis and hybrid continuous analysis techniques. The framework design and module design of the system for hybrid continuous analysis of medical ECG will be introduced below.

Frame Design of ECG Hybrid Continuous Analysis System

The ECG hybrid continuous analysis system needs to obtain ECG signals from various ECG devices. For these signals, CEP (composite event processing) technology cannot directly process and obtain information, and there may be factors such as signal sources that affect the quality of judgment. , so the first step is to perform a series of preprocessing on the ECG signal so that the quality of the signal meets the requirements of the detection method. Only then can the ECG detection method be used to extract the signal characteristics of the ECG for analysis by the CEP system. The CEP system finds abnormal heartbeats after analyzing the sent ECG data, and gives an alarm to serious abnormal heartbeats. And compare the performance of abnormal heartbeat with the patient's medical records and the patient's past heartbeat history data, and give an auxiliary pre-diagnosis for each patient, helping doctors better understand the patient's condition.

Based on the above analysis, the CEP-based ECG real-time monitoring and analysis system can effectively process ECG signals, identify abnormal heartbeat events, compare heartbeat events with historical information such as patient medical records, and provide pre-diagnosis information. Provide support to help ordinary patients understand their physical conditions and make timely decisions about seeking medical treatment. The framework of ECG hybrid continuous analysis system is shown in Figure 1. The system is divided into five modules, which are electrocardiogram signal acquisition module, electrocardiogram detection module, CEP heartbeat recognition module, historical database module, and pre-diagnosis output module. The following will introduce the main functional modules and related technologies of the system respectively.

Design of electrocardiogram signal acquisition module

The acquisition of ECG signals is done by ECG monitors, portable ECG instruments, etc. These signals are transmitted to mobile phones or PCs through Bluetooth, and ECG detection is performed on the mobile phone or PC side. In order to save bandwidth, the 360HZ ECG signals are not real-time After being transmitted to the server, the program on the mobile phone or PC detects and compresses the characteristic information of each heartbeat, and then uploads the ECG characteristic signal of this cycle to the server for processing every minute or longer; if If an abnormal heartbeat is found in the detection of the terminal, the ECG compressed data in the current period will be immediately transmitted to the server for analysis. The information received by the serial port is used as the data source for the next signal preprocessing through the monitoring program on the PC side. The monitoring program can record the frequency of the signal source and the strength value of each signal for use by the preprocessing module.

The ECG database generally stores the ECG data of one or more leads, and the contained signal sources need to be decoded according to different databases. Generally, the ECG database will contain header files and diagnostic files. The header files generally give the ECG data. Relevant information, such as the signal frequency used by the ECG, the position of the signal base point, the unit, the number and location of the leads, and some relevant information of the patient: age, gender, medication, etc. The diagnosis file will indicate the type and occurrence time of the abnormal heartbeat, and mark the signal points and related values of the specific abnormal heartbeat. Reading annotation information will greatly help developers verify their ECG monitoring algorithms or programs.

Design of Electrocardiogram Detection Module

The function of the electrocardiogram detection module is to identify each heartbeat from the voltage signal sequence of the heart, and extract some characteristic parameters such as heart rate, RR wave interval, P wave, QRS wave group, and T wave. These characteristic parameters constitute basic heart rhythm events and are transmitted to the complex event processing heartbeat recognition module for matching of abnormal heart rhythm events. The ECG detection module is mainly composed of two parts: a signal preprocessing module and an ECG detection module. The ECG detection module includes two parts: QRS wave group detection and identification and P, T wave detection and identification. The function and design of these three parts will be described respectively below.

Signal preprocessing module design: The quality of ECG recording has an important impact on the recognition rate of ECG signals. Usually, the actual ECG signals have various interferences and drifts, so it is very important to preprocess the signals. necessary behavior. ECG signal has two important sources of interference, one is the power frequency interference of 50Hz/60Hz and its harmonics; the other is some baseline drift less than 1Hz. Firstly, wavelet transform is used to remove power frequency interference, and polynomial fitting is used to adjust the baseline for the problem of baseline drift.

ECG detection module design: it includes signal QRS wave group detection, because the analysis of real-time data cannot extract the characteristics of the complete heart rate record, so a dynamic threshold detection method is used to point-by-point the real-time transmitted data Operations such as variance, point-by-point square, and point-by-point square of signal amplitude make the output data positive, and nonlinearly amplify the differential output signal, highlighting the high-frequency part of the signal, highlighting the R peak more, and reducing The false positive caused by the T wave is eliminated; the integration of the moving window is performed at the R peak to extract other information of the R wave, such as slope, width, and improve the accuracy of QRS complex detection. It also includes P and T wave detection. Through the detection of QRS wave group and the calculation of RR wave interval, it is easy to roughly calculate the window period of a complete heartbeat. Analyze this window period to find the possible positions and other information of the P and T waves corresponding to the heartbeat from the signals before and after the QRS complex; but not all P and T signals can be identified in this way, such as cardiac Acceleration, tremor, and other situations where P and T waves overlap with other waves are difficult to identify, but usually P and T waves that cannot be identified can also become a sign, and can also be used to judge abnormal ECG .

In view of the need to compress the basic information details of the heartbeat from the voltage values of hundreds of sample points to the basic data information of each heartbeat in this system, the main description of the dynamic threshold detection method is as follows.

Step1. Establish a window with a size of 1000 sample points, and as new signals are input, old sample points are moved out of the window.

Step2. The preprocessed signal is subjected to sliding average processing, and the sliding window of 11 sample points is averaged. The acquisition of the moving average can effectively reduce the error of the dynamic threshold and filter out the signal part below its average value.

$\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 11</span>}}\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 5</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 5</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span>$

n=1,2....1000

Step3. The dynamic threshold THR1 is equal to the sum of the average value of the sample points in the sliding window and the sample average value to filter out the signal part below the threshold.

$\mathrm{<span\; class="notranslate">\; T</span>}\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left[\begin{array}{cc}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>& \mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ></span>\mathrm{<span\; class="notranslate">\; T</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; R</span>}\mathrm{<span\; class="notranslate">\; 1</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span>& \mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>\end{array}\right]$

n=1,2....1000

Step4. The dynamic threshold THR2 is equal to the difference between the average value of the sample points in the sliding window and the sample average value to filter out the part of the signal higher than the threshold value. The location of the R wave can be located.

$\mathrm{<span\; class="notranslate">\; T</span>}\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left[\begin{array}{cc}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>& \mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; T</span>}\mathrm{<span\; class="notranslate">\; h</span>}\mathrm{<span\; class="notranslate">\; R</span>}\mathrm{<span\; class="notranslate">\; 2</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span>& \mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>\end{array}\right]$

n=1,2....1000

Step5. Compare the interval between the non-zero intervals generated in Step3 and Step4. When the interval between non-zero intervals is less than 50 sample points, set the next non-zero interval to zero, and combine the results in Step3 and Step4. The main purpose of doing this is to abstract the actual QRS complex and quickly locate the position of the QRS complex. Prepare for the next step to estimate the position of the P wave.

Step6. The part after Step5 processing is in the QRS wave group, find the most value point in the wave peak as the point of R peak.

Step7. After Step4 processing, if the non-zero interval does not fall back at the edge of the sliding window, it is considered that the QRS complex may not have reached the highest point, that is, there may be an error in the latest R peak value found in Step6, and the analysis in the seventh step The latest R peak marker for .

Step8. If the non-zero interval processed by Step4 has fallen back at the edge of the window, perform Step6 again to compare the new result with the R peak position already found in the sliding window to determine the latest R peak position.

Step9. Calculate the sample point where the sign of the second derivative changes in the two size of 20 sample intervals on the edge of the interval that is not zero after processing in Step5 as the interval of the R wave.

Step10. Calculate the interval of the complete heartbeat in the middle through the two RR intervals before and after, and analyze the windows before and after the QRS complex, and find out the interval that meets the sub-threshold THR3 as the interval of P and T waves for analysis.

Step11. Save the detailed data of QRS wave group, P wave, T wave (if P wave, T wave can be detected) in each heartbeat interval, such as interval time, peak value, start and end sample point position information and send it to the recognition module.

The electrocardiogram detection module is the cornerstone of the system's accuracy in heartbeat type analysis and plays a vital role.

Design of Heartbeat Recognition Module for Compound Event Processing

The composite event processing (CEP) heartbeat recognition module is the core of the ECG hybrid continuous analysis system and the basis of the hybrid analysis. This module uses CEP technology to process and analyze abnormal events in real-time heart rate. A single abnormal heart rate event will cause a series of complex events of abnormal heart rate. By monitoring the basic events identified by the monitoring and identification of the ECG detection module, various abnormal heart rhythm events can be monitored, and the occurrence of multiple abnormal heart rhythm events with specific rules can be considered as abnormal heart rhythm complex events that conform to the expression pattern in the EPL sentence . The system's monitoring of these complex events can promptly notify medical staff and provide patient symptom data at the first time to help save the patient's life to a certain extent. Accompanying drawing 2 is a hierarchical diagram of complex events used in the system. Firstly, the basic heartbeat events are generated by the ECG detection module in the system, which are represented by boxes; all ovals represent complex events. In the figure, EPL statements are used for pattern matching, and the occurrence of some specific basic events will trigger EPL statements to generate complex events. , the complex event generated by the basic event We locate the first-level complex event, which is the first layer of the complex event in the hierarchy diagram. The generated complex events will also flow into the system. When an EPL statement monitors the occurrence of a complex event that conforms to the pattern, a higher-level complex event will be generated.

The higher the level of complex events, the more specific the symptoms of cardiac abnormalities and the more life-threatening the occurrence of this event for the patient. For example, an ordinary patient with coronary heart disease may experience thousands of premature ventricular contractions every day, commonly known as premature beats, which are too frequent to judge whether the patient has serious symptoms. However, if the frequency and pattern of premature contraction induces a secondary event such as a VT tachycardia event, clinically it means that an unpredictable and potentially fatal situation may occur in the patient; if the VT event lasts to a certain extent, it will The occurrence of a third-level complex event, such as a myocardial ischemic event, means that the patient needs immediate first aid, and if measures are not taken, it will be life-threatening.

The flow of the compound event processing heartbeat recognition module is shown in Figure 3. CEP is a real-time analysis of continuous signal sources, so the flow chart does not have a way to mark the end of the recognition process. When any ECG monitoring data source is connected to the system, the ECG detection module will perform preprocessing and waveform detection on the sampled signal. Then each detected heartbeat information is encapsulated into basic heart rhythm events and stored in the basic heartbeat. When a basic heart rhythm event occurs, the EPL statement that the system is responsible for identifying the basic heart rhythm event will identify whether the condition of a normal heartbeat is met, and if there is no abnormality, it will monitor the next heartbeat event. If the data monitored by the CEP does not meet the definition of a normal heart rhythm, it will match the abnormal heart rhythm according to some existing matching rules and store it in the abnormal heart rhythm table. The complex event system will judge whether the abnormal heart rhythm can form other more complex abnormal heart rhythm events according to the position of the abnormal heart rhythm event in the event tree or event graph. If the event does not reach the top layer of the event tree, the abnormal heart rhythm pattern of the previous layer is added to the CEP monitoring mode. And if it has reached the top level of the event tree, it means that the event has been monitored, and the corresponding pattern matching statement needs to be removed. Finally, the new complex event is sent to the complex event processing heartbeat identification module. Afterwards, the compound event processing heartbeat recognition module will execute the steps in the flow chart in a continuous cycle.

The electrocardiogram detection module makes the basic information of the heartbeat into basic heart rate events and sends them to the composite event processing heartbeat recognition module, and compresses the ECG data into the historical database module for later search; the composite event processing heartbeat recognition module receives the transmitted Pattern matching will be performed after the heartbeat event, and all basic heart rhythm events of each heartbeat will be stored in the database as historical data. If the characteristics of abnormal heartbeat are met, the characteristics of abnormal heartbeat will be stored in the abnormal characteristic database for later pre-diagnosis output module to make the call.

Heartbeat event composition: the data read from the ECG monitor, etc., can detect the detailed characteristic values of a heartbeat through the electrocardiogram detection method after preprocessing, such as the detailed information of QRS wave group, P wave, T wave as Basic heart rhythm events, basic heart rhythm events are only some parameters that can reflect the current heartbeat detected in real time. Although it can also provide many related detailed information, this information is still far from the goal of automatically detecting abnormal heartbeats. , unable to express a complete meaning. Therefore, it is necessary to aggregate basic heart rhythm events into complex events with practical medical meaning. Below is a basic description of the event.

Basic heart rhythm event: Each basic heart rhythm event has a unique name, which consists of patient Uid+BeatPart+BeatID. The source list indicates whether the ECG signal source comes from a mobile device or an ECG monitor, and the device number; and the medical record database name indicates the location where the medical record is stored. The attribute list includes: ecg_id is the information indicating which lead; Beat_ID is used to mark which heartbeat attribute event it is; strat_time and end_time indicate the start time and end event of the event, and end_time can be empty if the event has no duration.

Complex events: Compared with basic events, there are more subevents and constraint list constraints. Complex events are composed of multiple subevents and their constraint relationships. Complex events are mainly used to represent some unique abnormal heart rate events, such as VT ventricular tachycardia, ventricular bigeminy, ventricular trigeminy, etc.

The function of the hybrid continuous analysis module in the complex event processing heartbeat identification module is to extract the information of the patient's historical data for continuous analysis of real-time data. The mixed continuous analysis module records the medium and long-term indicators of the patient's heart rate, and uses these indicators to adjust the screening of cardiac physiological events caused by individual differences, and applies them to complex event processing to make the matching used in complex event processing Modes can be changed and adjusted dynamically.

History database module design

If the analysis system wants to compare with the historical ECG data, it needs to add the ECG historical database module to store the historical data. The historical database module is mainly divided into five types of data tables.

The first category is the basic information table of the patient, which contains basic information such as the patient's patient number, name, age, sex, and medication history. Remarks on daily behavior information, such as whether there is any physical activity or labor behavior. What is stored in the basic heart rate table is the basic information parameter of each heartbeat identified by the electrocardiogram detection method and sent to the compound event processing heartbeat identification module.

The second type is to store the basic heart rate monitor, which refers to the relevant data of each heartbeat, such as date, heart rate ID, RR interval, detailed parameters of P wave, QRS wave group, T wave, etc.; and this type of basic heart rate storage table can be divided into There are two types, one is the data storage table of the current month, and the other is the data storage table of historical ECG data. The stored basic heart rate table stores the basic heart rate of each month. The table field structure is consistent with the basic heart rate. Every month, the system will copy the basic heart rate table of this month to the stored basic heart rate table and clear the basic heart rate table. to store all heart rate information for the month.

The third category is the abnormal heart rate table, which stores all abnormal heart rates, including the ID of abnormal heartbeat, abnormal heart rate type, duration and abnormal heart rate ID, etc.; it can also be divided into two categories, one class stores the abnormal heart rate of the current month , another type of abnormal heart rate storage history.

The fourth category is the abnormal heart rate sub-table, which divides and stores different abnormal heart rates according to the three regions of the ventricle, atrium, and junction area, such as ventricular tachycardia, atrial premature beats, and junctional area escaping. The sub-table database is divided into three, namely the ventricular abnormal heart rhythm database, the atrial abnormal heart rhythm data and the junction area heart rate abnormal database.

The fifth category is the HRV analysis value table, which is a heart rate analysis value based on the electrocardiogram as the data source. These calculated values can well represent some characteristics of the measured heart, and these characteristics can be used to analyze the measured heart rate. The possibility and degree of chronic heart failure. HRV (heart rate variability) reflects the activity of the autonomic nervous system and quantitatively evaluates the tone and balance of the cardiac sympathetic and vagus nerves, so as to judge the condition and prognosis of cardiovascular diseases. It is an extremely valuable indicator for predicting sudden cardiac death and arrhythmia events.

Pre-diagnosis output module design

When the complex event processing heartbeat recognition module recognizes an abnormal heart rhythm, it will start a mixed query, and at the same time query the historical ECG data, abnormal heart rate pattern library, and medical records and compare it with the real-time ECG situation. When the relevant part of the patient's medical record is found by comparison, and there is a high risk level, the abnormal heartbeat alarm and pre-diagnosis information will be transmitted to the pre-diagnosis output module, and the pre-diagnosis output module will send an alarm to the physician and patient's family members set in the patient's medical record Or a message for help, and wait for others to receive the message for treatment.

If you want to understand a person's heart rate, medium and long-term ECG data will bring great help and benefits. The change of a person's heart rate in the medium and long term can directly reflect the change rule and pathological changes of the patient's heart. And it can be used as the person's heart rate benchmark index to be used in individual heart rate monitoring.

As shown in Figure 4, it is a heartbeat rate mixed monitoring flow chart. In the figure, when the basic heart rhythm event is generated by the electrocardiogram detection module, it will be monitored by the compound event processing heartbeat recognition module. The heart rate is slightly different during the day and night, so after judgment The corresponding medium and long-term historical averages will be taken from the historical data of night or day respectively. Real-time basic heart rhythm events are mixed with historical database data for judgment. When the heart rate is less than 65% of the average historical heart rate, it is considered bradycardia; when the heart rate is 200-250% of the historical heart rate, it is considered tachycardia; 250% of the historical heart rate is considered to be a tremor of the heart. With the continuous development of technology today, the normal heart rate range introduced in traditional textbooks and the defined ranges for tachycardia, tremor, and cardiac escape are no longer applicable to the needs of individual monitoring. The mastery of medium and long-term heart rate indicators will be helpful for personalized heart rate monitoring. Historical data allows for individualized threshold setting and event analysis against each individual's heart rate baseline.

Complex Events - CEP Heartbeat Recognition

Composite event processing is the data source of the heartbeat recognition module, and the electrical signal of each lead of the ECG monitor will be encapsulated into a basic heart rhythm event after passing through the ECGD detection module. A complex event is composed of one or more basic rhythm events or complex events. The generation of complex events is generated by the CEP system parsing the pattern in the EPL statement and matching the corresponding pattern in the basic event flow. After the complex event is generated, it will also be sent to the CEP system in the form of an event flow, and may be Other pattern detection and matching;

Fig. 5 is a schematic diagram of an event tree of a ventricular tachycardia event, which shows the basic composition of the complex event VT ventricular tachycardia and the events of the next level. Ventricular tachycardia (VT): referred to as ventricular tachycardia, refers to the cardiac rhythm that originates from the ventricle, spontaneously, and consists of 3 or more consecutive premature contractions with a frequency greater than 100 beats/min.

First, the compound event processing heartbeat recognition module detects the characteristics of the basic heart rhythm event to match the premature ventricular contraction, and adds the second abnormal heart rhythm event to the event flow. When the abnormal heart rate event of premature ventricular contraction is added to the event stream, the EPL statement related to it will be dynamically activated, that is, the subsequent matching pattern monitoring related to premature ventricular contraction will be increased. For example, the mode monitoring of VT ventricular tachycardia shown in FIG. 5 , ventricular bigeminy shown in FIG. 6 , and ventricular trigeminy shown in FIG. 7 .

Afterwards, the compound event processing heartbeat recognition module will continue to monitor the patterns contained in all EPL statements on the line. If there are two consecutive basic events identified as ventricular premature contraction, and the average frequency of the RR wave intervals of the three consecutive heartbeat events is higher than 100 per minute, that is, the average value of the RR wave intervals is less than 216 Match the characteristics of VT ventricular tachycardia. In this case, the system will generate a VT complex event and add it to the current time stream. If there are more complex patterns based on VT ventricular tachycardia, CEP will activate the monitoring of the relevant patterns.

Since the cardiac characteristics of each patient are different, usually not all heart rhythm events are monitored, but only events that can be obtained by monitoring a single basic heart rhythm event in the event tree are monitored. When a specific abnormal heart rate event is activated, the EPL statement at the parent node of each event tree containing the abnormal heart rate event will be activated and monitored in the CEP. This can effectively save resources and reduce system burden.

The complex events of VT only need the following information, patient ID, patient heartbeat number, date, and start time and end time of the abnormal heart rhythm event. It can be used as the start time of the VT event, and then record the end time of the event when the heart rate event no longer matches this complex event. Afterwards, the abnormal heart rhythm complex event is stored in the database corresponding to the abnormal heart rhythm.

The characteristics of ventricular premature contraction are: the QRS wave group appears early, its shape is abnormal, and the time limit is mostly >0.12 seconds, the T wave is opposite to the main wave of the QRS wave, the ST wave shifts with the T wave, and there is no P wave before it. In the case of premature ventricular contractions at the proximal end of the bundle branch, the QRS complex may not be widened. Most of the premature ventricular contractions have a complete compensatory interval. When the basic rhythm is slow, premature ventricular contraction can be inserted between two sinus beats, forming an inserted premature ventricular contraction. Occasionally retrograde P waves that travel retrograde to the atrium, often appearing on the ST segment.

Usually, as the research progresses, the number of layers of the event tree will gradually increase or the event tree will grow, and the number of event trees will also increase. This can continue to support more complex events without making changes to the system, which is also the flexibility of CEP technology.

hybrid continuous analysis

The implementation of complex event analysis for ECG monitoring has been described in the previous section. This is very helpful for patients with general arrhythmia, but in practical applications, each person's heart rate benchmark index and the ECG patterns produced by the same abnormal heart rate phenomenon are different, which virtually brings the system to the accurate matching of individuals. Here comes the challenge. The original intention of the application of hybrid continuous analysis is to enable this ECG analysis system based on complex event processing to have stronger individual compatibility and accuracy. And it can realize the analysis and prediction according to the physical condition of different patients.

Based on complex events, the implementation of the hybrid continuous analysis module will be divided into three parts: hybrid basic events, hybrid complex events and comprehensive pre-diagnosis.

The detection of the basic heart rate is done by the electrocardiogram detection method, which has a high accuracy for the detection of the QRS wave group; but the accurate detection of the P wave and the T wave is still a medical problem, because most In some cases, the different cardiac pathological characteristics of patients lead to great differences in the relative positions of P waves and T waves relative to the QRS complex in different patients. It is difficult to accurately locate them with fixed detection methods and provide them to CEP for analysis, so hybrid continuous analysis will solve this problem to a certain extent.

The electrocardiogram of each person's single normal complete heartbeat is the easiest to locate the P wave and T wave. The analysis and statistics of the historical normal heartbeat can get the specific P wave and T wave relative to the QRS wave in each person's normal heart rate. The location of the group. Using this more accurate distribution position interval unique to each person can solve the positioning problem of P wave and T wave to a large extent.

Further, the ECG of each person's single complete heartbeat can be extracted, corresponding to the characteristic value of the heartbeat type. If all individual heart rates in the past are excavated or analyzed, the historical characteristic value of a specific type of heart rate can be obtained. By comparing the two, the basic heart rate type and other attributes can be analyzed using a hybrid continuous analysis method.

Each person's basic heart rate and the presented ECG pattern have independent characteristics. The extraction of each complete heartbeat feature value of a patient can make the relatively fixed abnormal heart rhythm standard for the general public in the medical field more individualized.

On the basis of the original complex event processing system, the feature values of different abnormal heart rhythms that have occurred in the patient's historical records are extracted, and these feature values are used for real-time abnormal heart rhythm detection of the patient. The patient's history is kept in the database by day. A person's heart rate will be higher and more excited during the day than at night. So historical records need to separate day and night to analyze eigenvalues. Medically, the daytime interval is defined as 9AM to 8PM, and 9:00 am to 8:00 pm on the same day; the night time interval is defined as 9PM to 8AM the next day. So it can be seen that although the data is stored by day, when analyzing historical data, it is necessary to refer to the date and the interval between day and night. Therefore, it is necessary to modify the EPL statement of the CEP system and add an access module to the historical database in the system.

The processing of mixed and complex events needs to enable EPL statements to support the acquisition of historical data; Nesper supports the introduction of custom functions in EPL statements, and the use of custom functions to link corresponding database, take out the average characteristic value of the corresponding abnormal heart rhythm of the category in the database and use it as the EPL to perform pattern matching with the threshold value in the play.

The task to be completed by the pre-diagnosis output module is to quickly judge whether there is life-threatening when an abnormal heart rhythm is detected in real time and transform it into an abnormal heart rhythm complex event, and send information related to the symptoms to the medical staff. If the patient’s basic information includes the addition of a pacemaker and previous historical data, the reference value will be lost, and what drugs the patient has been taking can help medical staff quickly judge the diagnosis and treatment method and the safety of the prescription.

System effect and performance test

Database selection: There are three mainstream standard ECG databases in the world: the European AT-T ECG database, the AHA database provided by the American Heart Association, and the MIT-BIH database provided by the Massachusetts Institute of Technology. Including arrhythmia, ST segment changes, atrial fibrillation and other databases.

Evaluation of ECG Detection Methods

Table 1 The detection results of QRS complexes in the MIT database

The detection results of the ECG detection method on the MIT-BIH heart rate database are shown in Table 1, and the error rate is obtained by the following formula. It is not difficult to see from the results that the heart rate of most patients can be detected, but the error rate will increase to a certain extent for the heart rate with serious heart rate compensation and the disappearance of p waves and t waves. At the same time, the accuracy of ECG detection for patients with frequent premature ventricular contractions needs to be further improved.

Accompanying drawing 8 shows the detection error statistics of ten ECG records selected from the MIT-BIH ECG database. The left picture shows the statistics of the number of missed detections and the number of false detections. The figure shows the number of missed detections and the number of false detections of each record respectively. The number of missed detections is the number of missed heartbeats without recognition, and the number of false detections is the plus or minus 5 of the R wave peak. Signal points other than signal points are identified as the number of R peaks; the figure on the right is the total error rate statistics, and the error rate is obtained by dividing the sum of the false positive rate and the false positive rate by the total number of heartbeats.

It can be seen from FIG. 8 that the detection method has relatively large errors in the identification of the record 200 and the record 223, reaching more than 4% respectively. Therefore, the characteristics of record 200 and record 223 were specifically analyzed. In the ECG record of record 200, there were twelve minutes of ventricular bigeminy symptoms accompanied by severe electromyographic interference. For a 30-minute ECG recording, the twelve-minute bigeminy already occupies a large part of the time. In record 223, there is a VT tachycardia lasting 1 minute and 50 seconds, during which the detection accuracy of the heart rate will be greatly affected. From the analysis of these two records, the current ECG detection method still has some problems in the identification of relatively harsh signal environments and rapid continuous abnormal heart rhythms. The success rate and correct rate of the heart rate recognition method are the cornerstone of the heart rate analysis software, and also determine the accuracy and usability of the system. The performance and effect of the method can meet the standard of system design.

ECG complex event analysis system evaluation

CEP complex event detection results: In order to verify the role of CEP technology in heartbeat monitoring and analysis, the following experimental results were obtained after experimenting with the heartbeat records of MIT-BIH. Abnormal ventricular heartbeats were counted for records 200, 208, and 223. The number of occurrences is shown in Table 2. It can be seen from the table that there were 826 premature ventricular beats in the heart rate record No. 200, and the system identified 719 of them, of which 71 occurred for bigeminy, 63 of which were actually recognized, and 7 of which were ventricular tachycardias. Recognized 6 times.

Table 2 Statistics of identification of abnormal ventricular heart rate

As shown in Figure 9, the left figure shows the total number of premature ventricular beats and the number recognized by the CEP system in the three heartbeat records. The figure on the right shows the recognition error rate of ventricular premature beats, ventricular bigeminy, and ventricular tachycardia in record 200 and record 223. It can be seen from FIG. 9 that the recognition error rates of ventricular premature beats in record 200 and record 223 respectively reach 13% and 7.6%, while the error rates in the ECG detection method are 4% and 4.18%. This shows that the detection accuracy of the CEP system is affected by the ECG detection method, and there are still some problems in the identification of relatively harsh signal environments and rapid continuous abnormal heart rhythms.

The average recognition rates of the three sets of records for MIT abnormal heart rate are all higher than 85%. The experimental results well illustrate the applicability of the CEP technology.

Hybrid continuous analysis results evaluation

The statistics of historical data are added to the ECG hybrid continuous analysis system, and the statistical values of the basic characteristics of abnormal cardiac rhythm based on historical data can be obtained. Through these statistical values, the matching attribute value in the EPL statement responsible for pattern matching of abnormal cardiac rhythm can be modified. Thereby improving the accuracy of recognition.

As shown in FIG. 10 , the left figure shows the total number of premature ventricular contractions in the three heartbeat records and the identification number of the hybrid continuous analysis system. The graph on the right shows the comparison of the error rate of the recognition of ventricular premature beats by CEP and hybrid continuous analysis of record 200, record 208 and record 223.

It can be seen from Figure 10 that compared with the CEP system, the hybrid continuous analysis technology has a higher recognition accuracy rate for abnormal heart rate, and the recognition error rate of record 200 has been reduced from 12.95% to 6.77%; the corresponding errors of record 208 and 223 Rates also decreased from 3.12% and 6.71% to 2.72% and 5.07%, respectively. The decrease in the error rate is obvious, and the main reason is that there are a large number of bigeminy phenomena in the record 200, and the alternate occurrence of bigeminy phenomena will interfere with the phenomenon of CEP searching for premature heart rhythm events. Through the mixed continuous technology, the appropriate attribute matching range can be found for the statistics of the patient's historical heart rate, so that the identification of the heart rhythm is more accurate.

Experiments have proved that the statistical analysis of historical data is helpful for the identification and analysis of real-time heart rate. It can not only identify the individual's physical fitness, but also improve the accuracy of identification, thus proving the desirability of hybrid continuous analysis technology.

Claims (9)

1. the application process mixing continuous information analytical technology medically, it is characterised in that: by electrocardio mixing continuously ECG signal acquisition module in analysis system obtains ECG signal from electrocardio equipment, then detects mould by electrocardiogram Block carries out pretreatment to ECG signal, and the signal after process extracts signal characteristic by electrocardiogram detection method and makes one Each and every one basic heart rate event is analyzed for complex event processing heart beating identification module, complex event processing heart beating identification mould Block by mixing therein analyze continuously module basic heart rate event and the information in patient's history's DBM are carried out right Ratio, identifies abnormal cardiac rate event, alert to pre-diagnosis output module transmission abnormality heart beating when there being serious abnormal cardiac rate event Report and pre-diagnostic message, the pre-diagnosis output module doctor to patient and/or family members send alarm and distress signals.

Mixing continuous information analytical technology the most according to claim 1 application process medically, it is characterised in that: institute Stating ECG signal acquisition module and obtained signal by electrocardiogram monitor or portable electrocardiograph, signal passes through Bluetooth transmission to mobile phone Or PC.

Mixing continuous information analytical technology the most according to claim 1 application process medically, it is characterised in that: institute State electrocardiogram detecting module and identify each heart beating from the voltage signal sequence that the transmission of ECG signal acquisition module comes, and Extract the heart rate of heart beating, RR wave spacing, P ripple, QRS complex and T wave property parameter, and these characterisitic parameters are fabricated to base This heart rate event is also transferred to the complex event processing heart beating identification module coupling for abnormal cardiac rate event；Simultaneously by electrocardio Data are stored in history database module.

Mixing continuous information analytical technology the most according to claim 3 application process medically, it is characterised in that: institute State electrocardiogram detecting module and include signal pre-processing module and electrocardio detecting module；Electrocardio detecting module includes that QRS complex is examined Survey module and P, T ripple detection module；The voltage signal that wherein transmission of ECG signal acquisition module is come by signal pre-processing module Remove Hz noise and for baseline drift problem first by wavelet transformation, and adjust baseline with fitting of a polynomial；QRS Wave group detection module use dynamic thresholding detection method data that real-time Transmission is come carry out pointwise variance, pointwise square and Signal amplitude pointwise square operation, is having the integration doing moving window at R peak；P, T ripple detection module is for a complete heart The position of P, the T ripple that the window phase jumped is analyzed the signal before and after QRS complex finding heart beating corresponding and information.

Mixing continuous information analytical technology the most according to claim 4 application process medically, it is characterised in that: institute State dynamic thresholding detection method and include following steps:

A, set up the window that 1 size is 1000 sample points, along with new signal input, old sample point grand window；

B. pretreated signal carries out sliding average process, and the sliding window of 11 sample points is averaged；

C, use dynamic threshold THR1 equal to the meansigma methods of sample point in sliding window and sample average and filter out and be less than The signal section of threshold value；

D, use dynamic threshold THR2 equal to the meansigma methods of sample point in sliding window and sample average difference filter out height In threshold signal part；

The interval in the non-zero interval that E, contrast step C and step D produce, when being spaced non-zero interval less than 50 sample points, will Adjacent latter one non-zero interval zero setting, and step C is merged with the result of step D；

F, part after step E processes are all to find in crest value point as the point at R peak in QRS complex；

G, step D process after, if non-zero interval do not fall after rise at sliding window edge, then regard as QRS complex and be not reaching to High point, there is error in the up-to-date R peak value that both step F had been found, up-to-date R peak markers step G analyzed；

If the non-zero interval after H step D processes has been fallen after rise at window edge, perform step F the most again by new result pair Than the R peak position found in sliding window so that it is determined that the position at a up-to-date R peak；

Two sizes at interval edge that J, calculation procedure E are not zero after processing are just to find out second dervative in 20 sample intervals The sample point that negative sign changes is as the interval of R ripple；

K, interval by former and later two RR ripples, calculate the interval of middle complete heartbeat, and the window before and after QRS complex is carried out point Analysis, finds out and meets the interval of subthreshold THR3 and be used as the interval of P, T ripple and be analyzed；

L, the QRS complex by each heart beating interval, P ripple, the interval time of T ripple, peak value, the sample point position letter of state pause judgments Breath preserves and issues complex event processing heart beating identification module.

Mixing continuous information analytical technology the most according to claim 1 application process medically, it is characterised in that: institute State complex event processing heart beating identification module by mixing therein analyze that electrocardiogram detecting module exports by module continuously every One basic heart rate event is monitored, by EPL each base of statement identification in complex event processing heart beating identification module Whether this heart rate event meets the condition of normal heartbeat, for abnormal cardiac rate event by with in patient's history's DBM Abnormal cardiac rate table carries out matching judgment, and it is any abnormal cardiac rate event, and is in event tree or thing according to abnormal cardiac rate event Position in part figure judges that can this abnormal cardiac rate event form other increasingly complex abnormal cardiac rate event, and judgement is No remove corresponding pattern matching statement, finally complicated event new for gained is sent to complex event processing heart beating identification mould Block；And all basic heart rate events of each heart beating are stored in history database module as historical data.

Mixing continuous information analytical technology the most according to claim 1 application process medically, it is characterised in that: institute State that history database module includes the Basic Information Table of patient, storage basal heart rate table, abnormal cardiac rate summary table, abnormal cardiac rate divide Table and heart rate variability analysis numerical tabular；Each table is divided into two storage parts of day and night according to time difference.

8. according to the application process medically of the mixing continuous information analytical technology described in any one in claim 1-7, It is characterized in that: described mixing is analyzed module continuously and the medium-term and long-term index of patient heart rate recorded, and utilizes these indexs Adjust the examination of the heart pathology affair caused because of individual variation, and applied to complicated heart rate event handling makes The match pattern used during complicated heart rate event handling can be changed dynamically and adjust.

Mixing continuous information analytical technology the most according to claim 8 application process medically, it is characterised in that: institute State mixing to analyze continuously the realization of module and include three parts: mixed base present event, mixing complicated event and comprehensively examine in advance Disconnected；Wherein mixed base present event is by the analysis of the normal heartbeat to history and statistics, obtain in everyone normal cardiac rate P ripple and The distributing position relative to QRS complex of T ripple, and single heart rate all of to the past excavates or analyzes and obtain the rhythm of the heart History feature value, both use the continuous analysis mode of mixing to compare, thus analyze the category attribute of basic heart rate；Mixed Close complicated event by, in the complex event processing heart beating identification module snoop procedure to abnormal cardiac rate event, extracting patient The eigenvalue of the different abnormal cardiac rate events occurred in history database module, and it is the most different that these eigenvalues are used for patient Often heart rate event detection；When real-time detecting abnormal heart rhythm and being converted into the complicated event of abnormal heart rhythm, by comprehensively Pre-diagnosis judges whether to be in peril of one's life rapidly, and the information relevant to symptom is sent to medical personnel.