WO2019219035A1

WO2019219035A1 - Vital sign data processing method and system based on cloud platform

Info

Publication number: WO2019219035A1
Application number: PCT/CN2019/087109
Authority: WO
Inventors: 陈韵岱; 隆云; 翟茜; 韩宝石; 黄晓波; 吴屹; 王新康; 孙思楠; 江永; 郝晓宁; 吕卫华; 唐祖骏; 杨昊宇; 张锦景
Original assignee: 上海术木医疗科技有限公司
Priority date: 2018-05-16
Filing date: 2019-05-15
Publication date: 2019-11-21
Also published as: CN108847274B; CN108847274A

Abstract

A vital sign data processing method and system based on a cloud platform, which belong to the field of medical cloud computing, solve the problem of a user not having the ability to analyze and process vital sign data, and also solve the problems of data formats and communication protocols of vital sign monitoring devices from different manufactures being incompatible with one another and being difficult to integrate and efficiently process. The method comprises: acquiring user data in real time for preprocessing and uniform encapsulation, and storing same in a vital sign database; applying a learning framework based on distributed parallel computing to the reading and analyzing of massive vital sign data in real time; screening abnormal data; automatically generating an electronic data analysis report; and sending, in a timely manner, an event abnormity prewarning to a user to prompt a healthcare professional to intervene quickly, thereby improving medical quality and working efficiency, reducing false alarm events of a device, and reducing the labor intensity and working pressure of the healthcare professional. Applying quantitative and qualitative data in a vital signal database to train and optimize a central model in real time further improves the data processing efficiency of a cloud platform.

Description

Method and system for processing vital sign data based on cloud platform

Technical field

The present application relates to the field of medical cloud computing, and in particular, to a cloud platform-based vital sign data processing method and system.

Background technique

Vital signs monitoring equipment including bedside multi-parameter monitor, respiratory function monitor, intracranial pressure monitor, fetal heart monitor is the hospital intensive care unit (ICU), as well as cardiology, respiratory, neurosurgery, The main equipment of the ICU in emergency department, obstetrics and gynaecology, etc., is used to monitor the vital signs of patients in real time, which plays an important role in saving the lives of patients. In the United States, there are millions of ICU custodial services per year. China is the country with the largest number of ICUs in the world. There are millions of types of vital signs monitoring devices in hospitals across the country, and they are growing rapidly.

The problem with global vital sign monitoring equipment is that each device generates hundreds of MB of data every day, but there is no long-term data storage capability, the device false alarm rate is extremely high, and it is impossible to generate an electronic data analysis report, which requires the hospital ICU medical staff to manually real-time. Analysis of the identification data, manual transfer of relevant information and data, the global hospital ICU medical staff for a long period of high load, exhausted. In addition, the hospital lacks experienced ICU medical staff for a long time, unable to cope with the complicated and heavy vital signs data processing work, directly affecting the quality of medical care. In recent years, the Clinical Information System (CIS) of the intensive care unit has been able to reduce the labor intensity of the ICU, but it does not solve the problem of data analysis and data management of users' vital signs, and requires users to have expensive hospital information. Supported by the Hospital Information System (HIS). In the prior art, data on vital signs monitoring equipment of lower-level hospitals are generally sent to higher-level hospitals remotely, helping ICU doctors in lower-level hospitals to analyze and diagnose vital signs data and guide clinical medical work. However, when a large number of ICU patient data in lower-level hospitals are concentrated in the upper-level hospitals, it will bring unbearable pressure to the higher-level hospitals, and it is difficult to achieve scale.

The global vital sign monitoring device already has a variety of data communication interfaces, but each manufacturer has its own communication protocol and data format, which are not compatible with each other. Some existing technologies can solve the problem of data receiving services of other manufacturers' devices. However, the problem of different data formats of different manufacturers is not solved, and multiple processing softwares need to be set up, and the data storage format is matched, and the efficiency is obviously low. At the same time, there are many manufacturers' communication protocols that do not support the input of patient information. This technology cannot guarantee the identification of different patients in the same device (same bed). In hospitals, the same device often has to serve many patients. In the prior art, the information module and the communication module are connected to the vital sign data source module, and the communication module is connected to the Internet to send the data to the cloud computing module and the cloud database. The technology is in each device and the cloud platform of each manufacturer. Between the two, the intermediate processing link has been added, not only the complexity and cost of each device have increased significantly, but also the reliability has decreased.

Summary of the invention

In view of the above analysis, the present application aims to provide a cloud platform-based vital sign data processing method and system for solving the problem that the hospital lacks the ability to analyze and interpret vital signs data, and to solve the data format of different manufacturers' vital signs monitoring equipment, Communication protocols are incompatible with each other and it is difficult to concentrate and efficiently deal with them.

The purpose of this application is mainly achieved by the following technical solutions:

In one aspect, a cloud platform-based vital sign data processing method is provided, including the following steps:

Step S1: acquiring a plurality of user data in real time; the user data includes vital sign monitoring device ID code, patient information, and vital sign data;

Step S2: pre-processing each acquired user data and performing unified encapsulation and storing the data into the vital sign database;

In step S3, the vital sign data in the vital sign database is read in real time, and the deep learning framework of the distributed parallel computing is used for analysis and screening processing, and the analysis screening result is obtained, and a data analysis report is generated.

The beneficial effects of the application are as follows:

1. Through the real-time analysis and processing of massive vital signs data, the problem of lack of ability to analyze and interpret vital signs data in hospitals is solved, and medical quality and work efficiency are improved.

2. Real-time analysis and calculation of the original alarm event data of the equipment, effectively reducing the false alarm events frequently occurring during the monitoring process, improving the accuracy of the abnormal event warning, and reducing the labor intensity and work pressure of the medical staff.

Based on the above scheme, the application also made the following improvements:

Further, in step S2,

Binding the vital sign monitoring device ID code and the patient information according to the system coding table rule, generating a patient service serial number, and maintaining bidirectional mapping conversion with the device ID code;

Parsing, classifying, and normalizing the data format of the received vital signs data, and retaining the original alarm event flag of the device;

The patient business serial number and the preprocessed vital sign data are uniformly encapsulated and stored in the vital sign database.

The beneficial effects of using the above further solution are:

The user information, clinical information, and data information are linked by the service serial number, and the device ID is bidirectionally mapped and converted, thereby solving the problem of identifying different patients of the same device (the same bed), and establishing a reliable and efficient internal data query and data interaction. Relationship with external logic to meet internal data query and interaction with external data.

Further, the real-time reading of the vital sign data in the vital sign database and the use of the distributed parallel computing deep learning framework for analysis and screening processing are implemented by an online real-time data analysis processing method and a deep learning framework based on the Spark engine:

The Spark distributed parallel computing deep learning framework reads the vital sign data in the vital sign data. According to the set micro-batch processing interval, the Spark engine creates multiple tasks in parallel, triggering Spark Streaming to cut the data by type. It is divided into RDD data sets, and the central model of the corresponding type is controlled to calculate and process the type data.

Further, the central model is divided into two categories, one is to analyze and calculate the shape, rhythm and rate of the waveform data, and the other is to analyze and calculate the amplitude of the numerical data; the central model includes a second-order differential calculation tool and / or logic analysis tools, real-time calculation and analysis of the shape, rhythm, rate, and value of vital sign data, classification and labeling of waveforms, statistical summarization of logarithmic values, real-time analysis and screening of abnormal data beyond the benchmark.

Further, when the central model calculation process finds abnormal data exceeding the set reference, the abnormal data feature is analyzed, the duration is calculated, and the abnormal data attribute is marked.

Further, when abnormal data is found, the abnormal data is generated into a real-time data analysis report, an abnormal event warning is issued to the user, and the real-time data analysis report is sent to the user.

The beneficial effects of using the above further solution are:

The deep learning framework based on Spark engine has the advantages of distributed, high throughput and self-learning. It realizes real-time analysis and processing of massive vital signs data, solves the problem of lack of ability of hospitals to analyze and interpret vital signs data, and timely discovers abnormal data and supports medical care. Rapid response interventions improve the quality of medical care and work efficiency.

Further, using the vital sign data after the analysis and screening process in the vital sign database, each type of central model is trained and optimized in real time to obtain a new central model of the type data.

The beneficial effects of using the above further solution are:

The central model applies quantitative and qualitative vital sign data for training optimization, which can further improve the analysis and calculation accuracy of the central model, and thus effectively improve the data processing efficiency of the cloud platform, reduce the frequent equipment false alarm events during the vital sign monitoring process, and reduce Medical staff's labor intensity and work pressure.

Further, the entire life vital sign data of each user after the analysis and screening process is integrated, the dynamic data analysis report is automatically generated, and the dynamic data analysis report is sent to the user according to the patient service serial number.

The beneficial effects of using the above further solution are:

It solves the problem that the vital signs monitoring equipment lacks the summary data of the electronic data in the application process. Through the dynamic data analysis report, the medical staff can analyze and diagnose the patient's condition, evaluate the clinical treatment effect, adjust the treatment plan, and effectively improve the user's Medical quality and work efficiency.

On the other hand, a cloud platform-based vital sign data processing system is provided, including: a cloud platform data communication subsystem and a cloud platform data support subsystem; the cloud platform data communication subsystem includes a data communication module and a data pre- a processing module; the cloud platform data support subsystem comprises a message bus module, a data storage module, and a real-time analysis processing module;

The message bus module is configured to connect data transmission between the control data communication module, the data preprocessing module, the real-time analysis processing module, and the data storage module;

The data communication module is configured to receive a plurality of user data in real time, and is also used for data interaction with the user, and the received data is transmitted to the data preprocessing module, where the user data includes a vital sign monitoring terminal device ID. Coding, patient information and vital signs data;

The data pre-processing module includes a system coding table, configured to perform pre-processing on each acquired user data, and perform unified encapsulation and then store the data in the vital sign database;

The data storage module includes a vital sign database, a file database, a business information database, and a cache database for data storage and calling;

The real-time analysis processing module includes a deep learning framework for distributed parallel computing, which is used for real-time reading vital sign data in the vital sign database, performing analysis and screening processing, generating a data analysis report, and transmitting the analysis report to the user. Save to the file database.

The beneficial effects of using the present application are as follows:

Further, the data preprocessing module is configured to:

Binding the vital sign monitoring terminal device ID code and the patient information according to the system coding table rule, generating a service serial number, and performing bidirectional mapping conversion with the device ID code;

The beneficial effects of using the above further solution are:

Further, the real-time analysis processing module reads the vital sign data in the vital sign database in real time, and performs online analysis and processing by using an online real-time data analysis processing method and a deep learning framework based on the Spark engine;

The Spark distributed parallel computing deep learning framework reads the vital sign data in the vital sign data. According to the set micro-batch processing interval, the Spark engine creates multiple tasks in parallel, triggering the Spark stream to divide the data into RDD data by type. The collection, while controlling the central model of the corresponding type, performs an analysis screening process on the type of data.

Further, when the central model calculation processing finds abnormal data exceeding the set reference, the real-time analysis processing module analyzes the abnormal data feature, calculates the duration, and marks the abnormal data attribute.

Further, when the abnormal data is found, the real-time analysis processing module generates a real-time data analysis report of the abnormal data, issues an abnormal event warning to the user, and sends the real-time data analysis report to the user.

The beneficial effects of using the above further solution are:

The deep learning framework based on Spark engine has the advantages of distributed, high throughput and self-learning, realizing the real-time processing of massive vital signs data, solving the problem of lack of ability of hospital to analyze and interpret vital signs data, and timely discovering abnormal data, prompting medical staff Rapid response intervention to improve medical quality and work efficiency.

Further, the real-time analysis processing module uses the vital sign data after the analysis and screening process in the vital sign database, and performs real-time training on each type of central model to obtain a new central model of the type data.

The beneficial effects of using the above further solution are:

The real-time analysis and processing module trains and optimizes the quantitative and qualitative vital sign data after the analysis and processing of the central model application, which can further improve the analysis and calculation accuracy of the central model, thereby effectively improving the data processing efficiency of the cloud platform and reducing vital signs. Frequent equipment alarm events during the monitoring process reduce the labor intensity and work pressure of medical staff.

Further, the real-time analysis processing module integrates the whole vitality data of each user after the analysis and screening process, automatically generates a dynamic data analysis report, and stores the data in a file database, and sends the dynamic data analysis report according to the patient service serial number. To the user.

The beneficial effects of using the above further solution are:

In the present application, the above various technical solutions can also be combined with each other to achieve more preferred combinations. Other features and advantages of the present application will be set forth in the description which follows. The objectives and other advantages of the present invention are realized and attained by the description and the appended claims.

DRAWINGS

The drawings are only for the purpose of illustrating the embodiments, and are not intended to

1 is a flowchart of a method for processing vital sign data based on a cloud platform in an embodiment of the present application;

2 is a structural diagram of a vital sign data processing system based on a cloud platform in an embodiment of the present application;

3 is a deep learning framework based on Spark distributed parallel computing in the embodiment of the present application;

4 is a flowchart of real-time data analysis processing in the embodiment of the present application;

Detailed ways

The preferred embodiments of the present application are described in detail below with reference to the accompanying drawings, wherein the accompanying drawings illustrate,

A specific embodiment of the present application discloses a cloud platform-based vital sign data processing method, as shown in FIG. 1 , including the following steps:

Step S1: acquiring a plurality of user data in real time; the user data includes vital sign monitoring terminal device ID code, patient information, and vital sign data;

Step S2: pre-processing each acquired user data and uniformly encapsulating the data into the vital sign database;

In step S3, the vital sign data in the vital sign database is read in real time, and the deep learning framework of the distributed parallel computing is used for analysis and screening processing, and the analysis screening result is obtained, and the data analysis report is generated.

Compared with the prior art, the method processes and processes massive vital signs data (including raw alarm event data of the device) in real time, screens abnormal data, and prompts abnormal data warning to the user, prompts the medical staff to quickly respond to the intervention, and generates a data analysis report. The user can receive, read and download and download and print, as a medical basis, effectively reduce equipment false alarm events, improve the user's medical quality and work efficiency, and reduce the labor intensity and work pressure of medical staff.

In step S1, data interaction is performed by interacting with the vital sign monitoring terminal device (exemplarily, the vital sign monitoring device may be a multi-parameter monitoring device, a respiratory function monitor, an intracranial pressure monitor, a fetal heart monitor, etc.) User data.

It should be noted that in order to solve the problem of identifying different patients of the same device (same bed), and solving the problem of addressing and identifying data query and data interaction; the following steps are also included:

Binding the vital sign monitoring device ID code and the patient information according to the system coding table rule, generating a service serial number, and bidirectional mapping conversion with the device ID code;

The received vital sign data is parsed, classified, and the data format is standardized, and the original alarm event flag of the device is retained; the parsing refers to extracting data (including timestamps and numerical values) from the received data packet, and the classification refers to according to the data protocol. The parsed vital sign data is classified according to the parameter type; the data format normalization process is converted into the cloud platform vital sign data standard format by digital change, rate conversion, and re-encoding;

The patient business serial number and the processed vital sign data are uniformly encapsulated and stored in the vital sign database.

It should be emphasized that the service serial number is associated with user information, clinical information, and data information, and is associated with the vitality monitoring device ID, and bidirectional mapping is converted, thereby establishing a reliable and efficient internal and external logical relationship of data query and data interaction. It solves the problem of identifying different patients on the same device (same bed), meeting the internal data query of the system, and the need to interact with external data. The data format standardization process solves the problem that the data formats of the devices of different manufacturers are different and difficult to be processed centrally, and the data processing efficiency of the cloud platform is improved.

It should be noted that, in step S3, the vital sign data in the vital sign database and the original device alarm data included in the vital sign database are read in real time, and the real-time analysis and screening process is performed by using the deep learning framework of the distributed parallel computing, and the online process is adopted. Real-time data analysis and processing and deep learning framework based on Spark engine (you can also use one of the common Storm, Flink, and Samza frameworks), as shown in Figure 3, which has distributed, high-throughput, self-learning Advantages, greatly improve the real-time processing speed of massive vital signs data, and at the same time review the parameters, waveforms and marks of the original alarm event data of the equipment, effectively reduce false alarm events and reduce the labor intensity and work pressure of medical staff.

Specifically, the deep learning framework of the Spark distributed parallel computing reads the vital sign data in the vital sign data and the original alarm data of the included device, according to the set micro-batch processing interval (0.01 seconds or more), the Spark engine. Create multiple tasks in parallel, trigger the Spark stream to divide the data into RDD data sets by type, and control the central model of the corresponding type to analyze and process the type data. The central model can be divided into two categories, one for the waveform class. The shape, rhythm and rate of the data are analyzed and calculated, and the other type is used to analyze and calculate the amplitude of the numerical data. The central model includes a second-order differential calculation tool and/or logic analysis tool to calculate and analyze the shape and rhythm of the vital sign data in real time. , rate, value, classification and labeling of waveforms, statistical summation of logarithmic values, real-time analysis and screening of abnormal data beyond the benchmark.

It should be noted that the vital sign data is divided into a waveform class and a numerical class; wherein the waveform type vital sign data includes: total heart rate, ECG interval, QRS time limit, ST segment shape, QT interval, total respiratory time , respiratory wave interval, pulse volume peak-to-valley value, intracranial pressure peak-to-valley value, end-tidal carbon dioxide partial pressure peak-to-peak value, end-tidal carbon dioxide partial pressure wave interval; numerical vital sign data including: non-invasive/inclusive Systolic blood pressure and diastolic blood pressure, pulse rate, blood oxygen saturation, body temperature value, fetal heart rate; non-invasive cardiac output, stroke rate, cardiac index, total peripheral resistance value; airway pressure value of respiratory mechanics, Airway flow value, airway volume value.

Further, when the central model calculation analysis process finds abnormal data exceeding the set reference, analyzes the abnormal data feature, marks, calculates the duration, and marks the abnormal data attribute; meanwhile, generates the real-time data analysis report of the abnormal data, and sends the data to the user. The abnormal event is alerted, and the real-time data analysis report is sent to the user; it should be noted that the benchmark uses the diagnostic criteria of internationally-used vital sign data.

Abnormal data characteristics include: tachycardia, bradycardia, flutter, frequent premature beats, cardiac arrest, RonT, QT interval prolongation, ST segment elevation/depression, apnea, hypopnea, and rapid breathing. Increase/decrease in blood oxygen saturation, increase/decrease in systolic and diastolic blood pressure, increase/decrease in mean arterial pressure, increase/decrease in peak volume of pulse volume, increase/decrease in peak intracranial pressure wave, end-tidal carbon dioxide The peak value of the partial pressure wave rises/falls, the fetal heart rate increases/decreases, the non-invasive cardiac output decreases, and the respiratory mechanics value increases/decreases.

In order to improve the efficiency and accuracy of the central model analysis screening process, the method further comprises using the quantitative and qualitative vital sign data after analysis and screening in the vital sign database, and training and optimizing each type of central model in real time to obtain the type data. New central model.

Through the deep learning framework of Spark distributed parallel computing, the analysis and processing of massive vital signs data is realized, which satisfies the needs of user vital sign data analysis and interpretation, and improves medical quality and work efficiency. The central model uses real-time training and optimization of quantitative and qualitative vital signs data after analysis and screening, further improving the analytical calculation accuracy of the central model and improving the efficiency of data service of the cloud platform.

Further, the biometric data of each user of the analysis and screening process is integrated to generate a dynamic data analysis report, and is sent to the user according to the patient service serial number.

It should be noted that the user can read the data value corresponding to the vital sign database according to the dynamic data analysis report template content, including: the whole process of dynamic electrocardiogram data, dynamic blood pressure data, respiratory data, blood oxygen saturation data, and invasive blood pressure data. , intracranial pressure data, end-tidal carbon dioxide data, body temperature data, fetal heart rate data comprehensive analysis and calculation, waveform classification marks, waveform patterns, etc.;

It may also include the following steps: counting data and generating statistical charts of data, including: trend graph, histogram, scatter plot, variability analysis graph;

The method may further include the steps of: providing the dynamic data analysis report to the user for reading and downloading, downloading and printing according to the patient business serial number.

Through the dynamic data analysis report, the problem of lacking electronic data summary analysis and record in the application process of vital sign monitoring equipment is solved. At the same time, as the data basis of clinical medical treatment, the medical staff can analyze and diagnose the disease state of the patient and evaluate the clinical treatment effect. Making or adjusting medical plan decisions can effectively improve the quality of medical care and work efficiency of users, and reduce the workload of medical staff.

A second embodiment of the present application discloses a cloud platform-based vital sign data processing system, as shown in FIG. 2, including: a cloud platform data communication subsystem, a cloud platform data support subsystem, and a cloud platform data communication subsystem. The utility model comprises a data communication module and a data preprocessing module; the cloud platform data support subsystem comprises a message bus module, a data storage module and a real-time analysis processing module.

Specifically, the data communication module is configured to receive a plurality of user data and data interaction in real time, and deliver the received data to a data preprocessing module, wherein the user data includes a vital sign monitoring terminal device ID code, patient information, and vital signs. data;

The above data communication module supports a plurality of communication protocols, including: TCP/IP protocol, instant communication protocol, HL7 protocol, DICOM protocol, multimedia communication protocol, device manufacturer communication protocol, automatic identification of user identity and device ID coding, Establish a network connection, receive data, and send it to the data preprocessing module. By supporting a variety of communication protocols, the business service surface is effectively expanded to meet the data interaction of various vital vitality monitoring devices and external systems of the user, including the hospital information management system (HIS) and the intensive care clinical information system ( CIS), and the application programming interface (API) of the platform of the medical examination organization, health management organization, and insurance organization.

The data preprocessing module is configured to preprocess each acquired user data and perform unified encapsulation and then store the data in the vital sign database; specifically:

Binding the obtained vital sign monitoring terminal device ID code and the patient information according to the system coding table rule, generating a service serial number, and bidirectionally mapping and converting with the device ID code;

The patient business serial number and the preprocessed vital sign data are uniformly encapsulated and sent to the vital sign database for reading, invoking, calculating analysis, and retrieving statistics.

It should be noted that the service serial number includes time stamp, patient information, user information, device information, and quantity counter; the parsing is to extract data (including timestamp and numerical value) from the received data packet, and the classification is parsed according to the data protocol. The vital sign data is classified into a data type; the data format normalization process is converted into a vital sign data standard format of the embodiment by using a digital bit change, a rate conversion, and a re-encoding method.

It should be emphasized that the data format standardization process solves the problem that the data format of each manufacturer is incompatible. The unified data format is adopted to improve the operating efficiency of the cloud platform; the service serial number and the vital sign monitoring device ID maintain bidirectional mapping conversion. It establishes reliable and efficient internal and external logical relationships of data query and data interaction, satisfies the requirements of internal data query and interaction with external data, and solves the problem of identifying different patients of the same device (same bed).

The message bus module is used for connecting data transmission between the control data communication module, the data preprocessing module, the data real-time analysis processing module, and the data storage module;

The data storage module includes a vital sign database, a file database, a business information database, and a cache database for data storage and invoking; the data storage module integrates the advantages of various databases and data storage service systems, and solves the massive vital signs of the system. The problem of storage and access of data, as well as the management of large-scale data collections, diverse data structures, and multiple data types.

Specifically, the vital sign database belongs to a structured database for storing vital sign data after unified encapsulation; providing massive storage capacity and real-time query capability, featuring high concurrency, low latency, and flexible support;

The file database belongs to the object storage database, and is used for storing the vital sign data analysis report file, the clinical information file, the patient information file, the video image file, and the medical tool data file generated by the system;

The business information database belongs to a relational database for storing structured business data and business logic relationship data; based on the relationship model, it has the advantages of maintaining data consistency;

The cache database belongs to a non-relational database. It acts as a cache for data exchange and state preservation between modules. It is also used to cache the query results of the data storage module, which reduces the number of database accesses and improves the response speed of the system.

The real-time analysis processing module includes a deep learning framework for distributed parallel computing, which is used for real-time reading of vital sign data in the vital sign database for processing, generating a data analysis report, and transmitting the analysis report to the user, and simultaneously storing it in a file database; As shown in Figure 4, specifically:

Real-time reading of vital sign data of the vital sign database (including the original alarm data of the device), and using the central model of the vital sign data of the Spark engine-based deep learning framework for screening processing, obtaining analysis screening processing results, and abnormal data Attributes, and the data processed by the analysis and screening is stored in the vital signs database.

It should be emphasized that the real-time analysis processing module adopts a deep learning framework based on distributed parallel computing, and can be one of the general Spark, Storm, Flink, and Samza frameworks.

Specifically, the deep learning framework of the Spark distributed parallel computing reads the vital sign data in the vital sign data, and according to the set micro-batch processing interval (0.01 seconds or more), the Spark engine creates multiple tasks in parallel, triggering Spark. The stream divides the data into RDD data sets by type, and controls the central model of the corresponding type to analyze and process the type data; the central model includes second-order difference calculation tools and/or logic analysis tools to calculate and analyze vital sign data in real time. Morphology, rhythm, rate, and numerical values are used to classify the waveforms, statistically summarize the logarithmic values, and analyze the abnormal data beyond the baseline in real time.

It should be noted that the vital sign data is divided into waveform class and numerical class; specifically, the waveform type vital sign data includes: total heart beat, ECG interval, QRS time limit, ST segment shape, QT interval, full breathing Total number, respiratory wave interval, pulse volume peak-to-valley value, intracranial pressure peak-to-valley value, end-tidal carbon dioxide partial pressure peak-to-peak value, end-tidal carbon dioxide partial pressure wave interval; numerical vital sign data including: non-invasive Systolic and diastolic blood pressure, pulse rate, blood oxygen saturation, body temperature, fetal heart rate in non-invasive blood pressure; stroke volume, cardiac index, total peripheral resistance value of non-invasive cardiac output; airway pressure value of respiratory mechanics , airway flow value, airway volume value.

The above abnormal data features include: tachycardia, bradycardia, fluttering, frequent premature beats, stop, RonT, QT interval prolongation, ST segment elevation/depression, apnea, hypopnea, and rapid breathing. Increase/decrease in blood oxygen saturation, increase/decrease in systolic and diastolic blood pressure, increase/decrease in mean arterial pressure, increase/decrease in peak volume of pulse volume, increase/decrease in peak intracranial pressure wave, end-tidal carbon dioxide The partial pressure wave rises/falls, the fetal heart rate increases/decreases, the non-invasive cardiac output decreases, and the respiratory mechanics value increases/decreases.

It should be noted that the central model is divided into two categories, one is to calculate and analyze the shape, rhythm and rate of waveform data, and the other is to calculate and analyze the amplitude of numerical data; when the central model finds that it exceeds the set reference When abnormal data is used, the abnormal data characteristics are analyzed, the abnormal event duration is calculated, and the abnormal data attribute is marked; the real-time analysis processing module generates the real-time data analysis report of the abnormal data, stores it in the file database, and simultaneously sends it to the user.

In order to further improve the efficiency and accuracy of the analysis screening process and reduce false alarm events, the real-time analysis and processing module uses the quantitative and qualitative vital sign data after analysis and screening in the vital sign database to train and optimize each type of central model in real time. Get a new central model of this type of data.

The real-time analysis processing module of the embodiment may further integrate the whole life vital sign data of each user subjected to the analysis screening process, generate a dynamic data analysis report, and store the data in the file database, and send the dynamic data analysis report according to the patient service serial number. To the user.

It should be noted that the dynamic data analysis report generated by the real-time analysis processing module includes: full-range dynamic electrocardiogram data, dynamic blood pressure data, respiratory data, blood oxygen saturation data, invasive blood pressure data, intracranial pressure data, and exhalation. Carbon dioxide partial pressure data, body temperature data, fetal heart rate data, non-invasive cardiac output data, comprehensive analysis and calculation of respiratory mechanics data, waveform classification markers, waveform patterns, and their trend graphs, histograms, scatter plots, variability diagram.

The real-time data analysis report includes: abnormal ECG data, abnormal blood pressure data, abnormal respiratory data, abnormal blood oxygen saturation data, abnormal intracranial pressure data, abnormal end-tidal carbon dioxide partial pressure data, abnormal body temperature, abnormal fetal heart rate data Abnormal non-invasive cardiac output data, real-time computational analysis of abnormal respiratory mechanics data, waveform classification markers, abnormal waveform graphics, and trend graphs.

The dynamic data analysis report of the embodiment solves the problem that the vital sign monitoring device lacks the summary and analysis record of the electronic data in the application process, and at the same time, as the data basis of the clinical medical treatment, the clinical treatment effect and the patient state can be evaluated, and Adjust medical plan decision-making; real-time data analysis report solves the problem of lack of electronic abnormal data analysis report record when abnormal event occurs, and serves as the basis of abnormal event data to support rapid intervention of medical staff; the above-mentioned electronic data analysis report effectively improves medical treatment Quality and work efficiency reduce the labor intensity and workload of medical staff. Users can also send request instructions to the cloud platform, perform search queries, statistical analysis, and review and summarize clinical experience.

It should be noted that the method and system for processing vital signs based on the cloud platform can be implemented and executed on a public cloud or a private cloud, and can be implemented by using a cloud server, a database, and an application service system, and implemented in the form of a cluster. The functionality of the modules involved in the system.

The above method embodiments and system embodiments are based on the same or similar principles, and the similarities can be learned from each other and can achieve the same effect.

It will be understood by those skilled in the art that all or part of the process of implementing the above embodiments may be performed by a computer program instructing related hardware, and the program may be stored in a computer readable storage medium. The computer readable storage medium is a magnetic disk, an optical disk, a read-only storage memory, or a random storage memory.

The above description is only a preferred embodiment of the present application, but the scope of protection of the present application is not limited thereto, and any person skilled in the art can easily think of changes or within the technical scope disclosed in the present application. Replacement should be covered by the scope of this application.

Claims

A cloud platform-based vital sign data processing method, comprising the following steps:

Step S1: acquiring a plurality of user data in real time; the user data includes vital sign monitoring terminal device ID code, patient information, and vital sign data;

Step S2: pre-processing each acquired user data and performing unified encapsulation and storing the data into the vital sign database;

In step S3, the vital sign data in the vital sign database is read in real time, and the deep learning framework of the distributed parallel computing is used for analysis and screening processing, and the analysis screening result is obtained, and a data analysis report is generated.
The method of claim 1 wherein in step S2,

Binding the vital sign monitoring terminal device ID code and the patient information according to the system coding table rule, generating a patient service serial number, and maintaining bidirectional mapping conversion with the device ID code;

Parsing, classifying, and normalizing the data format of the received vital signs data, and retaining the original alarm event data flag of the device;

The patient business serial number and the preprocessed vital sign data are uniformly encapsulated and stored in the vital sign database.
The method according to claim 1 or 2, wherein the vital sign data in the vital sign database is read in real time, and the deep learning framework of the distributed parallel computing is used for analysis and screening processing, which is an online real-time data analysis. Processing method and deep learning framework based on Spark engine:

The Spark distributed parallel computing deep learning framework reads the vital sign data in the vital sign data. According to the set micro-batch processing interval, the Spark engine creates multiple tasks in parallel, triggering the Spark stream to divide the data into RDD data by type. The collection, while controlling the central model of the corresponding type, performs calculation processing on the type data.
The method according to claim 3, wherein the central model is divided into two categories, one class analyzes and calculates the morphology, rhythm and rate of the waveform data, and the other analyzes and calculates the amplitude of the numerical data. The central model includes a second-order differential calculation tool and/or logic analysis tool to calculate and analyze the shape, rhythm, rate, and value of vital sign data in real time, classify and mark the waveform, and statistically summarize the logarithm. Real-time analysis and screening exceeds the benchmark. Exception data.
The method according to claim 4, wherein when the central model calculation process finds abnormal data exceeding a set reference, analyzing the abnormal data feature, calculating the duration, and marking the abnormal data attribute.
The method according to claim 5, wherein when abnormal data is found, the abnormal data is generated into a real-time data analysis report, an abnormal event warning is issued to the user, and the real-time data analysis report is sent to the user.
The method according to claim 6, wherein the vital sign data after the analysis and screening process in the vital sign database is used to perform training optimization on each type of central model in real time to obtain a new central model of the type data.
The method according to claim 7, wherein the entire life vital sign data of each user subjected to the analysis screening process is integrated, a dynamic data analysis report is automatically generated, and the dynamic data analysis report is sent according to the patient service serial number. user.
A cloud platform-based vital sign data processing system, comprising: a cloud platform data communication subsystem and a cloud platform data support subsystem; the cloud platform data communication subsystem comprises a data communication module and a data preprocessing module; The cloud platform data support subsystem includes a message bus module, a data storage module, and a real-time analysis processing module;

The message bus module is configured to connect data transmission between the control data communication module, the data preprocessing module, the real-time analysis processing module, and the data storage module;

The data communication module is configured to receive a plurality of user data in real time, and is also used for data interaction with the user, and the received data is transmitted to the data preprocessing module, where the user data includes a vital sign monitoring terminal device ID. Coding, patient information and vital signs data;

The data pre-processing module includes a system coding table, configured to perform pre-processing on each acquired user data, and perform unified encapsulation and then store the data in the vital sign database;

The data storage module includes a vital sign database, a file database, a business information database, and a cache database for data storage and calling;

The real-time analysis processing module includes a deep learning framework for distributed parallel computing, which is used for real-time reading data in the vital sign database for analysis and screening processing, generating a data analysis report, and transmitting the analysis report to the user, and simultaneously depositing the file database.
The system of claim 9 wherein said data preprocessing module is operative to:

Binding the vital sign monitoring terminal device ID code and the patient information according to the system coding table rule, generating a service serial number, and maintaining bidirectional mapping conversion with the device ID code;

Parsing, classifying, and normalizing the data format of the received vital signs data, and retaining the original alarm event flag of the device;

The patient business serial number and the preprocessed vital sign data are uniformly encapsulated and stored in the vital sign database.
The system according to claim 9 or 10, wherein the real-time analysis processing module reads the vital sign data in the vital sign database in real time, adopts an online real-time data analysis and processing method, and a deep learning framework based on the Spark engine. Perform analytical screening processing:

The Spark distributed parallel computing deep learning framework reads the vital sign data in the vital sign data. According to the set micro-batch processing interval, the Spark engine creates multiple tasks in parallel, triggering the Spark stream to divide the data into RDD data by type. The collection, while controlling the central model of the corresponding type, performs calculation processing on the type data.
The system according to claim 11, wherein the central model is divided into two categories, one class analyzes and calculates the morphology, rhythm and rate of the waveform data; and another class analyzes the amplitude of the numerical data. The central model includes second-order differential calculation tools and/or logic analysis tools to calculate and analyze the shape, rhythm, rate, and value of vital sign data in real time, classify and mark the waveform, and statistically summarize the logarithm. Real-time analysis and screening exceeds the benchmark. Exception data.
The system according to claim 12, wherein the real-time analysis processing module analyzes the abnormal data feature, calculates the duration, and marks the abnormal data attribute when the central model calculation process finds abnormal data exceeding the set reference.
The system according to claim 13, wherein the real-time analysis processing module generates a real-time data analysis report of the abnormal data when the abnormal data is found, sends an abnormal event warning to the user, and sends the real-time data analysis report to the user. .
The system according to claim 14, wherein the real-time analysis processing module uses the vital sign data after the analysis and screening process in the vital sign database to perform training optimization for each type of central model in real time, and obtains new data of the type. Central model.
The system according to claim 15, wherein the real-time analysis processing module integrates the global vital sign data of each user subjected to the analysis screening process, automatically generates a dynamic data analysis report, and stores the data in a file database, according to The patient business serial number sends a dynamic data analysis report to the user.