See the medical health big data (below)
Author: Zhang Jiwu April 2017 published in the "World of Medical Devices" special issue (as cited article, please indicate the source, thank you.)
It is recommended to break through the following technical directions:
(1) Big Data Platform Technology
(2) Building a big data model
(3) Big data acquisition technology (establishing data standards, interconnection standards)
(4) Big data preprocessing technology (feature parameter extraction)
(5) Data mining and analysis, artificial intelligence
(5) Quality control of clinical process (quality control of critically ill single disease process, electronic disease of single disease type)
(7) Clinically assisted diagnosis and treatment (CDSS)
(8) Integrated management, interconnection, mutual recognition and sharing, analytical search, standard specification, privacy protection
The details are as follows:
3.1 Big Data Platform Technology
At present, new technologies related to big data are emerging one after another, and clinical medicine is also developing. We need to grasp the main line and top-level design of the overall structure, comprehensively consider various needs and trends, and improve the health of patients and improve the operational efficiency of the medical system. Carry out the necessary trade-offs and trade-offs to form a master plan design and build a sustainable development platform for new functional development and new technology applications.
Big Data Platform technology research is the infrastructure design and architecture planning for big data-based healthcare applications. Including the collection, transmission, storage, processing and management of big data in the field of acute and critical care, involving patient management, diagnosis and treatment, nursing monitoring, clinical quality, medical research, medical operations, etc. Application scenarios. Help more medical institutions to continuously improve the quality of medical care for critically ill patients, reduce overall medical costs, and improve the ability of society to respond to acute and severe diseases that pose a major threat to life.
The master plan of the medical big data platform includes the following levels:
1. Data integration: Data collection from a variety of data sources, including information systems, imaging equipment, bedside equipment, wearable devices and other digital systems.
2, process management: process-centric, based on these data integration, support clinical business process improvement, efficiency improvement and refined management.
3, data fusion: patient-centered, a variety of heterogeneous data fusion, reflecting the continuity of the diagnosis and treatment process.
4. Quality management: On the basis of data fusion, medical behavior monitoring, clinical quality measurement and clinical decision support are combined to improve the compliance of evidence-based guidelines and continuously improve medical quality.
5. Data mining: further explore the deep value of big data, and discover and verify new evidence-based evidence.
3.2 Big Data Model Establishment
The establishment of big data model is a difficult point. Medical experts understand that the application does not understand the data structure. The engineering and technical personnel understand that the data structure does not understand the clinical. The excellent data model needs to be closely combined with long-term cooperation efforts by clinical experts and engineering and technical experts.
The data model is an abstraction of information objects that need to be processed in specific areas of the real world (such as medicine, engineering, business), and related processing flows. At present, computers cannot automatically understand the laws of medicine, engineering or commerce in the real world. They can only assist humans in dealing with these fields according to pre-set patterns and procedures. They are suitable for processing with computers instead of human brains. That part of the data.
Therefore, the data model is the core and soul of computer information systems. The process of analyzing the operational laws of medicine, engineering or commerce in the real world and abstracting out the data model is very important for constructing information systems that serve the field. In the theoretical system of software engineering, there is a special field engineering direction, which is to study this analysis and abstract process (data model abstraction is one of them). Domain engineering will occupy an important position in large-scale software development; in general commercial software development, some domain engineering methodology will be selectively used according to the actual project input and product ratio to ensure the quality of the data model.
In the process of building a big data platform, the paper analyzes and abstracts different medical fields (first from the critically ill critical medical field), establishes a data model, and forms a methodology for continuous improvement of the data model. When the platform is installed in different hospitals, in different regions, and when new advances are made in the medical field (such as proposing new medical treatment routes and clinical guidelines), the model can be smoothly expanded to quickly adapt to changes in demand. Including the in-depth study of the acute and severe medical business model, the establishment of an integrated, scalable emergency critical EMR data model. Integrate heterogeneous data from existing information systems into CDRs as a big data base for quality control and research. Support multi-condition query to screen case samples, establish BigTable data mart, support export to SPSS, and facilitate scientific analysis.
3.3 Big data acquisition technology (establishing data standards, interconnection standards)
3.3.1 Clinical information system integration in the hospital
The big data platform needs to exchange data with various systems such as hospital HIS, LIS, RIS, PACS, EMR, etc., so as to realize the comprehensive integration of patient-centered cross-ward clinical information.
Here are some key technologies for clinical information integration:
2, unified dictionary directory technology
3, patient main index technology
3.3.2 bedside monitoring data integration in the whole hospital
At present, the large amount of patient data generated by bedside equipment (such as monitors, ventilators, and monitors) in most hospitals is free from electronic medical records, and the order of treatment and monitoring of critically ill patients is difficult to guarantee. The big data assets generated by the bedside equipment have not been effectively utilized. The manual readings and records occupy a large amount of time for the nurses. When the condition suddenly changes, it is difficult to find and promptly intervene in time. There are many types and models of bedside equipment, lack of objective statistics and effective supervision, which is not conducive to hospital asset management and effective use.
Establish an enterprise-level device data integration bus (Enterprise Medical Device Hub) as an information infrastructure with IoT features to serve electronic medical records and clinical information systems. At the same time, it can monitor the use of bedside equipment in real time and improve asset utilization.
The main functions of the hospital bedside data integration platform include:
1. Management of bedside equipment in the whole hospital, including patient identification, time synchronization, operation and maintenance management.
2, the entire hospital bedside equipment data collection, including protocol analysis, term mapping, data correction.
3, the hospital monitoring data management, including route cache, centralized storage, backup archive.
4. Patient monitoring data access in the hospital, including data query, data push, and data maintenance.
The construction of the bedside data integration platform of the whole hospital requires reference to the following two technical frameworks in the IHE international specification.
1, Patient Care Device (PCD) Domain
It is used to acquire, process, alarm, store and exchange data of ICU related equipment (such as monitors, ventilators, etc.).
2. IT Infrastructure (ITI) Domain
IT infrastructure such as time synchronization, patient ID management, log management, etc.
IHE (Integrate Healthcare Enterprise) is an international organization that is initiated by medical and information experts and is a cross-vendor and multi-disciplinary informatization field. She helps hospitals by developing and publishing technical frameworks, organizing integration testing and demonstration activities. A medical and health information system that integrates with manufacturers and has a smoother workflow.
3.4 Big Data Preprocessing Technique (Feature Parameter Extraction)
Doctors need to process large amounts of clinical data every day, and they must quickly analyze these massive amounts of data to form accurate judgments. According to statistics, each patient produces tens of thousands of objective physical signs, hundreds of subjective observations, and nearly 100 medical records in the intensive care unit every day, plus hundreds of test results and several copies from the auxiliary department. Check reports and hundreds of medical images. The constant changes in the condition of critically ill patients and the discovery of critical information in these rapidly generated massive data are challenging tasks. Once some key information is overlooked, it is likely to miss the best opportunity for intervention.
The processing and judgment of these clinical data relies mainly on the doctor's rich clinical experience. The traditional assistant decision-making software can provide calculation tools to help doctors sort out a large number of scattered vital signs data and test results, and comprehensively calculate some scores (such as APACHE II score, MODS score, etc.) to form a condition and organ function status. The overall judgment and evaluation; can also provide visual analysis tools, doctors according to different patient needs, select relevant physical signs and monitoring items, laboratory indicators, medication, draw trend charts, for circulation, breathing, metabolism, organs Detailed analysis and evaluation of functions and therapeutic effects.
With the rapid development of medicine, various new scores and evaluation indicators, various analysis charts, and constantly being discovered and promoted, doctors are once again submerged into the ocean of data. Therefore, these traditional tools are far from meeting the requirements. Modern medical data processing systems should have the following three characteristics:
The first is to provide doctors with high-value information. The information related to the patient's current treatment and intervention goals, the patient's sudden critical value and alarm value, the patient's allergy and medication contraindications, etc., should be displayed at the first time. Information that is not relevant to conventional concerns can provide comprehensive query retrieval for clinical research.
The second is to provide doctors with integrated information. For a variety of clinical topics, such as circulation, respiration, kidney function, liver function, internal environment, nutrition and metabolism, infection control, DIC, etc., doctors can classify and process information according to clinical needs, rather than messy The ground is piled up to help doctors form an overall judgment of the condition.
The third is to provide doctors with real-time updates. The treatment of acute and severe patients, time is life. The information system must be able to process massive amounts of data in real time. In particular, the calculation of various critical scores and disease assessment indicators manually must be automated, push the calculation results in real time, and obtain the next intervention from the knowledge base. Reminder.
3.5 Data Mining and Analysis, Artificial Intelligence
Further explore the deep value of big data, discover and validate new evidence-based evidence, and forward-looking research in the field of medical big data applications. The current conditions are relatively comparative, and the direction of achievement transformation and product incubation potential is as follows:
3.6 Clinically assisted diagnosis and treatment (CDSS)
The clinical decision support system, CDSS, is a medical information technology application system based on human-computer interaction. It aims to provide clinical decision support (CDS) for doctors and other health practitioners, and assists in clinical decision making through data and models. The concept of the “Clinical Decision Support System” is still being updated. The current mainstream work definition is proposed by Robert Hayward of the Center for Health Evidence: “Linking clinical observations to medical knowledge, thus affecting doctors’ choice of programs and improving The quality and effectiveness of medical services." Take the critical illness as an example:
3.6.1 Emergency clinical decision support
The existing emergency department system of the Big Data Hospital takes the doctor's advice as the core process and meets the requirements of the thick line management of the emergency clinical process. Through the application of clinical decision support tools, from pre-examination and triage management to ward management, process efficiency can be further optimized and improved; from rescue monitoring to green channel, process quality can be refined and controlled.
First, the pre-screening triage is the emergency window and the starting point for process optimization. Using the "three districts and four levels" norms, the non-emergency patients are managed well. The patient's condition changes are timely discovery and rapid circulation, in order to give full play to the value of emergency medical resources, so that the emergency department is really "urgent". The intelligent triage system automatically collects patient's vital signs data, and provides a recommendation for disease grading and triage diversion with the help of a scalable emergency symptom knowledge base to help nurses complete the triage quickly and accurately.
Secondly, emergency medical thinking and clinical intervention often begin with physical symptoms rather than a clear pathological diagnosis. From pre-examination and triage, to rescue and observation, surgery and monitoring, based on the diagnosis and treatment path and timeline of the emergency department, the medical staff is provided with appropriate quality control reminders and decision support in the clinical process, reflecting the emergency clinical process management. Refined needs. According to the emergency route, the doctor's meal package is recommended in real time in the diagnosis and treatment session, and it can be quickly released through touch screen selection. According to the critical value knowledge base, timely reminders when emergencies occur, help doctors and nurses prevent medical errors, push the intervention checklist, and improve compliance.
In addition, the ED Dashboard of the emergency department has been used as a method of big data utilization at home and abroad. By installing a large-screen smart TV that supports touch at the appropriate location in the ward, the patient information in each area of the emergency department is displayed in real time, and the current diagnosis and treatment status and circulation status of the patient are prompted by different colors and icons to realize visual management of the emergency department. On this basis, automatically generate and update departmental statistical reports at any time, without having to wait until the shift to manual statistics, reducing the workload of medical staff, improving the real-time information, facilitating timely adjustment of medical resource allocation, calmly responding to sudden groups event.
3.6.2 Critical clinical decision support
At present, most hospitals, patients' disease assessment and medical decision-making, doctors need to consult information from multiple sources, the data has not been extracted, organized and presented according to the needs of clinical decision-making and the logic of doctors' thinking.
Design patient information summary, according to the needs of organ function support, according to the classification of multiple organ functions and physiological systems, the mass data is classified and displayed. Help doctors build a general understanding of the patient's condition and assist doctors in making rapid clinical decisions.
On the big data platform, the data visualization tool is used to assist the disease analysis, which facilitates the discussion of the condition of the house and the shift; the system can maintain the data visualization template required by the specialist, on the same time axis, for multiple organ functions, multiple treatment plans, etc. The comparison was made to review the factors related to the rescue process and changes in the condition parameters.
Quantitative scoring tools are used to assess the condition to assist in medical decision-making; the system can customize the scoring algorithm according to the latest evidence-based basis, automatically extract and select the data required for the scoring, greatly improving the efficiency and accuracy of the disease assessment. To enhance the willingness of doctors to use.
Conditional departments can further strengthen the implementation of severe clustering therapy through big data-based process quality control tools, including multi-parameter-based disease changes and intervention reminders, rapid connection of related medical orders, closed-loop medical management, and cluster treatment Time management and quality indicator calculations.
3.7 Integrated management, interconnection, mutual recognition and sharing, analysis and retrieval, standard specifications, privacy protection
3.7.1 Integrated hospital integrated management
The integrated management of the patient's diagnosis and treatment process needs to start from the emergency department and go through the whole process of follow-up treatment and monitoring. Including the connection between the emergency department and the specialist department, the operation room, the connection between the general ward and the ICU, and the multidisciplinary consultation process. It is convenient for the medical department to meet the unified management requirements for the quality (structural quality, process quality, terminal quality) and operation (person, finance, and materials) of critically ill and critically ill hospitals.
3.7.2 Data Interconnection Standard System
Data interconnection and interoperability, adopting an application-oriented, open and gradual standardization strategy. First, rely on the underlying standards and typical customer application requirements to build a scalable data model. As the application expands, the latest standards are adopted in due course. Actively participate in and promote the evolution of standardization in applications with standardized vacancies.
3.7.3 Data Security and Privacy Protection
1. Data security: It is necessary to establish a massive database to centrally manage medical health data of a large number of people. Data leakage may reveal patient privacy and even endanger public safety. Construction of relevant information systems in accordance with national information security laws and regulations. All data access is logged and the audit log is kept for tracking.
2. Data quality: In the process of data collection, transmission, processing and management, errors may be introduced due to human or objective reasons. The reduction of data quality will directly affect the credibility and application effect of the analysis results. Standardization and semantic processing techniques are used to standardize and verify, and an integrated data governance system is established.
3. Patient privacy: Patients are the owners of their medical data. As privacy awareness increases, big data platforms may face challenges when collecting data. Under the relevant laws and regulations, the use of open, transparent, effective data desensitization, encryption and other information protection mechanisms.
(2) Typical big data application scenarios
We propose application scenarios for big data (1) telemedicine
(2) Regional cooperation
(3) Quality resources drive remote areas
(4) Precision medicine
Tele ICU is a telemedicine form that is booming internationally. It is usually based on regional core hospitals, establishing an ICU service center, using computer, network, camera, microphone and information highly integrated clinical information system software, etc. Medical staff conduct real-time audio and video calls and monitoring data transmission, enabling high-quality critical medical resources to cross geographical barriers and serve more patients.
On the Tele ICU platform, the patient is looked after by a bedside medical staff and a remote specialist. Experts from the ICU Service Center conduct virtual rounds at intervals to guide bedside medical staff to conduct medical treatment and care; in case of emergency, bedside medical staff call remote experts for timely assistance. Some foreign practices have proved that this model can shorten the hospitalization time by 20% to 30%, and more importantly, it can greatly improve the survival rate of patients.
During the “Thirteenth Five-Year Plan” period in China, the four major projects have been identified in the medical and health service plan, and it is clearly pointed out that “it is important to improve the level of medical services in county-level hospitals for children and critically ill medicine”. Drawing on foreign advanced experience and combining with domestic actual needs, carrying out the practice of Tele ICU, giving full play to the value of high-quality critical medical resources, and promoting the improvement of regional critical medical services will be of great significance.
4.2 Regional synergy
In order to improve the level of emergency treatment in the region, big data platforms need to open up barriers between medical and health institutions. By constructing a regional information platform for the treatment of critically ill and critically ill patients, it is possible to integrate social first aid, 120 first aid, in-hospital emergency and critical care related resources in the region, aiming at emergency and critically integrated closed-loop treatment to achieve first aid, emergency, and ICU integration. Based on the latest mobile internet technology, through the APP and WeChat platform, the pre- and post-hospital processes are opened to achieve a complete closed loop. Form synergistic amplification effects, optimize resource utilization, and create better social and economic benefits.
Due to the large number of relevant parties involved in regional coordination, coordination is difficult. In the beginning, you can choose a relatively simple scenario to pilot and gradually promote. Regionalized emergency triage management is a relatively easy to start scenario. Through the sharing of triage data in cooperative hospitals in the region, unified supervision and coordination of medical resources can be carried out to improve the utilization efficiency of regional emergency resources. The accumulated data can also be used for emergency traffic forecasting and process optimization.
4.3 Quality resources drive remote areas
Establish a portal to expand the radiation of quality medical resources. The website section can include medical forums, case discussions, information release, hotspot analysis, expert reviews, expert consultation, continuing medical education, training and teaching, walking into the community, supervision and inspection, emergency consultation, reporting of key diseases, patient follow-up, etc. The disease report needs to provide a formatted, structured report, and can be modified to facilitate post-analysis statistics. The specific function realization is customized according to actual needs.
With the booming mobile Internet, mobile applications are no longer limited to gaming and entertainment. Professional mobile applications have emerged in all walks of life to provide deep customized information services to the public. Many domestic hospitals have also begun to provide services such as information release, report inquiry, and appointment registration through mobile phone applications, extending the reach of hospital information systems to the fingertips of patients, which not only improves patient satisfaction, but also enhances the hospital brand image. This new medical service model has become the trend of the mobile medical era.
With the rapid development of the WeChat platform, the value of big data can be further expanded based on the mobile services provided by medical staff and patients based on the WeChat public platform. It can help hospitals establish a connection between emergency medical staff, patients and hospital IT systems. With the enterprise number, the hospital can quickly realize high-quality enterprise mobile light applications and realize the mobilization of production, management, collaboration and operation. As the enterprise IT mobile solution, the enterprise number has obvious advantages compared with the enterprise's own development of the APP, specifically:
1) Fast mobile office. After opening the enterprise number, you can directly use the basic capabilities of WeChat and enterprise number to strengthen communication and collaboration among employees, enhance cultural construction, announcement notice, knowledge management, and quickly realize the mobilization of enterprise applications;
2) Development costs are low. It only needs to interface with the existing system according to the standard API of the enterprise number; zero threshold is used. The user WeChat scan code can be used for attention. When playing WeChat, the enterprise number message can be processed at will, and it can be used smoothly without learning. Through the emergency and heavy WeChat enterprise number, it is possible to establish an emergency first aid expert, an emergency emergency medical staff and a patient's connection with each other, and a closed-loop service for the first-aid full-process process can be realized.
4.4 Precision Medicine
Precision medicine is an emerging approach to disease prevention and treatment that takes into account differences in personal genetics, environment and lifestyle. Applying precision medicine to the research and diagnosis of high altitude disease will be of great significance and is therefore an important part of the big data platform construction project.
The precise medical research that starts with the current genetic testing is to find the corresponding clinical diagnosis information after reading the gene fragment, and the clinical information big data we have established has the information of multi-module and time axis, and correspondingly extracts the blood of the patient. Genetic testing and other marker testing have higher research efficiency.
The purpose of MDT is to provide patients with a personalized medical plan that is the most effective, has the least side effects, and has the best quality of life on the basis of multidisciplinary discussion and argumentation. MDT requires multidisciplinary team collaboration, so it requires a multifaceted clinical data support. Using the big data platform, all Asian specialists can access all the patient's medical data (such as medical records, medications, laboratory results, pathology reports, video images, surgical procedures, endoscopic images, genetic counseling reports, etc.) anytime, anywhere.
The bright future brought by big data is constantly being described and introduced. Just as the maturity of genetic diagnosis technology is exaggerated, it should be said that these beautiful futures exist and deserve and need our hard work, and the basis of struggle is to find it on the ground. A good entry point, correctly construct the information foundation needed for big data, and let each step of the effort be the cornerstone of the next step, and also produce a good realistic social application effect, form a virtuous circle, and promote big data. The development of the cause.