In the face of the new tens of billions of new markets, how do the electronic medical records “diggers” show their magic?

The artificial intelligence in the medical circle is so big that CDSS (Clinical Decision Support System) is destined not to be a supporting role. After experiencing the re-engineering of the expert rule system in the early years, CDSS urgently needs to use the concept of artificial intelligence to bring order out of chaos. The hospital informationization construction policy with electronic medical records as the core has just sounded this "counter-attack horn".

Although there are policy blessings, the landing of CDSS in hospitals is by no means smooth sailing. The high clinical requirements of the hospital, the different degrees of acceptance of medical AI by doctors, and the slow progress in the construction of hospital data warehouses restrict the rapid development of the industry. Under these conditions, domestic companies such as Science and Technology, Kangfuzi, etc., took the lead in playing the role of “digger” with innovative technologies such as medical record structure, “plug-in” components and knowledge maps.

In September 2018, the State Health and Health Commission Medical and Medical Administration issued the "Notice on Further Promoting the Informatization Construction of Medical Institutions with Electronic Medical Records as the Core" (hereinafter referred to as the "Notice"). Because it involves the core issues of hospital ratings and compliance, it is highly concerned by hospitals at all levels.

The "Notice" clearly requires: By 2020, the tertiary hospitals must reach the graded evaluation level 4 or above, that is, the hospitals can realize the information sharing of the whole hospital and have the medical decision support function. This is enough to see that medical decision support is becoming an important part of hospital electronic medical record construction.

In fact, medical decision-making has received attention, as early as the beginning of this year. On March 23, at the China Hospital Information Network Conference (2018CHINC), experts from the National Institute of Health and Family Planning Hospital Management Institute conducted an evaluation on the methods and standards for the evaluation of functional application of electronic medical record systems (2018 revised draft). Interpretation.

Comparison of the difficulty of the old and new versions

From the comparison of the basic content of the old and new versions, it is not difficult to find that after the standard adjustment, the medical decision support has been upgraded in overall. The following are the specific content of different levels of medical decision support:

It can be seen that primary clinical decision-making is mainly for reminders of simple conditions, such as rational use of drugs, realization of drug incompatibility reminders, providing a knowledge base for diagnosis and treatment; intermediate decision support can handle relatively complex conditions, such as: mentioned in previous history Disease, diagnosis, age, gender and other conditions for drug use taboo reminders, provide evidence-based medicine based knowledge; advanced decision support, through big data processing, machine learning, based on guidelines, evidence-based medical knowledge systems and real-world data , for early warning of clinical behavior, prognosis analysis, recommendation of similar medical records, etc.

With a clear definition, hospitals with different levels of electronic medical record construction can flexibly consider the reporting level according to their own circumstances.

According to the relevant data in the “2017-2018 China Hospital Informationization Survey Report” issued by CHIMA of the China Hospital Association Information Management Committee, it can be seen that the CDSS system construction of domestic tertiary hospitals and tertiary hospitals is still in its infancy. Among them, the proportion of tertiary hospital construction is 20.15%, and that of hospitals below grade 3 is only 8.6%. It is still far away from the popularization of the word.

The arterial network has investigated the market space of CDSS in the article "How to Choose the Two Billions of Markets for Medical Informationization in the Electronic Medical Record New Deal and How to Choose the Standard for Hospitals", mainly judged by the number of hospitals and the unit price of the products.

In terms of unit price, according to industry insiders, the electronic medical record grading evaluation requires more than four levels of clinical decision support. The clinical decision support function is not equivalent to the knowledge base. It is an application of artificial intelligence, such as being able to deal with non-institutionalization. Multi-dimensional clinical data, which can automatically predict the list of diseases to be identified based on the patient's clinical data and automatically recommend personalized treatment plans. Advanced decision support not only covers all medical sessions, but also requires an evidence-based, real-time updated medical knowledge base. The authority of the knowledge base, the ability to process unstructured data, and the ability to predict models are the reasons for the difference in CDSS prices.

Many times, CDSS is packaged into a large project and not sold separately. If sold separately, the price of CDSS (non-pure knowledge base) is usually between 500,000 yuan and 1 million yuan. Therefore, the average price of CDSS is estimated to be about 750,000 yuan. Combined with the National Health and Health Commission Statistical Information Center released at the end of June 2018, the number of national medical and health institutions (the number of tertiary hospitals is 2,439), the arterial network estimates that by 2020, the market space of CDSS in tertiary hospitals is about 1.9 billion yuan. .

However, considering that CDSS has a division between general and single disease, if a company launches a specialized disease CDSS for different clinical departments, then under the original market scale, it is almost a multiplication of 10 to be a potential market. Space, so to see, the market space of CDSS is close to 20 billion yuan.

Although there are policy blessings in the implementation of CDSS in hospitals, it is bound to be hindered by market environment and technical problems.

From the perspective of market environment and cognition, the overall level of informatization construction of domestic hospitals is relatively low. If there is no good information platform based on it, the clinical assistant decision system is difficult to advance. In the process of product docking, the first problem that CDSS needs to solve is the problem of matching with different vendors' systems. The implementation cycle may be longer or shorter depending on the strength of the hospital and the degree of cooperation of the manufacturers. The progress of hospital data warehouse construction also affects and restricts the full process application of CDSS system.

For example, when the CDSS analyzes whether the patient's electrocardiogram is completed within 10 minutes of the visit, it is necessary to retrieve the emergency registration time and the completion time of the electrophysiological examination outside the electronic medical record recording system, and whether there is a cross-system data interface. And whether the doctor entered as required will affect the auto-complete of this analysis.

In addition, CDSS products involve more clinical departments, and doctors have different levels of acceptance of medical AI. From landing to practical application, an education-to-recognition process is needed. In particular, some tools that can save work and improve efficiency, most doctors may first reject and then selectively try, which to some extent lengthens the verification cycle of CDSS products.

Due to the high clinical requirements of the hospital and the high labor intensity of the doctor, the CDSS must be able to help the doctor effectively and integrate the product with the clinical, rather than rating for rating. This process is difficult to accomplish overnight.

There is also a contradiction in the needs of clinicians and administrators for CDSS. Managers often prefer strict control of the entire process, while clinicians focus more on saving work and reducing the burden. In some cases, the needs of the two will oppose each other.

From the perspective of product development, for an AI-assisted diagnosis and treatment system, it is the biggest difficulty to digest medical knowledge like a clinician.

Taking the well-known medical AI company Huiwei Technology as an example, on the basis of the Mayo disease diagnosis and treatment path, Huiyi Technology turned a piece of clinical knowledge, guidelines and specifications into a "computer-readable and understandable" one. Set a set of clinical ideas, and use the clinical ideas to extract logic, and work with the cooperative hospital to improve the knowledge base of specialist and specialized diseases.

Hui each technology obtains real information from the medical record data, continuously trains the AI ​​“brain”, perfects the diagnostic model, and turns the experts' thinking and experience into a diagnosis and treatment path with hospital characteristics and advantages. This method and logic is the core value of the current technology.

It is reported that Hui has more than 40 large-scale top three hospitals in cooperation with CDSS. Among them, 6 hospitals passed the electronic medical record application level evaluation grade 5 and 6 level, 4 passed the HIMSS level 7 review, and 2 passed the interoperability level 5 B evaluation. In addition, it also includes the First Affiliated Hospital of Zhejiang University, Xiangya Hospital of Central South University, the Second Affiliated Hospital of Zhejiang University Medical College, China-Japan Friendship Hospital, and Jiangsu Provincial People's Hospital.

Hui CDSS is a clinical-centered intelligent CDSS. In the view of Zhang Qi, CEO of Hui Technology, the essence of CDSS application is to help doctors to solve the consistency and standardization of clinical diagnosis and treatment, and thus improve clinical quality. To achieve this, first consider whether the vendor's knowledge base is sufficiently leading and authoritative. The introduction of Mayo's complete knowledge system has laid the foundation for every CDSS and is one of the reasons why it can be recognized by clinicians.

The two-way blessing of policies and markets does not mean that companies have already done their homework and have enough technical reserves to deal with them. To truly meet the decision-making requirements of hospital clinical departments, companies need to change the "old routines" of the past expert rule system, and achieve a qualitative leap with new technologies such as deep learning and big data. In this regard, some companies have already taken the lead...

Kang Fuzi CDSS

In the first half of 2017, a well-known information technology manufacturer in China should promote the six-level electronic medical record review project of a top three hospital, so it is necessary to provide technical support. After learning about this situation, Kang Fuzi, a well-known medical AI company in China, successfully developed a new CDSS product after half a year of trial. The product has developed 30 interfaces to help the project implement smoothly.

In the early 2018, Kang Fuzi re-examined and examined the needs of clinical assistance, and officially launched Kangfuzi clinical decision-making products at the end of June 2018. Unlike traditional expert rule systems, this is a clinically assisted decision-making product driven by artificial intelligence.

According to the arterial network, the entire auxiliary decision-making system highlights the practical and expandable capabilities. In addition to supporting the hospital to meet various informationization rating requirements, the product can achieve clinical application functions such as auxiliary diagnosis, rational drug use, risk reminder, medical advice, knowledge push, medical record quality control, etc., and also focus on the easy implementation of hospitals and information manufacturers. The scalable and convenient management requirements have created a CDSS knowledge management platform and a CDSS data analysis platform.

Not only that, Kang Fuzi uses the "plug-in" component on-line support in product design concepts, and some specialized components such as "institutional sense", "VTE", "single disease diagnosis", "sickness classification", etc. The results of the medical records standardized by Kang Fuzi are input, the custom template style is output, and a variety of optional algorithm-driven models are quickly developed and launched.

At the level of clinical aided decision-making, essentially two things are handled:

1) Engine level: Real-time recognition and understanding of the current doctor's clinical content, giving decision-making results; Kang Fuzi uses leading artificial intelligence technology to achieve the results that are beyond the reach of expert rule engine and traditional machine learning methods.

2) Knowledge base level: Explain based on decision results and provide doctors with authoritative static knowledge base content. Kangfuzi cooperates with People's Health Publishing House, the largest and most authoritative medical publishing institution in China. At the same time, the products also support the docking with the mainstream knowledge base in the market, and also provide the editing platform inside the hospital. The experience of the doctors in this hospital can also be added to the platform.

The auxiliary decision engine consists of 2 major parts:

1) Semantic understanding engine; 2) Inference decision engine.

Semantic understanding: The information recorded by the medical staff is a large segment of continuous text. If the computer wants to process and understand the text, the first thing to do is to extract the knowledge features. This work, Kang Fuzi through the medical record structure engine to achieve. The Kangfuzi medical record structured engine can be structured and analyzed from more than 200 dimensions of the whole disease medical record. It has been applied to the information systems of dozens of large tertiary hospitals in China, including the top two hospitals in China, 301 Hospital and Beijing. Xie He Hospital.

If there is no medical record structure engine, companies can only use the medical record text to match some dictionary glossaries to approximate similar functions. However, for the medical records written in free text, there is a very rich way of expression, and the way of keyword matching basically does not work.

Inference Decision Engine: After solving the feature extraction required for computer calculations, the rest is passed to the inference decision engine for analysis. Medicine is a very complex subject that poses enormous challenges for reasoning decisions. Kang Fuzi's inferential decision-making uses a set of algorithms developed by himself, and is called the deep Bayesian network within the company. The network as a whole is based on Bayesian inference, but when calculating the probability of joint distribution, Kang Fuzi made a great innovation.

In computational mathematics, Kang Fuzi made many assumptions for simplifying the calculation, such as introducing the naive Bayesian hypothesis, introducing the Markov hypothesis, and so on. For a patient, “cough” and “fever” are not independent. In the deep Bayesian network, Kang Fuzi uses layering to deal with the relationship between these features.

In addition, the entire decision algorithm runs on top of the knowledge map. With the constraint and guidance of the knowledge map, we can ensure that the reasoning is carried out in the correct direction without low-level errors.

Take a real medical record as an example:

Male, 50 years old, was admitted to hospital due to intermittent episodes of abdominal pain, jaundice, and fever for 3 months. The patient had no obvious cause 3 months ago. Sudden upper abdominal pain after meal, radiation to the back and shoulders, more severe, with fever around 38 °C, the next day found sclera, yellow skin stains, no nausea, vomiting. Symptoms were relieved after the application of antibiotics and choleretic drugs in local hospitals. In the following 2 months, there was a similar episode 2 times, still anti-inflammatory, choleretic, liver protection treatment, and the symptoms were alleviated. In order to further clarify the diagnosis and treatment to our hospital. Half a year ago, he underwent cholecystectomy for "chronic cholecystitis and gallstones." Smoke-free wine hobby, no history of hepatitis and tuberculosis.

The semantic understanding engine needs to extract knowledge such as: gender male, age 50 years old, symptomatic abdominal pain (March, intermittent episode, back radiation, shoulder radiation, severe pain), negative symptoms, nausea, ..., historical treatment plan " Cholecystectomy" and so on.

First, the decision engine will get possible diagnosis based on information such as symptoms and signs, such as gallstones and common bile duct stones. After that, the system will further analyze the historical treatment plan and find that the patient has undergone a cholecystectomy, thereby eliminating the suspected diagnosis of "choledocholithiasis."

Less than three months after its release, the product has been launched in nearly 30 top three hospitals across the country. Such as: Nanjing Gulou Hospital, Shengjing Hospital and so on. Not only that, Kangfuzi CDSS also served the province's 10,000 village clinics through a provincial village medical platform, and signed a deep cooperation with a district health committee in Beijing, four district hospitals and more than 100 community villagers. The CDSS system of Kokangzi is also in line.

Hui per CDSS

According to the arterial network, in 2018, in addition to helping a large number of hospitals to achieve rating requirements, Huizhou Technology is more helping to achieve clinical value. The clinical value is reflected in the important features that distinguish each other from other CDSS vendors. Taking the AI ​​quality control project of each technology and Xuanwu Hospital as an example, the company and the Department of Neurology of Xuanwu Hospital jointly combed 11 quality control points for the treatment of acute cerebral infarction.

Such as: NIHSS score, the current NIHSS score results are recorded in each course of the disease; antithrombotic therapy within 48 hours, antithrombotic therapy should be started within 48 hours after admission, and the medical record text is monitored in real time. If there is a treatment defect, Hui CDSS will Remind doctors to improve treatment on the medical record page (admission record, first day course, daily course of illness, discharge summary).

Application benefits per CDSS quality control reminded less than one month, Xuanwu Hospital neurology department stroke 11 medical treatment norm rate increased from 70.19% to 93.85%, an increase of 33.7%. Among them, the intensive intensive lipid-lowering treatment in the hospital increased from 72.73% to 92.16%, and the discharge antithrombotic treatment increased from 65% to 100%. Song Haiqing, deputy director of the Department of Neurology, Xuanwu Hospital, said that the improvement of the treatment standard rate will greatly improve the patient's prognosis and reduce the recurrence rate.

From clinically assisted diagnosis and treatment to medical record quality control, Hui CDSS effectively improved the consistency and standardization of clinical diagnosis and treatment, and created a quality management closed loop from pre-diagnosis medical consultation, medical record examination, and post-diagnosis data reporting.

In this regard, Zhang Qi said: "From the application point of view to evaluate CDSS actually two points: the first is accurate, this is very understandable. The second is real-time, fully integrated into the clinical diagnosis and treatment process, real-time interaction, rather than jumping out of the system and then go Query, or find errors afterwards, and trace back. Only by achieving these two points can we have a real impact on clinical quality."

He also gave a practical case to illustrate: a patient admitted to the hospital due to acute cerebral infarction, Hui CDS through the real-time identification of the subject, current medical history, treatment and medication, and other medical records, according to the guidelines, recommended for doctors Treatment options such as automatic intravenous thrombolysis, endovascular intervention, and antiplatelet therapy. However, when the doctor continues to supplement the patient's medical history and writes that “the patient had a head trauma two months ago,” each CDSS will judge the patient's contraindications for intravenous thrombolytic therapy in real time, and tell the doctor that the patient should not be treated. At the same time, the recommended treatment plan will be updated automatically, and the treatment of intravenous thrombolysis will not appear again.

Recommended routine treatment plan for acute cerebral infarction

Identify patient contraindications in real time and recommend precise personalized treatments

In combination with the pre-existing arterial network, it is believed that CDSS must truly help doctors and integrate well with clinical practice in order to truly achieve the tens of billions of markets. In this regard, CDSS needs to meet at least the following conditions:

1. The progress of the hospital data warehouse construction is fast enough;

2. Solve the problem that CDSS matches different informatization vendor systems;

3. Meet the different types of needs of clinicians and hospital administrators;

4. Use the clinical ideas to extract logic, and work with doctors to improve the knowledge base of specialists and specialized diseases;

5. Through the structure of the medical record, try to ensure the accessibility of clinical data;

6. Cooperate with hospital departments with sufficient authority to ensure the authority of the decision-making knowledge base;

In other words, although there is a large enough market space, the foundation of the enterprise and the hospital itself needs to be improved. If you are able to grasp your own defects and are willing to improve at the right time, you can take the first step and taste the sweetness of CDSS.

Mulch Film

Mulch Films are a revolution in the agriculture industry as it presents an economical and environment friendly protective solution for crops and soil. It not only maintains the quality of soil but also aids in the faster growth of plants. By reflecting light, it ensures control on growth of weeds as photosynthesis process will not be carried out.

Mulch Film,Sliver Black Mulching Film Sheet,Mulch Film For Greenhouse,Polyethylene Mulch Film

JIANGSU SKYPLAN GREENHOUSE TECHNOLOGY CO.,LTD , https://www.skyplantgreenhouse.com