Release date: 2016-06-13
In recent years, medical big data has been hot. So what is medical big data?
Information related to health care can be classified as medical big data. Data sources can be medical and scientific research institutions, and can also be derived from individuals and even governments. Its application scenarios are very extensive, such as drug research and development, precision medicine, medical insurance control fees, personal health management, personalized medicine, and even graded diagnosis and treatment, telemedicine. From the experience of the ether, there are some common pain points in the application of China's medical big data.
Data island
First of all, from the institutional level, under the leadership of government policies, more than 70% of hospitals have realized medical informationization, and the infrastructure construction level has already begun to bear fruit.
However, less than 3% of hospitals currently achieve data interoperability: the number of traditional system vendors is large. According to incomplete statistics, there are currently more than 500 domestic HIS system vendors and more than 200 PACS system vendors. The manufacturers design and implement systems for hospitals. Different versions will appear in different systems. It takes 2-6 months to aggregate a system using traditional methods, which is time-consuming and labor-intensive for hospital data collection and aggregation. Although some startups with mobile medical and telemedicine will deploy new systems to obtain data, the access of many Internet medical products has created an additional burden for both doctors and patients.
Secondly, from a personal point of view, there are two main channels for direct access to personal data: wearable devices or mobile devices, and home detection to obtain user data. However, these two acquisition channels are in the market introduction period for the users, the stock market is minimal, and the cost of importing and persuading is huge.
For example, in 2016, China's wearable devices only had a market size of 20 billion. From the perspective of usage habits, even in the US 2015 data, only 3% of Americans have wearable devices. In addition, users' habits and willingness are also the enemy of data acquisition. According to a survey conducted by NMD Group, a US market research company in 2015, more than 40% of sports bracelet users will abandon the device within six months and cannot obtain periodicity. Continuous data.
On-site detection of user data is currently common in blood testing and genetic testing, from the service cut-through, the future through data accumulation applied to scientific research, medical supplies research and development, user personal health management and other scenarios, however, due to the high cost of such services, low user acceptance Data acquisition is also subject to many restrictions.
In fact, in response to the pain points of institutional data proposed above, some startup companies have begun to integrate data to create data middleware. However, due to barriers between medical institutions, detection levels, reagent selection, etc., data caliber and related indicators cannot be used directly. At the personal data source level, different data analysis vendors will also cause problems with non-uniform standards.
At the specific implementation level, how to integrate these scarce personal data, how to define the topics in different fields and topics, and the coordination of statistical caliber will become an important part of the application. In the business environment, this will also face the distribution of benefits, how to work together and ethical issues.
Missing tool
Medical big data can play a role, but also depends on the actual application and interpretation, how to correctly interpret the data, and let the data guide the clinical, personal health, public health, medical insurance and other fields is an essential part of the final landing of medical big data.
In any vertical application scenario, "know-how" is an important part. Analysts in the field of good medical data should also have the basic knowledge of vertical domain expertise and data analysis to effectively integrate business and data. However, as the most suitable composite subject biostatistics is still in its infancy in China, the number of people able to land in the vertical field is increasingly restricted due to the lack of overall data analysis talents in the data market.
The second is the tool-level short board. When I do data work, I have a qualitative discovery: SAS, R, MATLAB and other relatively high threshold tools are widely used in the medical industry, and the limitations of the discipline lead to high tools. Requirements, however, visual reports, intuitive presentation methods, can effectively improve the efficiency of analysis, "let professional people do professional things."
Privacy boundary
After data analysis, how to land, how to reflect the value of patients is also a problem of current data applications. Some of the government's data will be pushed back into the formulation of policies, and the final landing and verification cycle of drug R&D data will be longer. How the value of the data is reflected in the individual patient, can it be the output that the individual user can perceive, and how the actual role is evaluated and measured is the problem faced by medical big data.
In addition, in the process of medical data collection, processing and application, the protection of user privacy is also very important. Based on the attributes of patients, such as detection indicators and results, the user behavior can be directly predicted, such as the leakage of personal information. False sales and other issues, therefore, attention to how data is applied to the individual level, information leakage is also one of the focus of attention.
Third-party data vendors enter medical institutions and enable data to be effectively applied. They also face pressure to promote: the highly resource-oriented medical industry, industries, institutions, institutions, and geographies face numerous barriers, even listed companies are still unable to Breaking into certain areas, the difficulty of business development and landing is extremely difficult. On the one hand, it is difficult to obtain an effective data source, and on the other hand, there are not enough scenes to train its own model.
In this case, some entrepreneurs have taken a two-legged approach: for example, the “Establishing Yikangâ€, which is successful in financing the Ether, which integrates its own resources and continuously expands its business with the Health Planning Commission and the hospital; In terms of effectively utilizing the network of commercial organizations, it breaks geographical restrictions and rapidly realizes its own growth. Or, such as "permanent number", directly through product innovation, obtain support from national benchmarking institutions, and bind for promotion.
Source: Bio-Exploration
Yancheng Rongtai Labware Co.,Ltd , https://www.shtestlab.com