Establish search barriers based on medical knowledge map

With the development of the Internet, the content of network data is growing explosively.
because of the large-scale, heterogeneic and loose organizational structure of Internet content, it challenges people to obtain information and knowledge effectively.
Graph lays the foundation for knowledge-based organization and intelligent applications in the Internet age with its powerful semantic processing and open organization capabilities.
to realize medical intelligence, it is necessary to construct medical knowledge map and meet the application needs of different levels of difficulty in medical field.
01 Medical Information Retrieval When it is necessary to retrieve target information from a large amount of medical information, some key objectives contained in the result information and their association can be displayed in the search results.
02 Robotics Medical Question and Answer, in order for robots to better "understand medicine", need to use a large amount of medical data to train the model.
this way, when a robot needs to answer a new question based on an existing "question-answer" list, the algorithmic model can calculate the existing questions that are closest to the new question and return the answer to the question.
knowledge map is often used to help define some key concepts in the field of the problem and the synonyms between words in order to improve the search for answers and improve accuracy.
03 Medical knowledge base in the specialized medical field, how to quickly provide the doctors group with the latest medical literature, guidelines and other information, identify and answer professional medical information, it is necessary to structure medical information and form a knowledge base to support algorithms to automatically deal with common problems.
, for example, establishing a map of medical knowledge related to disease, symptoms, drugs, laboratory methods, clinical studies, etc., can answer common questions related to symptoms and the logic of the disease.
01VIEWPOINT builds a medical knowledge map on the premise that it is a full understanding of the medical profession - in Wikipedia's official terms: Knowledge map is Google's knowledge base for enhancing the functionality of its search engine.
Essentially, the knowledge map is designed to describe the various entities or concepts and their relationships that exist in the real world, which constitute a huge semantic network diagram in which nodes represent entities or concepts, and edges are made up of genus relationships.
: refers to something that exists differentiatedly and independently.
is just a node in the map.
such as "omeprazole," "heterointestinal ulcers," "prescription drugs" and so on.
entities are the most basic elements in the knowledge map, and there are different relationships between different entities.
: is the edge of the knowledge map, which refers to the relationship between an entity and an entity.
the relationship between the term penicillin and penicillin (drugs) is that penicillin is the name penicillin (drug).
Shapin's knowledge map example shows that penicillin (drug) an entity, hemolytic infectious diseases are an entity, adaptation is a relationship, indicating that the relationship between hemolytic streptococcus infectious diseases is hemolytic streptococcus infectious disease is penicillin adaptation.
of penicillin-adaptation-hemolytic infection-type diseases is a thromymose group (entity-relationship-entity).
the core of the medical knowledge map is the understanding of the business and the design of the knowledge map itself, which is particularly critical, depending on the in-depth understanding of the business and the prediction of future business scenario changes.
Based on its own strong medical team, firestone number wisdom builds a medical term set to explore the discovery data laws, couples data elements with business production and organization, completes the cognitive leap from objective data aggregation to abstract knowledge precipitation, and provides knowledge-driven auxiliary decision-making for the organization.
02VIEWPOINT through the medical knowledge map construction and traditional search for different intelligent search - Firestone Intelligence Intelligent Medical Content Center (AIMed) in the construction process, training AI model based on PICOs decomposition method of multi-source heterogeneous literature data to deconstruct, extract the indicator information contained in the literature, form a label, into a highly structured data, to support the rapid call and retrieval of information.
, through the construction of medical term set and medical knowledge map, will achieve more in line with the field of intelligent search in the field of medicine, for users to present more "medical" search results.
Traditional search: from the past we look up information online, enter keywords will appear after a lot of related links, such as in the knowledge of the internet space input keywords "Omerazo" search results are as follows: in this search environment, users need to open these links one by one to find the information they want to know, if you want to know about omeprazole adaptation, you need to search for "Omerazosis."
traditional search engine simply extracts the user's input keywords from each page to match, then sorts the results by relevance, and then returns the searched entries by sorting.
, the computer doesn't "understand" the words people enter, it just treats them as a field-by-field match.
Knowledge Map Search: 1) Search for a keyword that returns all aspects of the keyword, as shown below: The search results for the knowledge map are to connect the knowledge related to "Omerazo" to the network, from the keyword "Omerazo" can be connected to the "Omerazo" related to other knowledge For example: introduction to omeprazole compounds, pharmacopeia standards, drug descriptions and expert reviews, and some characteristic properties of omeprazole are displayed: molecular, CAS number, Chinese alias, English name, drug name, etc.;
these are in the knowledge map formed in the semantic network to show the nodes connected to omeprazole, is not possible with traditional search.
2) In knowledge map search, search engines "understand" what the user enters more accurately by trying to identify and understand the user's intent.
for example: want to know the omeprazole's adhesive, direct search for "omeprazole's adhesive" accurately presents the adaptive disorder is: the treatment of hetero-fingered intestinal ulcers, stress ulcers and so on.
03CONCLUSION conclusion - In summary, the knowledge map is a large knowledge network, and to a certain extent can "understand" what people enter.
search based on knowledge map has become the main form of search engine, and its technical framework is constantly improving and perfecting.
As a pioneer of intelligent innovation in medicine, Firestone is adhering to the belief of "making medical evidence generation simpler and medical value transmission more efficient", with professional precipitation in the field of medicine, the medical knowledge map developed not only provides intelligent search of knowledge map of medical terminology, but also provides intelligent search of knowledge map of medical health non-professional terms commonly used by the public.
terminology data sources include not only international standard terms such as WHODrug, HPO, MedDRA, UMLS, but also from pharmaceutical, health literature, dictionaries, books, clinical guidelines, etc., which are used to build the data base of the three product lines of Firestone Intelligence Medical Robots: AIMed, Social Listening, and Chatbot.
non-professional terminology data sources are mainly derived from medical and health-related social media commonly used terms, so that Firestone can not only achieve intelligent search, but also enable intelligent medical robots to provide a more user-friendly intelligent question and answer.
terminology expert Yu Cuilan