Open Corpus will make medical knowledge available to all
Among all the fields of science, medicine is the most well guarded. Students have to go through a sharp selection process, and most of the medical knowledge is still passed down orally and through obscure books.
You can become a computer scientist from high-quality courses available online, there are MOOCS for most area of studies, but not for medicine. Most medical articles on Wikipedia are lacking and do not help you get the big picture.
This relative opacity leads engineers and researchers to think medicine is still in need of structure. Through the window of pharmaceutical research, they might think that medicine has more to do with heuristics than with a formal and complex relational system that has evolved over hundreds of years. They mistakenly try to create new ways to detect symptoms and classify diseases using large neural nets and big data from millions of medical records. This will be useful, but it is not at all the most pressing nor will it be the most efficient way to make an AI system learn.
Knowledge in medicine is semantically organized, in a well-defined web of logical patterns. This system is often learned through training until an intuition is built over it, and even if through experience older specialists would sometimes tend to forget it and rely on keen intuition, the system is solid and remains central in how doctors conceptualize, communicate, and reason about medicine. You need learning agents to first learn this system before going for unconstrained learning and try to create new classifications. Such agents would also be able to help us go beyond our current knowledge and learn from medical records and experience. But by integrating structure into learning agents, you gain time and reduce the space of your search using a rich knowledge base. You would also enable communication with students and physicians, in a language they can understand, learn from, and use in their daily practice.
Open-Corpus is a platform that will democratize medical semantic knowledge
Using graphical models able to represent both complex nested objects and statistical relationships, Open-Corpus will offer an interactive visual access to medical knowledge. It will be dynamic and users will be able to improve the knowledge base through a curated contribution system. The knowledge will be semantic in nature and we will work to meet the constraints of the Wikidata format in order to offer our knowledge base to the wikidata users. We hope that this will help disseminate the knowledge across platforms.
Upon this large semantic database we will build tools to help students learn, and physician work in the hospital. We are presently building a medical assistant using Statistical Relational Artificial Intelligence (StarAi) that will be able to gather individual patients’ data using multiple entry points and traverse the knowledge graph to assist doctors in their daily work.
Open-Corpus will be completely free and open-source, but building commercial extensions using the knowledge base and free tools will be allowed.
The revolution is moving forward, and we hope you will join us.