GraphConnect

Our goal is to revolutionize surgical education by representing surgical knowledge in a Neo4j graph. Our use case is education - all surgical learners keep a detailed log of the operations they perform. These surgical logs are a potentially rich source of information about the education of surgeons but they go largely unused because they are traditionally saved using medical billing terms that computers can't understand. To make full use of valuable surgical information, computers would need to understand queries like: 

"What steps does a surgeon take occur during this operation?"
"What objects does a learner use in this procedure?
"What anatomic structures does this disease affect?"

To solve this problem, we mapped about 3500 surgical findings (i.e. fractures, tumors) and 4000 surgical procedures in the domain of maxillofacial surgery. In a Neo4j graph, each finding was graphed with its relationships to diseases and anatomic structures. Then each surgical procedure was graphed with its processes, objects, and structures. All processes, objects, and structures were then mapped to each other.

The result is a useful graph of surgical knowledge that a machine can use to answer highly specific queries about surgical education, like "Does this surgeon need more experience with nerve grafts?", "Is this learner getting enough operations that use bone plates?", or "Has this learner had enough experience operating in the eye socket?" For over two years, we have been using this graph in our surgical training program to understand the experience of each individual learner and guide their education. Nothing similar exists in any medical domain.