Selected among top 10 finalists in the 24hr Hackathon for building a Contextual Care Protocol using weighted decision trees and neural networks. Such a protocol had many features like:
- Used existing data and methods to leverage itself for optimal solutions
- Could learn while running the protocol using corrections from experts.
- Still be accurate in diagnosis.
As one of the challenges was to directly edit the decision tree and have the decision tree understandable by humans, we had to implement our own method of decision trees.