timhannifan / symptom-tree Goto Github PK

As I understand it, this excellent tool you've developed requires users to input all of their symptoms up front? I was wondering, although it might be outside of the research scope, whether this tool or data could be made to support the following use case: an interactive decision tree which asks the user whether they have a given symptom, and repeats until one or more diagnoses are found. It would try to order the questions so as to shorten the (avg, worst-case, etc) number of questions necessary, based on some probabilistic methods. Most symptom checkers I've seen require a person to enter the symptoms they are having, but this can be hard for people (like me) who don't know how to put into words their symptoms. Whereas if they are asked, do you have this symptom (and if necessary refer them to an explanation and or picture of what that symptom entails and what it doesn't entail, such as from an encyclopedia, such as perhaps by embedding google searches of webmd or mayo clinic, etc), it might be able to more completely elicit all corresponding symptoms (avoiding the need to match make with the data), and therefore, diagnose more diseases overall. An iterated version could theoretically be created which knows how dynamic certain symptoms are, and can prioritize asking about the more dynamic symptoms, in a mathematically sound fashion. Would you know where one could begin to implement such a system? I would possibly be interested in collaborating on such a system, although I don't know much Python.