Home

Welcome to "AEdetector" the first open source software that automatically detects if in an input string there is mentioned a potential adverse drug event. Several papers have shown that one can automatically
a) detect if a document is talking about potential adverse drug events
b) identify the sentences in a document that are about potential adverse drug events
c) extract the DRUG - PROBLEM relation
d) predict also based on c) if a DRUG will have a change in its label due to the PROBLEM
Those work made use of several softwares, some of them however not available under GPL since relied on the LVG system of UMLS (NLM), that requires a separate licensing. Furthermore, the LVG system is really "fat" (800+MB) and therefore were barely portable to smaller devices.

This software follows the so called "Machine Learning" strategy, that grounded on an extensive and rigorously created corpus, allows software machine to "learn" the patterns that are then needed to accomplish the task.

The first deliverable for this project is the User requirement specification. I will post it as soon as ready and am looking forward comments from the other contributors.

[luca]

Please have a look at the Discussion since initial thoughts are already starting discussing there