Abstract:
The rapid development of statistical learning methods and an increase in computational power and medical data repositories have significantly contributed to a plethora of...Show MoreMetadata
Abstract:
The rapid development of statistical learning methods and an increase in computational power and medical data repositories have significantly contributed to a plethora of applications for medical diagnostics prediction. In this paper we present a solution for an interesting problem which, surprisingly enough given its importance, has not gained much attention: detecting unexpected co-occurrences of data features. As standard data analysis methods do not directly produce a solution for the problem, we propose a measure for direct discovery of unlikely co-occurrences of data instances. We show how our method can be directly applied to detect possible incorrect medication for diabetes patients.
Published in: 2017 9th International Conference on Electronics, Computers and Artificial Intelligence (ECAI)
Date of Conference: 29 June 2017 - 01 July 2017
Date Added to IEEE Xplore: 07 December 2017
ISBN Information: