Hardalaç F., Kutbay U.

JOURNAL OF THE FACULTY OF ENGINEERING AND ARCHITECTURE OF GAZI UNIVERSITY, vol.29, no.1, pp.149-154, 2014 (SCI-Expanded) identifier identifier


Usage of drug has many risks. These risks are drug related problems in hospital admission, drug related problems during hospitalization, drug related problems at hospital discharge, medication errors and drug-drug interactions (DDIs) [1]. Because of the DDIs fatal effects, U. S. Food and Drug Administration (FDA) and European Medicines Agency (EMEA) are researching at this area [2]. Lazarou et al. investigated 6.7% of hospitalized patients, having a fatal DDIs with the rate of 0.32% [3]. The cost of DDIs related mortality is $136 billion annually in the USA [4]. Preventing the deadly effects of DDIs, classifying of DDIs using Neural Networks is aimed at this study. In this study, Jordan Elman Networks were applied for some DDIs and classification process trained for 1000 steps. At the end of 149 training step, using Levenberg Marquardt learning algorithm, Jordan network has been constituted with 0.0305 MSE and as a result of testing network, correlation coefficient was obtained as 0.8177. This study is also supported by Republic of Turkey Ministry of Science, Industry and Technology as "Drug Interaction (code number is 00912.STZ.2011-1)"