Abstract
The objective of the work reported in this paper was to develop a model that predicts the serious adverse drug reactions (ADRs) on medication uses. The predictive model is developed using the feed-forward back-propagation type of artificial neural network (ANN) using the Levenberg-Marquardt algorithm. The target and input data of the ANN model are derived from ADR data in FDA's adverse event reporting system. The target data contain the serious and non-serious ADRs. An ADR dataset consisting of 3,164 observations is used to obtain preliminary results. The preliminary results show that the ANN model provides 99.87% accuracy with the sensitivity of 99.11% for the serious ADRs and the specificity of 100% for the non-serious ADRs. These preliminary results will be further verified by a research using an ADR dataset containing 10,000 observations.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 53-59 |
| Number of pages | 7 |
| Journal | International Journal of Medical Engineering and Informatics |
| Volume | 3 |
| Issue number | 1 |
| DOIs | |
| State | Published - Mar 2011 |
All Science Journal Classification (ASJC) codes
- Medicine (miscellaneous)
- Biomaterials
- Biomedical Engineering
- Health Informatics
Keywords
- ADRs
- ANN
- Accuracy
- Adverse drug reactions
- Artificial neural network
- Feed-forward back-propagation.
- Levenberg-Marquardt algorithm
- Logistics regression
- Multiple layers
- Prediction
- Sensitivity
- Specificity
- Supervised training