Impact of missing data in evaluating artificial neural networks trained on complete data

Authors:
Mia K. Markey;Georgia D. Tourassi;Michael Margolis;David M. DeLong
Affiliations:
Biomedical Engineering Department, The University of Texas at Austin, 1 University Station, C0800, ENS617B, Austin, TX 78712, USA;Department of Radiology, Duke University Medical Center;Biomedical Engineering Department, The University of Texas at Austin, 1 University Station, C0800, ENS617B, Austin, TX 78712, USA;Department of Biostatistics and Bioinformatics, Duke University Medical Center
Venue:
Computers in Biology and Medicine
Year:
2006

Citing 4
Cited 6

Computer-aided diagnosis of breast lesions in medical images

Computing in Science and Engineering
Neural Networks for Pattern Recognition

Neural Networks for Pattern Recognition
Machine Learning

Machine Learning
Self-organizing map for cluster analysis of a breast cancer database

Artificial Intelligence in Medicine

Evaluating the Impact of Missing Data Imputation

ADMA '09 Proceedings of the 5th International Conference on Advanced Data Mining and Applications
Selection-fusion approach for classification of datasets with missing values

Pattern Recognition
A study on the use of imputation methods for experimentation with Radial Basis Function Network classifiers handling missing attribute values: The good synergy between RBFNs and EventCovering method

Neural Networks
Neural networks and other machine learning methods in cancer research

IWANN'07 Proceedings of the 9th international work conference on Artificial neural networks
Consequences of data uncertainty and data precision in artificial neural network sugar cane yield prediction

IWANN'07 Proceedings of the 9th international work conference on Artificial neural networks
Missing values: how many can they be to preserve classification reliability?

Artificial Intelligence Review

Quantified Score

Hi-index	0.00

Visualization

Abstract

This study investigated the impact of missing data in the evaluation of artificial neural network (ANN) models trained on complete data for the task of predicting whether breast lesions are benign or malignant from their mammographic Breast Imaging and Reporting Data System (BI-RADS^T^M) descriptors. A feed-forward, back-propagation ANN was tested with three methods for estimating the missing values. Similar results were achieved with a constraint satisfaction ANN, which can accommodate missing values without a separate estimation step. This empirical study highlights the need for additional research on developing robust clinical decision support systems for realistic environments in which key information may be unknown or inaccessible.