1Department of Industrial Engineering, Iran University of Science & Technology, Tehran, Iran
2Department of Industrial & Manufacturing Engineering, North Dakota State University, United States of America
Background and Objectives: Evaluating the performance of clinical units is critical for effective management of health settings. Certain assessment of clinical variables for performance analysis is not always possible, calling for use of uncertainty theory. This study aimed to develop and evaluate an integrated independent component analysis-fuzzy-data envelopment analysis approach to accurate the performance measurement of clinical units under uncertainty. Methods: Correlations between the input variables were calculated using Pearson’s correlation coefficient. Independent component analysis was used to extract independent components from input variables. Independent components were filtered against Gaussianity using Kurtosis parameter. An integrated independent component analysis-fuzzy-data envelopment analysis method was developed by using the uncertainty theory in the nonlinear fractional model proposed by Charnes, Cooper, and Rhodes (1978). The resulting fuzzy efficiency numbers were converted into normal ranking values by calculating amatrix of degree of preference. Findings: Under certainty, while data envelopment analysis identified 12 out of the 19 units as efficient units, independent component analysis-data envelopment analysis approach identified only three efficient units. On the other hand, under fuzzy conditions, while fuzzy-data envelopment analysis identified 12 efficient units, independent component analysis-fuzzy-data envelopment analysis identified only three units as efficient units. Conclusions: The results indicated that independent component analysis-fuzzy-data envelopment analysis offers the same efficiency measurement performance under fuzzy conditions as corresponding non-fuzzy method does under certain conditions. Our findings, hence, recommend use of the new approach in estimating efficiency of clinical units when access to reliable data is limited.