Based on the distributed assessment of each attribute, we can identify weak areas together with strengths for each alternative. Furthermore, to rank alternatives on one attribute or all attributes, a single score to represent the performance of each alternative FRAX597 necessary. Distributed assessment results, as discussed above, may not be directly used for ranking. Yang and Xu (2002) proposed employing the concept of expected utility to generate a numerical value from each distributed assessment to rank alternatives. For example, if the overall medical quality of one hospital is assessed as (excellent, 0.85), (good, 0.00), (average, 0.15), (poor, 0.00), (worst, 0.00) , and we assign a stomach utility of 100 to excellent quality, 80 to good quality, 60 to average quality, 40 to poor quality, and 20 to worst quality, then we can obtain a combined quality score of the hospital as follows: 0.85 ∗ 100 + 0.00 ∗ 80 + 0.15 ∗ 60 + 0.00 ∗ 40 + 0.00 ∗ 0 = 94. The quality score of 94 can be used to rank the medical quality of the hospital.