4. Discussion
In this study, 25 laboratories were nominated to form a laboratory network. The initial condition for determining the members of the network is to calculate the efficiency values of the laboratory units. For this reason, we used the efficiency values calculated by Ghafari Someh et al. 8. Then, by assigning the performance scores to the three clusters by the k-means algorithm, we determine the network members. In the first cluster we put members who can form a network, the second cluster includes members who need to have a strategy for improvement in order to be accepted in the network, and the third cluster includes members who have no chance of joining the network. Since the network depends on the cooperation of the members, we use a cooperative game called Shapley’s value which has the advantage of fair profit sharing, and we propose a method for dividing profit. The results show that managers can improve network performance by sharing fair profits among network members, and in addition to network survival, they can create incentive strategies for members of the network.
A limitation exists in this research. In a real-world situation, we are faced with risk and uncertainty in healthcare systems. Therefore, labs efficiency score might not correspond to actual values in this study. Therefore, several interesting directions can be further studied. The network of collaboration in an uncertain environment should be considered and the use of uncertain DEA methods, such as Fuzzy DEA, interval DEA, and robust DEA, may also be investigated and compared.