Journal
STOCHASTIC ENVIRONMENTAL RESEARCH AND RISK ASSESSMENT
Volume 32, Issue 6, Pages 1743-1758Publisher
SPRINGER
DOI: 10.1007/s00477-017-1459-z
Keywords
Gaussian mixture model; Dempster-Shafer theory; Clostridium difficile infection; Predictive model; Unsupervised learning
Categories
Funding
- National Natural Science Foundation of China [61573290, 61503237]
- China Scholarship Council
Ask authors/readers for more resources
Clostridium difficile infection is one of the major patient safety concerns in hospitals worldwide. Clostridium difficile infection can have high economic burden to patients, hospitals, and government. Limited work has been done in the area of predictive modeling. In this article, A new predictive model based on Gaussian mixture model and Dempster-Shafter theory is proposed to predict Clostridium difficile infection incidence in hospitals. First, the Gaussian mixture model and expectation-maximization algorithms are used to generate explicit probability criteria of risk factors based on the given data. Second, Dempster-Shafter theory is used to predict the Clostridium difficile infection incidence based on the generated probability criteria that have different beliefs attributing to their different credits. The main procedure includes (1) generate the probability criteria model using Gaussian mixture model and expectation-maximization algorithm; (2) determine the credit of the probability criteria; (3) generate the basic probability assignment; (4) discount the evidences; (5) aggregate the evidences using Dempster combining rule; (6) predict Clostridium difficile infection incidence using pignistic probability transformation. Results show that the model has a higher accuracy than an existing model. The proposed model can generate the criteria ratings of risk factors automatically, which would potentially prevent the imprecision caused by the subjective judgement of experts. The proposed model can assist risk managers and hospital administrators in the prediction and control of Clostridium difficile infection incidence with optimizing their resources.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available