This item is available under a Creative Commons License for non-commercial use only
1.2 COMPUTER AND INFORMATION SCIENCE, Computer Sciences
The objective of this study is to predict admissions from the paediatric Emergency Department in the Republic of Ireland to alleviate the problem of patient boarding that can lead to overcrowding. Providing advance notice to ED clinicians and bed managers can facilitate quicker decision making and planning. Two years of data from 2017 to 2018 was extracted from the hospital systems to generate the dataset. Logistic regression, naïve bayes and decision tree were applied to the dataset and evaluated for model performance. The logistic regression was the top performing model based on an AUC of 0.84 and sensitivity of 0.51. The variables of importance were identified as referral source, location type, re-attendance in the previous 7 days, triage category, presenting complaint and previous discharges (p< 0.0001 for all variables). A final phase of the project involved the development of a prototype for potential deployment of the prediction results in a user friendly visual medium for hospital staff.
Leonard, F. (2019) Prediction of Admissions from the Paediatric Emergency Department, Masters Thesis, Technological University Dublin.