Background Unplanned admission of a patient which is a vague or fuzzy event has important financial implications for efficient use of hospital resources. Patients at high risk of admission are of major concern due to heavy use of hospital resources. Traditional approaches are not capable of accounting for the complex uncertainty and vague nature of hospital admissions. Methods adapting fuzzy regression methods could be an alternative method for decision-making experts to predict patient admission.

Purpose To deal with uncertainty in health system variables, identify the relationship between risk of admission and risk factors associated with the admission of a patient, and capture a vague definition of admission of a patient.

Materials and Methods A modelling approach adapting a fuzzy regression method was designed and developed using UK Hospital Episode Statistics (HES) data to capture the vague definition of admission of a patient. This model deals with uncertainty in health system variables which act as input variables in the model. The data collected is fuzzified, upper and lower bounds of the fuzzy membership function are evaluated using a JAVA programme that uses fuzzy regression methods.

Results

1. The fuzzy membership function was evaluated for about 10,000 patient records.

2. 404 inpatient variables were scanned using HES data sets.

3. Significant risk factors were admission source, admission method, reference conditions, age, length of stay, disease diagnosis.

4. The uncertain relationship between predictors and outcome associated with it is shown with the help of upper and lower bound regression equations.

Conclusions The fuzzy regression model was found to be capable of quantifying and estimating the unknown relationships between input predictors and predicted outcomes. The findings suggest that the fuzzy regression approach provides a good way of dealing with uncertainty in health system variables and vagueness in the admission of a patient.

No conflict of interest.