Emergency Department Visit Forecasting and Dynamic Nursing Staff Allocation Using Machine Learning Techniques With Readily Available Open-Source Software

Abstract

Although emergency department visit forecasting can be of use for nurse staff planning, previous research has focused on models that lacked sufficient resolution and realistic error metrics for these predictions to be applied in practice. Using data from a 1100-bed specialized care hospital with 553,000 patients assigned to its healthcare area, forecasts with different prediction horizons, from 2 to 24 weeks ahead, with an 8-hour granularity, using support vector regression, M5P, and stratified average time-series models were generated with an open-source software package. As overstaffing and understaffing errors have different implications, error metrics and potential personnel monetary savings were calculated with a custom validation scheme, which simulated subsequent generation of predictions during a 4-year period. Results were then compared with a generalized estimating equation regression. Support vector regression and M5P models were found to be superior to the stratified average model with a 95% confidence interval. Our findings suggest that medium and severe understaffing situations could be reduced in more than an order of magnitude and average yearly savings of up to €683,500 could be achieved if dynamic nursing staff allocation was performed with support vector regression instead of the static staffing levels currently in use.

Artificial intelligence in emergency medicine: A scoping review

“…within the ED, including daily trauma volume. [48][49][50][51][52][53] These interventions have the potential to reduce ED wait-times through improved resource allocation and policy planning. Another 2 (1.3%) studies used ML to increase the efficiency of patient identification for ED clinical trials and research, with the goal of allowing research to be more accessible and standardized in an often fast-paced environment.…”

Design and evaluation of analytical tools for emergency department management based on machine learning techniques

“…The methodology we had developed for the forecasting of ED visits [101] could also be applied for the forecasting of ED census levels since, but with different dependent variables. Usually, data in ED information systems is stored in a format wherein each table row corresponds to one patient visit with a timestamp of entry to the ED and another timestamp of exit from the ED, among other variables.…”

“…We also had developed a model with such an approach [101]. However, it is more accurate to use the number of patients at the ED at a given moment (i.e.…”

“…Staffing costs should also be taken into account. Given an estimated average cost per nurse and 8-hour shift, an estimation of a staffing allocation policy can be obtained by means of adding up the predicted nursing personnel costs for all 8-hour shifts [101].…”

“…Parameterization could vary depending on the dependent variable (average or maximum ED census forecasts), the algorithm and the forecast horizon (N = 1, Table 5. Parameterization datasets Once models were parameterized (for maximum and average ED census forecasts stratified by 8-hour shifts), we used a variation of a "sliding window evaluation approach", similar to a previous methodological approach we had developed [101]. o the models were recalculated on this larger train set (for M5P, AMT and SVR, algorithms were retrained using the parameter selections for each dependent variable, algorithm and forecast horizon that had been obtained on the validation sets defined in Table 5 Notice that "macro test sets" stored different forecasts depending on the dependent variable (average and maximum census forecasts), algorithm (models retrained on increasingly larger test sets) and forecast horizon (N = 1, 2, 4, 8 weeks).…”