During my final year, I had the opportunity to work with Professor Michael Carter and fellow Engineering Science student Jonathan Wang on my undergraduate thesis with the support of a Toronto hospital’s Patient Safety Physician Lead:
Development of a Computer Model for [client hospital] to Predict and Plan for a Pandemic Influenza Surge
Abstract:
Pandemic influenza (PDI) is an infrequent but deadly occurrence of influenza and differs from seasonal influenza in its severity and transmissibility. Historically, there have been three occurrences of PDI in the 20th century—in 1918 with a worldwide mortality of 50 to 100 million people, in 1957 and in 1968 with a combined worldwide mortality of 1 to 6 million people. A major surge in PDI patients can result in shortages of hospital resources such as emergency department beds, intensive care unit beds and mechanical ventilators. It is therefore important to adequately plan to mitigate the shortages should a pandemic occur. This research focuses on developing an influenza prediction and resource allocation model to assist [client hospital] plan for future pandemics. The approach was to take FluSurge, a model built by the Centre for Disease and Control in the US, and build a real-time prediction mechanism so that hospital administrators could enter their current data to obtain an estimate of how long and how virulent the pandemic will be. The prediction mechanism is comprised of two parts: 1) an assessment of when a pandemic has started in the hospital (using moving seasonal decomposition methods and control charts) and 2) an estimate of the most likely scenario given user-inputted data (using RMS error minimization between real time data and predicted data). Preliminary validation testing indicates that the model is effective at predicting local surges in the data but may not predict larger scale trends in the data. However, the more data that was inputted into the model, the more accurate the prediction. Future work lies in expanding the resource allocation model and in running more diverse sets of data through the model for better validation. Ultimately, it is the hope of this thesis to be able to assist [client hospital] in developing their pandemic plan so that in the event of a pandemic, health care workers can provide quality medical care without being hindered by logistical resource issues.
TL; DR:
We built upon FluSurge: an influenza prediction software compiled by the the US Center for Disease Control and introduced new functionality on it (we bundled the new methodology as FluPredict) so we can answer a series of question to determine the potential impact of an influenza outbreak on a hospital resources.
