All frontline hospital staff need to undergo an annual Quantitative Fit Testing (QnFIT) for optimal Personal Protective Respirator (PPR). There is a need to ascertain whether or not the employee passes the QnFIT for using the employer provided PPR. Whilst individuals may have many different facial features, there are only a few standard models of the P2/N95 mask available at any given time. This is further dependant on the limitations of a supply chain to work in real time. In a pandemic like COVID-19, supply chain disruptions can make it difficult to ascertain and procure alternatives, highlighting the importance of conserving PPRs during the fit testing process (i.e. reducing the failure rates and the need to use multiple PPRs to identify the correct fit). Our UQ research identified that it is possible to use UQ developed AI facial imaging software to identify the most appropriate PPR (i.e. make, model and size) thereby improving the success rate of QnFIT. This AI Software may potentially help to reduce turnaround times, improve the rapid availability of staff to respond to the pandemic and reduce the onboarding time for new staff that need a QnFIT. Importantly it may also help reduce the high consumption of PPR during the QnFIT process. Our initial prototype system demonstrated 54% mask selection accuracy (human accuracy 39%). This project will deliver further analysis of the collected hospital data, will develop this strategic and close hospital collaboration and will continue to develop the concept of touchless online fitting of items to people and progress the initial prototype to find a solution to rapidly fitting N95 masks to the 90,000 frontline staff in Queensland Health.

Project members

Professor Brian Lovell

School of Information Technology and Electrical Engineering

Dr Patrick Zhang

Postdoctoral Research Fellow
School of Information Technology and Electrical Engineering