Caregiver workload monitoring in the ICU using Artificial Intelligence

This is an offshoot of a longstanding project that would suit those with a coding/hardware/engineering background as well as medical/nursing students. Caregiver workload in the ICU setting is difficult to numerically quantify. While there are numerical scores that can be manually applied to semi quantitatively assess workload, these are often inconsistently applied. Ambient Intelligence utilises computer vision-guided neural networks to continuously monitor multiple datapoints in video feeds, and has become increasingly efficient at automatically tracking various aspects of human movement. Specifically, opensource computer vision algorithms such as Google YOLO can be trained to recognize specific activities to automatically score workload. This, however, requires video to be manually annotated to recognize these activities. We are currently seeking either medical or nursing students with a familiarity of some of the activities involved in patient care in the ICU that can help label videos. A total of 7000 infrared videos need to be annotated to train this algorithm. This can be done from home via a secure VPN. If students do not have a familiarity with usual ICU activities but would like to learn, we can facilitate their participating and watching patient care in the unit. I have supervised multiple students over the years. Many of these have gone and progressed to research in their own right and many others have used the experience to get onto their training program of choice. All of them have authorship on at least one paper, with one student having authored 5 papers with me. There is scope to present your work both locally and internationally, collaborate with Healthtech organizations in Melbourne, as well as attend local and international medical or technology conferences. Please contact me on peter.chan@easternhealth.org.au as the Monash email is not checked very frequently.