Take ten minutes to get to know Harvey Jia Wei Koh, one of four new PhD students working on the DHCRC Practice Analytics project in collaboration with Monash University, the University of Sydney, the Royal Australasian College of Physicians, Cabrini Health, Sydney Adventist Hospital and St John of God Healthcare.
Tell us a little about your life before you started your PhD
I grew up in the north of Malaysia, near the Thai border, and my family emigrated to Australia when I finished high school, mostly so that my younger brother and I could get a better education. My mum is an accountant and my dad has a physics degree.
My Dad got me interested in science from a young age, buying me light bulbs and batteries, allowing me to learn how circuits work.
I did a Bachelor of Biomedical science at Monash University with the intent of finishing my education and studying medicine. However, during my honours project with the Prostate Cancer Registry, where I studied how data visualisations affected interpretation of clinical quality indicators, I got really interested in data science and its potential impact across multiple fields.
I’m a curious person by nature, and I felt that data allowed me to have an extra lens to view the world from a different perspective, which inspired me to do a Master of Data Science at Monash. On completing my Masters, I did a student internship with CSIRO. Here, I looked at datasets generated from the Australian Animal Disease Spread (AADIS) model to create an interactive geospatial visualisation tool to show the spatio-temporal spread of Foot and Mouth Disease.
I’m excited to do cross-disciplinary research that covers my interests in medicine and data science.
What got you interested in digital health?
I first became interested in digital health during my Honours year, doing a research project with the Prostate Cancer Registry, which gives feedback to hospital administrators and clinicians on their performance by collecting data from patients, hospitals and healthcare professionals. I had the opportunity to learn much more about healthcare systems and digital health from very talented people who were really passionate about improving clinical practice. That project made me think a lot more about the power of data and how to harness it, and I realised that I had zero skills, so I enrolled in my Master’s in Data Science.
What’s your thesis about?
The working title of my thesis is, Defining Clinical Practice Indicators. My work is part of a larger, cross-disciplinary project. I look at how we can identify potential clinical indicators from large hospital datasets, which can be reliably used to measure clinical performance. I will be working alongside three other PhD colleagues on this project and our research will feed into each other’s work.
Emma Whitelock-Wainwright, a psychologist, will look at how we make sense of health care data. Bernard Bucalon, an education technology engineer, will look at ways that data visualisation of physician performance can be used in education and in measuring patient outcomes. And medical anthropologist Carol Pizzuti, will investigate the attitudes of health professionals to performance data.
For my project, I will be looking at hospital administrative data, one of the most common bits of data collected by hospitals, and how we can use the data to generate quality indicators. The idea here is by utilising hospital administrative data, we can tap into larger datasets, allowing for more data to be analysed. A saying in data science is, “It’s not who has the best algorithms, it’s who has the most data.” That’s why Google generates such accurate search results – not only because they have such most sophisticated algorithms but through the fact that every day, we use Google to browse our internet. There’s a potential here to have much more accurate quality indicators by simply having more data.
There are limitations however to using hospital administrative data as the data tends to be used for billing and other administrative purposes. Studies have shown variability in administrative datasets between hospitals even within the same state. In statistics and data science, there’s a saying called GIGO, garbage in, garbage out: quality of data is crucial in producing a good statistical model. The same can be said about developing good quality indicators. If the quality of the data is suspect, then healthcare professionals won’t trust quality indicators to assess their performance. These variabilities would have to be standardised before good quality indicators can be generated and used to benchmark hospital performance. That will be a focus of my thesis, studying variability in hospital data across partner hospitals, standardising and generating good robust quality indicators that can accurately reflect quality of care provided to patients.
What are the benefits that we will get from this research?
I would like to think that my thesis may influence the way hospitals can collect and use data, and how it can translate into robust quality indicators that reflect the quality of care we provide patients, and that gives healthcare professionals information to help them improve the quality of care that they provide.
I’m part of a large project with collaborations with multiple PhD students and researchers, so it’s not just exciting to work with a talented team, I think that it’s also going to boost the value of our work for all of us.
When I worked with the Prostate Cancer Registry, an economic evaluation was done which calculated how much the Registry improved people’s lives, and found that for every dollar invested in the Registry, there was between a two and five dollar return in money saved and also in quality of life gained. I believe we will see a similar return from this work, in costs saved and in improved outcomes for patients.
What benefits have you seen from digital health?
The benefits keep mounting because the entire field of bioinformatics has grown exponentially and become more effective due to the increasing computing power that we have, so we are seeing improvements on an order of magnitude that’s quite amazing. At the start of the COVID-19 pandemic – it took just two weeks before the whole world had access to the genome sequence of the viral data. Just a few years ago that would not be possible.
Another example, the introduction of My Health Record should increase the entire efficiency of the health care system simply through having much more accurate patient records and less duplication of effort. Doctors have case histories at their fingertips and can make more accurate diagnoses.
What do you like to do when you’re not working or studying?
I’m very interested in data as I said and I have been inspired by Nate Silver, who wrote The Signal and the Noise. Currently I am working on analysing datasets from the North American Ice Hockey leagues for fun. Although I have lived in Melbourne for six years, I haven’t chosen an AFL team yet, so that’s another project, I’m going to do some analysis to work out which team to pick.
I look after my health and exercise each day and I also love to play guitar, I’m into blues music. Also as I come from a multicultural background, I speak a number of languages including Bahasa Malay, Mandarin Chinese, Hokkien and Cantonese and I’m now learning French.
by Fran Molloy