The "exponential rate" has been widely cited by all kinds of social media since the outbreak of the pandemic caused by COVID-19. The figure below provides a visualisation of the trending behaviour of the COVID-19 data of the European countries.
How to better understand the trending behaviour is crucial from the perspective of forecasting and policy making. This project explores some time series and panel data models to study the trending behaviour of COVID-19 using country level data. The main objective is to help us understand how fast the COVID-19 spreads within the community, and account for the varying policies which are frequently updated by each government.
Professor Di Cook has been monitoring the COVID-19 data with her own web app. A link is provided to the source code, so others can download, run, modify, and learn how different plots and data wrangling are done.
The data is automatically extracted from different public sources, and plots are created in order to understand the progression and effect of the pandemic across the globe, and at least initially in Australia. This has been useful for several of our classes, where developing web apps for communicating data is being taught. There are various conscious choices in the design: (1) focus on local, because we needed to get a better idea of the trend of incidence, testing and mortality in Australia in the early weeks when other web sites had little about us; (2) calibrate counts and tests by population so we could compare the response by state; (3) not process the data too much, which doesn't address the differences reporting between countries but also doesn't impose a belief; (4) focus on "not mortality rates" because that's more encouraging; (5) compare incidence and recovery rates for Australia with China, to get a sense for when we would expect the end of the crisis here; (6) pull data such as pedestrian traffic in Melbourne and compare with last year to see if people are heeding the government instructions. She has also pulled CO2 data to see if there is a climate benefit, but it's too early to tell. Data used in the app comes primarily from Johns Hopkins, with Australian data extracted from the rapid compilation made by the Monash FIT-led "The real-time COVID-19 status in Australia".
Professor Rob J Hyndman is part of the Australian government advisory panel on COVID-19 modelling and forecasting, working with epidemiologists and statisticians primarily from the University of Melbourne. His role has been to develop forecasting ensembles based on various epidemiological models and time series models, in order to provide a comprehensive picture of the likely evolution of the disease in Australia over a four week forecasting horizon.
He has also written some blog posts about COVID-19 data:
Professor Brett Inder has an ongoing contract with the Australian Aid Program to undertake economic and statistical research into economic development issues in Timor-Leste. This involves analysing large, official datasets such as Census data, plus overseeing data collection and analysis on a few specific projects. The work is designed to better inform the Timorese government and civil society on economic development issues and policy options for the future. In 2020 this work has switched to specific advice on the economic response to the COVID-19 threat. The Government has designed a range of short term responses to deal with the State of Emergency where many business activities have been shut down. Professor Inder’s role is to do some background research on options and likely impacts, as well as overseeing data collection on actual effects. This work provides input to Government ministers in their implementation of policies. He is also providing guidance on longer term responses to the global situation, as they are likely to affect the Timorese economy.
Dr Montero Manso has been part of the Spanish "Mathematics against coronavirus" action collaborative effort, in which he has been developing some new forecasting methods, and coordinating and combining the mathematical modeling efforts from research teams across Spain. Forecasts are now open to the public .
Accurate forecasting of the main COVID-19 variables is of special interest at the level of administrative regions, rather than at an aggregated national level, since it informs decisions around distribution of resources. Forecasting at this level is challenging due to regions having varying population, age profile, population density, mobility, etc. while simultaneously the data becomes less reliable the smaller the area.
Dr Montero Manso has developed some new forecasting methods to tackle these issues by leveraging the availability of data from regions with similar characteristics across the world to support the models of the regions of interest in a purely data-driven way. He has also assisted the Australian government forecasting panel to apply the same methods.
COVID-19 brings a ‘new normal’ with lots of new issues and more volatility to come in future – societal concerns on the wealth gap, global warming, geo-political uncertainty, digitalisation and changes in how we work. If our most important work is figuring out and addressing problems, then the challenge amidst risk, uncertainty and opportunity is to figure out the question, ‘How can we become a problem-solving company?’
This article is for leaders of businesses and focuses on key risks, uncertainties and opportunities to think about. Not just for now when organisations have been forced to act quickly by necessity but to prepare for the future and be better placed to act quickly by design.
With growing awareness of the impacts of COVID-19 and the various restrictions on people's mental health, the Government has responded with expanding access to mental health services, as well as more funding for telehealth. Brett has been part of a team looking at inequities in access to mental health services across Australia. Through comparing national surveys that show the spatial distribution of prevalence of mental health conditions, medicare data on service utilisation and ABS data on socio-economic disadvantage, the research has highlighted the reality of greater prevalence in disadvantaged areas, and much lower access to services. In our recent work, reported in an article in The Conversation, we show that recent changes that aim to increase access to services may actually increase the inequities, biasing the expanded access to those better able to use telehealth and those in higher socioeconomic areas. Some creative changes to medicare funding models may be able to reverse this increasing inequity.
Greater needs, but poorer access to services: why COVID mental health measures must target disadvantaged areas