Complex problems require sophisticated solutions, and when it comes to adapting to a global pandemic, capturing and making sense of vast amounts of data is critical to help mitigate societal and economic impactAlso, intelligently using non-traditional source of data such as social media and integrating it with traditional sources such as telecommunications and surveys to improve spatial and temporal granularity is extremely important. It is more important in countries such as Indonesia where accessing good quality data can be challenging. 

Indonesia has recorded roughly 155,000 cases of the COVID-19 virus, with a corresponding US$10 billion blow to its capital market and economic growth predicted to fall by 5.3-7.3% by the end of 2020. Low testing rates and high population density have hampered efforts to track the spread of the disease, with limited resources and distribution issues exacerbating the situation.

By bringing together a suite of data-driven technologies into a single decision support tool, a partnership between CSIRO’s Data61, The Department of Foreign Affairs and Trade, the UN Pulse Labs, Jakarta and the Indonesian Government is providing Australia’s neighbour with the tools needed to help tackle the pandemic.

Leveraging big data, artificial intelligencenatural language processing and agent based modelling, the system provides a combined alert and intelligence service to help local authorities understand and plan targeted action including directed testing, medical resource distribution and social assistance to where it’s needed most. The system will also allow scenario modelling to evaluate the effectiveness of a range of post lockdown strategies.

The web-based tool will provide crucial insights on the early detection of hotspots, population movement and country’s capacity to adapt to monumental change. The framework of the platform is grounded in two Data61 focus areas, social media analytics and evacuation modelling.

 

Data61 COVID-19 Indonesia

Watch the Flu uses statistical time series modelling and natural language processing to identify where and when an outbreak might occur.

“The social media analytics aspect will provide early warning and mobility insights, while agent-based modelling, which is what we use in our evacuation work, is being applied to better understand potential post-lockdown scenarios,” explained Data61 project team member Dr Mahesh Prakash.

“For example, the platform can be used to recognise and coordinate the populations’ movements to prevent the spread of COVID-19, but still ensure continued economic recovery, something that could be done through staggering arrivals, departures and time spent in offices and schools.”

A combination of natural language processing, data science and time series modelling underpin the social media monitoring feature, which can identify specific syndrome keywords and their context mentioned in Twitter posts.

Words such as ‘cough’, ‘wheezing’, ‘fever’, ‘headache’ and ‘breathing difficulties’ are will be monitored to determine the spread of potential COVID-19 symptoms across Indonesia, a technique that is also being applied to Australia’s COVID preparedness and response strategy.

“Any early warning monitoring system can be crucial to allowing hospitals and health institutions to be better prepared,” said platform co-developer and senior Data61 scientist Dr Cecile Paris.

“The advantage of this feature is that it can generate alerts prior to an influx in hospitals or clinics, thus providing authorities and institutions with some lead time for better preparation, and, ultimately, fewer deaths.” – Dr Cecile Paris

Social media data is also being incorporated within the platform to better understand the Indonesian population’s response to the outbreak. Data analysis will reveal key insights detailing population density, age profile, food security and availability of medical resources unique to specific areas.  

The platform’s movement modelling capability uses area insights derived from social media and aggregated telecommunications data to populate its community movement models.

These models will provide a greater understanding of spatio-temporal mobility of Indonesia’s population within and across various neighbourhoods, effectively enabling near-real-time virus spread assessments.

Decision makers and crisis management will use this vital information to understand where and how quickly the outbreak could be expected to spread, ultimately preventing the emergence of new hotspots.

“This tool could also be applied to other kinds of resource allocation, such as food provision and optimising transportation groups,” explained Dr Prakash.

Data61 COVID Indonesia

Project team member Dr Mahesh Prakash (pictured) will be drawing on his agent-based modelling expertise, which is used in natural disaster evacuation platforms at CSIRO’s Data61.

“There’s a range of possible applications from this project which we think will help with Indonesia’s short-term needs as well as medium to long-term. One of the many benefits of this work is enabling Indonesia to become more data aware and data-driven.”

Data61 is also collaborating with in-country research partner Pulse Lab Jakarta, who are renowned for harnessing alternative data sources and advanced data analytics methods to obtain actionable insights used to inform policy makers in Indonesia and across the Asia-Pacific region.

Dr Robertus Nugroho, Dean of Faculty of Computer Science at Indonesia’s Soegijapranata Catholic University and former Data61 PhD resident, is also supporting the development and translation of the natural language processing modules required for the social media component of the tool.

“As the 4th largest country in terms of population, Indonesia is one of the most active social media users,” he said.

“Having the CSIRO’s Data61 analytics and big data decision tools in Indonesia will support the authority’s ability to take appropriate action promptly, employing a strategic response plan based on real-time and accurate data analytics.”