On April 09, 2020, the Roche Data Science Coalition (RDSC) requested the collaborative effort of the AI community to fight COVID-19. The challenge, posted on Kaggle, consisted of a curated collection of datasets from 20 global sources and asked participants to model solutions to key questions developed by global frontline healthcare providers, hospitals, suppliers, and policymakers.
This dataset composed of a curated collection of over 200 publicly available COVID-19 related datasets from sources like Johns Hopkins, the WHO, the World Bank, the New York Times and many others. It also included data on a wide variety of potentially powerful statistics and indicators, like local and national infection rates, global social distancing policies, geospatial data on the movement of people and more.
As a response to this challenge, our Data Science team led by Ankan Ghosh posted a submission on how COVID-19 Mortality could be analyzed. The approach took into consideration that the date of Corononavirus’s spread is different for each country and data was available only till the end of March 2020. The main objective was to identify the countries in a pandemic state and determine the factors that contributed to mortality. These factors were then used to identify the countries/groups more vulnerable to mortality or were in a high-risk state.
- The study named Belgium, France, Germany, Iran, Italy, Netherlands, US and UK (listed in alphabetical order) as the countries as pandemic struck and would have higher mortalities.
- Female patients fared better than male patients, as males had a higher mortality rate. Smokers too had a higher mortality rate – 20 per 1000 people.
- Better healthcare infrastructure reduced mortality. Countries that had a higher number of beds per thousand people (the global average is only 2-5 beds per 1000 people) had lower mortality.
- The other factors that helped reduce mortality included higher per capita expenditure and per person availability of healthcare resources.
- Having said that, the mortality did not depend on the per capita income of the country or how much they invested in healthcare. This is evident from the impact that COVID-19 is having on top healthcare spenders like US and China. What mattered more was how the expenditure was distributed amongst the country’s population and how easily people could avail them.
Countries have a higher mortality risk due to low per capita expenditure on Healthcare
The submission made our team was selected by the Roche Data Science Coalition. The challenge itself featured 12 tasks and hundreds of entries from across the globe. “This challenge gave our Data Science team an opportunity to demonstrate their mettle,” said Arun G, Director – Technology & People Practice, Ideas2IT Technologies.
“On a larger note, this initiative clearly highlights how central Data is for organizations, governments and healthcare bodies for their resource planning and distribution initiatives.” quipped Murali Vivekanandan, Founder & CEO, Ideas2IT Technologies.