SUCCESS CASE #MachineLearning
Modeling the Spread of Covid-19 Infection
Spain has been one of the world’s countries most stricken by this new global pandemic and Madrid has been its epicenter, recording the most cases and deaths in the country.
Regional Government of Madrid
The Government of the Community of Madrid is the body in charge of directing the regional policy. It is responsible for the executive and administrative functions, as well as the exercise of regulatory authority in matters within its competences. Each community has a health department responsible for key areas such as healthcare planning, public health, and management of health services that have the primary aim of ensuring proximity of services for users.
In the current early phase of the COVID-19 outbreak, Madrid needed to prepare to face the escalation of cases, with a particular focus on the readiness of healthcare services. Containing and mitigating the spread and infection rate of the virus has been the first priority of Madrid public health authorities to distribute the number of infections over time and, if possible, reduce the incidence of the disease it causes.
Although new data on COVID-19 are available daily, information about the biological and epidemiological characteristics of COVID-19 remain limited, and uncertainty remains around nearly all parameter values.
For example, estimates of case fatality ratios must account for numerous biases, including high numbers of asymptomatic cases, under-reporting of symptomatic cases, under-reporting of COVID-19 associated deaths, and the delay between case reporting and death reporting. There is also likely regional variability in testing practices, reported incidence.
The machine learning perspective on modeling infectious disease spread involves consideration of these large number of modeling parameters detailing the spread of and recovery from the disease and additional different compartments corresponding to demographic and social factors.
Accurate forecasting of the number of COVID-19 cases is becoming the backbone to facilitate the use of the available resources in hospitals and improve management strategies to optimally manage infected patients. And for that purpose, Keepler has developed an interactive dashboard in Quicksight, since a data-driven approach is crucial to understand the evolution of the pandemic and the use of machine learning models is essential to develop regional-scale models for forecasting and assessing the course of the pandemic.
Machine learning models also can help evaluate the potential effects of different community mitigation strategies (e.g., social distancing) and simulated the future trajectory of the epidemic under different scenarios and project the impact and the implications for healthcare capacity and policy interventions.
The solution follows the AWS Well-Architected Framework best practices. A serverless solution has been implemented in order to optimize costs and unnecessary infrastructure maintenance. The services used in the solution are:
AWS S3: store the information.
AWS CloudWatch: trigger ETL, prediction execution and logs.
AWS Batch: preprocessing and cleaning data.
AWS Glue: construct Data Catalog.
AWS Athena: query SQL to analyze data in S3.
AWS QuickSight: build visualizations.
AWS SageMaker: create and deploy machine learning models.
Keepler is a boutique company of professional technology services specialized in design, construction, deployment and software solutions operations of Big Data and Machine Learning for big clients. They use Agile and Devops methodologies and native services of the public cloud to build sophisticated business applications focused in data and integrated with different sources in batch mode and real time. They have Advanced Consulting Partner level and have a technical workforce with 90% of their professionals certified in AWS. Keepler is currently working for big clients in different markets, such as financing services, industry, energy, telecommunications and media.