Leveraging Artificial Intelligence in SARS COVID 19 Data Analysis in Rwanda (LAISDAR Project)

LAISDAR aims to leverage Artificial Intelligence and Data Science Techniques to create a scalable framework for gathering, harmonizing, and accessing available SARS-CoV-2 / COVID-19 data in Rwanda, with a common interface for data querying, analysis and reporting on top of the data hub. The project will gather all existing fragmented data and will collect new enriched longitudinal data. We propose making optimal use of existing (open source) solutions, standards and tools offered by the Observational Health Data Sciences and Informatics (OHDSI) community especially the OHDSI Common Data Model (CDM) through the Observational Medical Outcomes Partnership (OMOP) initiative. This largely used OMOP CDM will help in pooling data from multiple existing cohorts and to standardize data elements creating a ready-to-use dataset. The supported data harmonization framework will help making the inventory and mapping of needed data elements/variables from each data (static and longitudinal) to OMOP CDM. After harmonizing, this project will create a common data portal interface for access, querying and analytics including OHDSI tools and machine learning techniques on top of the data hub portal. Within this framework all data will be securely accessed in federated manner, meaning that no data will leave its owner location; which will comply with data privacy regulations. This project will ultimately generate data to help in the identification of key factors influencing the susceptibility to infection and clinical manifestation, to assess the optimized therapeutic and clinical management options, and provide evidence-based recommendations for health policies in preventive strategies, protective actions, and disease management.

Project scope

1. Quantifying Physical Distancing

Physical distancing is key to avoid or slow down the spread of viruses. Each country has taken different policies and actions to restrict human mobility. In this project, we investigate how policies and actions affect human mobility in certain cities and countries. By referencing our analysis of policy and secondary impacts, we hope that decision makers can make effective and appropriate actions. Furthermore, by analyzing human mobility, we also aim to develop a physical distancing risk index to monitor the risk on areas with high population densities and probability of contraction. For further information and our up-to-date analytics results, please go here.

2. Emotion Analysis

For health, we have focused on emotion changes that people have experienced during this pandemic. Emotion changes have stemmed from various reasons such as unemployment, implementation of stay-at-home policies, fear of the virus, etc. We quantify emotion changes by using social media data, including Twitter and Instagram. Since the breakout of COVID-19, we have seen an increase in online discussions that use hashtags such as #COVID-19 and #depression. We believe it is vital to visualize and analyze the differences in people’s perceptions towards COVID-19.. We also hope to analyze overall responses to the pandemic by sentiment: sadness, depression, isolation, happiness, etc. Further detailed analysis will also look into specific keywords and corresponding trends. For further information and our up-to-date analytics results, please go here.

3. Impact on Human Life

Due to physical distancing and lockdown policies, people have begun relying on video conferencing tools for meetings, lectures, and conversations among friends more frequently than usual. Children are especially affected by the quarantine since many must refrain from going to their classrooms and take classes online. By leveraging various data sources, we will analyze how daily behavior has been affected by this pandemic, and also compare behaviors among different countries and cities. We will also measure online e-commerce and consumer behavior by analyzing sites such as Amazon. For further information and our up-to-date analytics results, please go here.

4. Prediction of Infected Cases

We will also attempt to predict future infections by consolidating various data sources. One of our target data is from crowd-sourcing data via social network services. We collect online posts about certain symptoms relevant to COVID-19 such as fevers and coughs under the assumption that there is approximately a 12-14 day window before people are found COVID-19 positive, and a 5 day incubation period. After collecting relevant posts and tweets, demographic features such as user location and age, combined with the number of confirmed cases, we will use machine learning algorithms to predict the number of COVID-19 cases that will occur in the future..