Covid Activity Risk Calculator (CovARC)

With the outbreak of SARS-CoV-2 in 2019, the world went through very drastic changes, and precautions against COVID infections are still very crucial for every individual in their daily life. Even today, there are new cases and a non-trivial chance of hospitalization from COVID-19. Though there have been several different tools developed to predict the active risk of COVID infection in a particular area at a given point of time[1], we target the identification of risk specific to day-to-day activities such as dining out, participating in social gatherings, or simply going out to the market, work or store. 

Our tool provides an easy way for every user to understand different levels of risk when doing a specific activity. It also helped users be more aware of precautionary measures they could take, such as wearing a particular type of mask or getting vaccinated to have safety from different variants and further avoid spreading the virus to others. As a result, we aim to ensure that people are affected less in their daily life by COVID and are more aware of measures they can take to ensure their safety against it.

With an estimated over 500 million cases, COVID-19 has affected a significant part of the world and has been one of the major pandemic outbreaks in recent years. Though the impact has been well controlled at the current stage in some parts of the world, the current solutions developed in the perspective of risk calculation are only limited to a few countries, with a majority of the world still having no tool to help in risk estimation as well as increasing awareness among people. CovARC, our Activity Risk Calculator, works in 200+ countries worldwide and at the regional level to ensure all of the different communities worldwide are positively impacted by this solution.

We take into account many factors, including the number of active cases, types of variants present at a particular location at a given point of time, how much the vaccines & masks reduce the risk, and how an individual’s age, gender, past chronic illness or past COVID illness changes the risk of infection, and chances of hospitalization or death when carrying out a specific activity. Apart from that, the number of people the individual will contact is also taken into account to further accurately predict the risk of COVID.

The sheer scale of the problem provides a considerable challenge to identify a fit-for-all solution that could scale to all parts of the world to improve safety and decrease risks related to COVID. Our solution uses various data sources to ensure that we create a one-stop solution that could be used in any part of the world to accurately estimate the different risks associated with carrying out daily life activities and helping people learn ways in which they can ensure more safety against COVID. This use as a learning tool in turn becomes a gamified public health intervention wherein people are encouraged to make safer choices that will reduce cases on a population level.

CovARC[18] is an easy-to-use everyday tool that calculates the risk of Covid-19 infection, death, and hospitalization risk based on the user's input and considers the number of active cases and variants at the location of the user at the given point of time as well as the user’s personal risk factors.

As a part of the initial pre-processing of the confirmed cases obtained through the Johns Hopkins dataset[19], we identify the number of active cases and compute a 14-day aggregate. Furthermore, we determine a range of 14-day aggregate active cases by taking the upper limit as the product of the 14-day aggregate active cases and the ratio between the confirmed cases obtained from Facebook's COVID-19[20]. Trends and Impact Survey and Johns Hopkins dataset[19]. To extract the final number of active cases, we subtract the number of confirmed cases from the previous day and take a 14-day aggregate. This process is carried out daily through GitHub actions so that we have updated active cases daily. 

We use the GSAID dataset[15] for calculating the presence of different variants in a particular region. The data is updated weekly to include new variants introduced in every region worldwide at a given time. We perform Gaussian smoothing over the variants dataset. Further, we use the country and region-wise population dataset to estimate the number of active cases per unit population to identify the COVID density in the given region. Since there are inaccuracies in reported number of confirmed cases, we cross-verify the confirmed cases using Facebook surveys[20] and generate a ratio between the number of confirmed cases by Facebook surveys[20] and the officially reported confirmed cases to calculate the variance in the number of confirmed cases. We use the ratio as a multiplication factor to calculate the upper limit for the active cases and the maximum range for risk of infection. This is particularly important in regions of the world with less ability to track and report cases and with underreporting now occurring due to rapid COVID tests.

In addition to these pre-existing datasets, we use information provided by several different research studies and articles to create custom datasets for the protection mask's fitted filtration efficacy (FFE), the efficacy of vaccines against different variants of virus, indoor and outdoor risk, risk of hospitalization and death with their dependency on age, gender, presence of variants, past COVID illness and past chronic illness. We use all of these datasets and take into account the number of people the user passes outdoors while traveling to the location and the number of people they are with at the destination (which can be indoors or outdoors) to estimate several risks associated with COVID (risk of infection, risk of hospitalization & risk of death).

Using different kinds of authentic data sources and user-provided information, we estimate the chances of getting an infection, hospitalization risk, and death. CovARC is also provided as a simple, lightweight web application that users can use on low bandwidth internet, which is crucial in many countries. Our tool provides a much more comprehensive approach to the problem than the existing methods. To our knowledge, this estimation method is the only one that considers most of the factors essential to estimate risk and further accurately makes it applicable for almost every country and region in the world. Due to its user-friendly interface, the risk estimation system is much simpler than the existing alternatives and can estimate risk within a few minutes. We aim to empower people to live their lives safely and get educated about risk reduction, which could ultimately decrease COVID-19 infections worldwide.

One of our main objectives while developing CovARC was to make this tool easy to access in every part of the world. Our tool used an accurate system to extract and correct COVID data from over 200 countries, including specific regions within countries, to generate accurate predictions. This solution coalesces with the requirements of underserved people by providing them better details about COVID-19 accurate to their current location and time. It holds a considerable promise to improve the lives of marginalized and underserved communities.

CovARC is also very lightweight and can be accessed with low-bandwidth internet. It enables the tool to be accessible in remote areas across the world. CovARC also ensures literacy and awareness regarding COVID variants and changes in risk of infection with time. This, in turn, helps users take safety measures to prevent the spread of disease and further prevent the virus from evolving into new, dangerous mutated variants. It is independent of a country or region’s ability and willingness to accurately report cases as cases are scaled using Facebook survey data. The user-friendly interface ensures that CovARC can be used on the go for daily activity risk prediction. As a result, CovARC proves to be highly scalable and easily adaptable across a wide range of the population. Furthermore, the robust, up-to-date risk prediction by the tool ensures that it is highly beneficial worldwide, especially for underserved communities.

Our team members comprise people across three continents of the world, each with different ethnicity, gender, socioeconomic, and cultural stratification. We have team members in the US, India, and Europe. The team is representative of different communities, which is why CovARC was developed to be a global solution rather than specific to a particular country or region. One of the team's goals in this project's initial stages was to understand the needs of different regional communities so the problem of COVID can be dealt with in different frames of the world. Various inputs, ideas, and agendas of several communities worldwide have been considered and used to compile CovARC. As the calculator’s active user base grows, having team members situated in different regions will allow us to actively monitor people’s response to the calculator and to reach out to local communities for detailed feedback in order to improve the calculator. In turn, we have ensured that our solution will motivate the masses to face the ongoing pandemic and also educate them about how to survive the challenges.

The sponsorship provided by the MIT Solve Challenge would help CovARC further update the infrastructure of the current prototype, access better paid-data resources, reach users across +200 diverse countries for awareness to help prevent infection spread, and implement our business strategy to grow the organization and beyond. This joint partnership through the MIT Solve challenge will allow us to provide better services to consumers in terms of support required for application and UI development and expand useability. Participation in this challenge will also give us exposure to different peer groups who are working towards similar goals, and will allow us to support one another, grow together and help make the world a safer place by making health resources and information more accessible.   

While starting the development process of this project, we identified several different risk calculators that users presently use in different parts of the world to calculate COVID risk and identify their limitations. From our background research, we generated the following table comparing current risk calculators and created CovARC to address the limitations of most existing risk calculators. More details can be found in our preprint, “A COVID-19 Activity Risk Calculator as a Gamified Public Health Intervention” at DOI: 10.13140/RG.2.2.23583.89765.

By addressing the limitations in the previous risk calculators, we could streamline the risk calculator further so that users could use it for daily life activity risk estimation. Our risk calculator also includes all the countries and regions in the world so that the reach of this technology is not limited to a specific country or region. Moreover, the minimalist design of the risk calculator and easy-to-interpret user interface enables people to easily understand measures they can take to ensure safety from COVID when doing a specific activity. 

Furthermore, our risk calculator updates daily, considering all the new COVID cases daily. This makes it easier and faster to interpret if there will be an outbreak in the future and ensures that users can take precautions early to reduce the outbreak's impact. Our solution provides a state-of-the-art system for everyday activity COVID risk calculation, making it an innovative and flexible solution that users could use in most parts of the world.

1. California covid restrictions. https://www.nytimes.com/2020/12/04/briefing/california-covid-restrictions-warner-bros-stimulus.html  (2021).

2. Centers for Disease Control, C. & Prevention. Covid data tracker. https://covid.cdc.gov/covid-data-tracker/#county-view&list_select_map_data_parent=Risk&map-metrics-cv-comm-transmission=community_transmission_level&list_select_map_data_metro=all  (2019).

3. Covid-19 event risk assessment planning tool. https://covid19risk.biosci.gatech.edu/ . Accessed: 2021-10-29.

4. 19 and me: Covid-19 risk score calculator. https://19andme.covid19.mathematica.org/ . Accessed: 2021-10-29.

5. Jin, J. et al. Assessment of individual-and community-level risks for covid-19 mortality in the us and implications for vaccine distribution. medRxiv (2020).

6. Mycovidrisk. https://mycovidrisk.app/ . Accessed: 2021-10-29.

7. Rowe, B., Canosa, A., Drouffe, J. & Mitchell, J. Simple quantitative assessment of the outdoor versus indoor airborne transmission of viruses and covid-19. medRxiv DOI: 10.1101/2020.12.30.20249058 (2021). https://www.medrxiv.org/ content/early/2021/01/04/2020.12.30.20249058.1.full.pdf.

8. Qcovid risk calculator. https://qcovid.org/Home/AcademicLicence?licencedUrl=%2FCalculation . Accessed: 2021-10-29.

9. Covid-19 risk calculator. https://covid-19.forhealth.org/covid-19-transmission-calculator/ . Accessed: 2021-10-29.

10. Covid19 exposure risk calculator. https://covidtracker.fr/covid19-risk-calculator/ . Accessed: 2021-10-29.

11.  Lelieveld, J. et al. Model calculations of aerosol transmission and infection risk of covid-19 in indoor environments. Int. J. Environ. Res. Public Heal.17, DOI: 10.3390/ijerph17218114 (2020).

12.  Bazant, M. Z. & Bush, J. W. M. A guideline to limit indoor airborne transmission of covid-19. Proc. Natl. Acad. Sci. 118,DOI: 10.1073/pnas.2018995118 (2021). https://www.pnas.org/content/118/17/e2018995118.full.pdf .

13.  Covid-19 dashboard. https://www.northwestern.edu/coronavirus-covid-19-updates/university-status/dashboard/ (2021).

14. Gisaid tracking of variants. https://www.gisaid.org/hcov19-variants/ . Accessed: 2021-11-20.

15. for Disease Control, C. & Prevention. Science brief: Community use of masks to control the spread of sars-cov 2. https://www.cdc.gov/coronavirus/2019-ncov/science/science-briefs/masking-science-sars-cov2.html  (2021).

16. Agency, U. E. P. Epa researchers test effectiveness of face masks, disinfection methods against covid-19. https://www.epa.gov/sciencematters/epa-researchers-test-effectiveness-face-masks-disinfection-methods-against-covid-19  (2021).

17. Covid19 activity risk calculator webpage. https://realsciencecommunity.shinyapps.io/riskcalculator/  (2021).

18. Covid-19 trends and impact survey. https://dataforgood.facebook.com/dfg/tools/covid-19-trends-and-impact-survey (2021).

19.  Gardner, L. Covid-19 data repository. https://github.com/CSSEGISandData/COVID-19 (2019).

Over the next year, we plan on carrying out wide-scale testing for CovARC to continually iterate over the user experience. We plan on using the first year to grow a large user base so that we can identify issues users face while using our technology and address these issues immediately to fuse this web application into the user’s life in a much more organic manner. We would also identify other factors that add up to the risk of COVID for the user. Over the next two years, we plan on implementing a location-based system that identifies the county, region, or country from the user’s location and deploying it in the form of a web application and an android and IOS smart application. We will use this opportunity to link Google mobility data to identify if a place where the user is visiting is busy or has a higher number of people to accurately identify the risk of infection. Over the next five years, we plan to ensure that we have scaled in remote areas of Africa, Asia, and South America, previously identified as hotspots for COVID spread. In the long term, we plan on reducing COVID infections by ensuring people are more aware and take safety precautions in case there is a high chance of COVID infection in their areas. We will run randomized control trials to demonstrate the efficacy of CovARC in reducing infection rates on a population level. This strategy can also easily be adapted to other infectious disease public health crises such as the rapidly evolving Monkey Pox outbreak.

Our current version of CovARC accurately determines the risk of infection, hospitalization, and death for over 200 countries, including regions worldwide. Our system is in-line to be validated by wide-scale testing by people in different parts of the world to identify needs and requirements that could make it more polished and fit for deployment over a larger scale. A smartphone application version of this tool would ensure that people can use it anytime, anywhere, even with limited internet availability. Our team consists of people from several different parts of the world, which ensures that we would be able to test across a diverse set of communities to tailor CovARC to the required needs of the masses. Hence, we are en route toward ensuring that we can achieve the goals we have set in the future.

In the past years, we have observed that during the COVID pandemic one of the most critical problems that caused its widespread was the incubation of the virus in densely populated areas, which led to different variants of the virus that were more resistant and highly infectious. At that time there was no channel that could be used to spread information to help people be more cautious and not spread infection. Main sources of information included major news outlets as well as social media websites. Misinformation spread in many communities across the world leading to worse COVID-19 infection rates. To identify and address this issue, we developed CovARC to ensure that accurate personal risk information is spread to the community in an easy-to-understand way so that it could be grasped by everyone in any part of the world. The streamlined UI provides a gamified feel to the tool, which draws the user in to try out different combinations of vaccines, masks, as well as other parameters.  This allows users to educate themselves regarding the safety measures they can take to minimize the risk of infection. As a result of this, people adapt safety measures or simpler techniques to minimize the risk of infection which prevents the spread of COVID in groups of people. This in turn helps to decrease the spread of COVID. Our calculator, and its methodology also serve as a suitable and reliable blueprint for effective disease risk estimation. It shows how to harness information from various data sources in order to create a streamlined model that considers important factors that affect the transmission of airborne diseases such as the use of masks and the number of people that a person comes into close contact with. We also show how an estimation model can be made to consider different variants of a virus. While we developed this calculator over the course of the pandemic, in the future, in the event of another viral outbreak, by referring to this calculator, researchers and scientists can collect relevant data and can create a similar calculator in the early stages of the outbreak in order to educate people before the outbreak has the opportunity to morph into a large scale pandemic. 

One of the core problems that we identified in the current situation with the pandemic was the delivery of information as well as the access of people to information. During our preliminary research with this project, we identified several false data sources as well as inaccuracies in estimation of risk of COVID infection. This lack of misinformation caused confusion among several people causing people to follow two extremities of either trying to be too safe or not taking any safety measure against COVID infection. The lack of reliable infomation was very severe in developing countries where due to the non-preparedness of the information supply system, the medical institutions collapsed as it was unable to accommodate the rising number of cases and treat everyone affected due to COVID. Our teammates in India experienced this acutely during the Delta wave when oxygen supplies ran out in hospitals. We believe that in order to ensure the safety of people, there is a need for reliable information that has been cross-verified by several sources as well as simplification of the process to access the information. As a result, we developed our core technology around data science and predictive modeling techniques that have been implemented over several datasets that we extracted from different sources. This was an essential step we took to ensure data we extract and process is accurate, realistic and has not been tampered with. In case of abnormalities in the data, we implemented different smoothing techniques to cross-check them and make the data accurate. After having this processed data, we created a risk estimator using references from different researches that were carried out to generate a risk factor. This risk factor was generated as a range to take into account the error margin for our risk calculator as well so that all the information provided by us is realistic and based on valid facts. Upon building this core technology to have an accurate estimation of different risks associated with going out of the home to perform a specific activity, we wanted to provide this information in a user-friendly manner. This is where we used web development to package our technology so that it is widely accessible and easily interpretable by users.

We obtain our data from GISAID variants database, JHU, and Oxford COVID datasets as well as Facebook’s COVID-19 survey. We extract the data regularly, update it using Github actions, and use the processed and filtered data to estimate the risk along with other inputs provided by the user, which are specifically based on the form that users fill when trying to find different risks associated with COVID.

Our team lead, Dr. Christin Glorioso, who is Head of Research has always taken up leadership by communicating to diverse partners and teammates to give better direction to projects. Open feedback creates a more welcoming and open environment, this also helps to put diversity policies in place. The Global Health Research Collective community includes people working together from different medical and technology backgrounds belonging to different countries. This creates a culture that values difference and builds things for a better medical future. 

We are considering two business models for CovARC: grants/joint partners and freemium. Practicing in MIT Solve will open doors for research grants using which we can work on better infrastructure for our prototype and help us gain consumers. We are associated with The Global Health Research Collective, a non-profit organization. The organization's primary goal is to bring COVID-related awareness to consumers and make places safe rather than monetization. Our current focus is on building and sustaining the product. We plan to beta-test the product to gain consumers by attracting a vast target audience. Additionally, finding joint partners with public/private health agencies or governments and private entities like restaurants will help us provide initial grants to build/maintain infrastructure for CovARC to reach 200+ countries worldwide.

As mentioned previously, we are associated with The Global Health Research Collective, a non-profit organization. Our focus is to bring covid-related awareness to consumers by building a sustainable version of our product CovArc rather than monetizing it for profit. We are participating in MIT Solve to get attention to our product. The grant will help us build a robust infrastructure for CovARC for data management, model update, website, and others. We intend to participate in more competitions/challenges or find a joint partner who will be able to sponsor us.

Currently, we are participating in competitions, for example, MIT Solve for potential possible grants for CovARC.  Dr. Glorioso, the founder of the non-profit The Global Health Research Collective is also a 2022 NIA Healthy Aging Start-Up Challenge finalist wherein she is representing her startup NeuroAge Therapeutics. As a team, we will continue participating in similar challenges for financial support and user awareness.