CoVis: real-time disease risk assessment

A highly accessible centralized system that tracks communicable diseases is lacking. While a lot of data is collected by several global and national organizations, its curation and accessibility for individuals, corporate entities and NGOs have not been built as has been made evident by the ongoing pandemic. 

At CoVis we will build this system by consolidating data streams, building AI-driven forecasting algorithms and incorporating scientific knowledge to understand the risk of diseases and monitoring potential outbreaks. As a first step, we have built our flagship app, CoVis, providing real-time risk assessment for individuals using intelligent algorithms without compromising data privacy. CoVis is being launched for iOS/Android in Germany and the USA.

Our solution can provide transparent and actionable insights into the imminent risk that one faces from disease helping people understand how their behaviour affects them. This will guide them towards safer behaviour and decisions that will help society collectively.

We are solving the problem of global disease risk management for communicable diseases. This will address the "Strengthen disease surveillance, early warning predictive systems, and other data systems to detect, slow, or halt future disease outbreaks." are of the challenge.

Our solution starts with addressing COVID-19 risks for individuals and our first product is an app for Germany and the USA available to about 70% of the population of these countries. We plan to rapidly launce in several other countries. Our solution includes disease surveillance and forecasting systems. We want to further build anomaly detection systems that will serve and early warning predictive systems that will slow or halt future disease outbreaks.

The CoVis app provides an assessment of location-based risk from COVID-19 for individuals. We collect data for the spread of the disease and make forecasts at a highly granular level geographically. We collect knowledge from emerging medical literature to build parameters that affect individuals of different demographics, health conditions and habits. We then send all this data every day to the app on the phone.

Since data privacy is important to us, we do not collect user data or their location history unless they choose to store their location data on our servers for device portability through the creation of an account. Instead, we send all the data we collect in a well-curated manner so that the Bayesian algorithms built into the app can compute a personalized risk score.

This is just a first step for CoVis. We want to build a global disease risk management platform that will monitor and forecast multiple communicable diseases and bring these insights to individuals, companies, government organizations and NGOs. We will leverage heavily on explainable AI and consolidate several data streams to achieve this. We will also recruit public health, medicine and life sciences experts to guide us in this endeavour.

The target population is the population of the world since a disease spread potentially affects everyone as is evident from the ongoing pandemic. However, such social and economic shocks affect the marginalized and poorer parts of society preferentially. Hence, a significant goal for building CoVis is to be able to reduce these burdens from the shoulders of those who can bear them the least.

Parts of the CoVis team has already studied and published (as a pre-print) work on the impact of COVID-19 on minorities in the USA [1]. Part of our goal is to use the understanding that we have gained from data to customize and deliver solutions. 

[1] A. Paul, P. Englert and M.Varga, Socio-economic disparities and COVID-19 in the USA, [Link] (under peer-review)

CoVis will contribute to strengthening disease surveillance and building early warning systems based on consolidating data streams. Our explainable AI systems will look for anomalies in population health metrics globally which will directly contribute the reducing the burden of communicable diseases in the future.

At CoVis we want to make this information accessible to individuals and organizations so that they can make informed choices on behaviour and policies leading to a society that is far less susceptible to disease outbreaks at the scale at which we have seen in the recent past.

The flagship CoVis app is waiting for final approval by the Google and Apple app stores. Being a COVID-19 related app there are very strict regulations for publishing an app (it needs to be backed by a government entity and cannot be used for monetization). While we meet all the criterion, the approval takes far longer than a normal app since we are not using contact tracing.

Meanwhile, we have been extensively testing the app since the concept of real-time location-based risk assessment is very new. We have not been able to allow open testing since the app is still not approved by the store. Hence, we have been performing closed testing with about 40 users spread primarily over the USA, Germany and Egypt. Our target communities are citizens of the USA and Germany, however, we are also testing for cases where the users travel.

The CoVis concept drives at making information on disease spread more accessible and presenting analytics in a way that is more actionable. There are several sources of existing data but they are not centralized in one system that is dedicated to disease risk management. Moreover, data streams are not consolidated for low and middle-income countries which are most affected by communicable diseases.

Secondly, while experts know where to find data and understand what it means, very little attempt has been made to process this data in a manner that is understandable by the population or entities that do not have access to a team of experts like small businesses or small-medium sized companies. This is where CoVis wants to step in. We want to build a bridge between the available data (also creating proprietary data streams) using expert knowledge and build actionable analytics that can be well understood by our target audience to enable them to make informed choices.

Thirdly, while AI-based forecasting algorithms have been widely used in finance, climate and consumer applications, they have not been well developed for monitoring disease outbreaks. The primary hurdle has been that this application is not seen as profitable. However, given the financial burden brought about by the ongoing pandemic, we are expecting that real-time global disease risk management will be seen as more instrumental in reducing the costs associated with a shock from such incidents in the future.

current number: about 40 testers

target number in one year: 1-10 million

target number in five years: 10-100 million

The numbers are an estimate based on:

1. the number of people that can potentially be using the CoVis app in a year or 5 years

2. the number of people that can be indirectly benefitted from our solution when adopted by corporate, government and non-government entities.

Once the CoVis app is launched we will conduct several surveys including a dictionary study of how it helps people change their behaviour. This will allow us to improve the UX/UI for the app and the algorithms that are used for the risk score computation. We will partner with universities to conduct these studies and will be an extension of the surveys and interviews that we have already conducted. Thomas Ribero, the CPO of CoVis is an expert in human factor engineering and will lead the studies. This will help us understand how well we align with the "Good health and well-being" UN SDG. More specifically it will help us understand how COVID-19 has affected individual lives and what CoVis can do to mitigate some of the problems faced by individuals.

In a longer-term perspective when CoVis is capable of forecasting and monitoring multiple communicable diseases, also in low and middle-income countries, we will have to understand how we calculate the impact that the CoVis platform has on different parts of the society. The impact from and impact on mitigating disease spread is difficult to measure given several confounding factors and counterfactuals rarely work and repeatable experiments designed to understand the impact rarely possible. However, it is well known that the right information and efficient organization are the keys to reducing the impact of diseases. Our progress will be measured by the amount of information we can gather and the degree to which we can convert it into actionable insights.

part-time staff:

Ayan Paul, PhD (CSO)
Rebecca Sereda (COO)
Jonathan Seward (CEO)
Geoffrey Blanding (President)
Thomas Ribero (CPO)
Christian Suharlim
Jessica Shen
Nichola Tran
Natasha Sham
Valeria Martínez
Melanie Muñoz

Contractor:

Sidebench LLC

The six CoVis founders came together at the MIT COVID-19 challenge in April 2020 with a common motivation of developing a digital solution for assessing risks from COVID-19. The team is multidisciplinary: Bayesian Inference, AI/ML, Public Health, clinical research, UX/UI design, Robotics and automation, backend architecture and development, cloud computing, marketing and social media outreach and administration for a healthcare tech company. The team is multicultural representing several countries and located in three different countries. Covis Inc. was formed remotely and we have built the CoVis app remotely. We share a common vision for the future of Covis Inc. and would like to see a consolidation of disease risk assessment at a global scale. The team is working part-time and contributing hours to build the solution since we decided to channel all the funding to build the app as quickly and as best as possible. Once we raise enough funding to pay salaries, several of us will work for Covis Inc full-time. Our cultural and intellectual diversity is our strength and we are building a people-centric company building solutions that will leave a large-scale social impact.

The CoVis team is built on diversity since it is spread over several countries including Germany, the USA and Mexico. In addition, we are working with developers that are based in the USA and Egypt. Having started with a global vision and a team that defied international boundaries diversity, equity and inclusion lie at our very foundation. 

The CoVis leadership team does not have a vertical hierarchy and we believe in collective decision-making. Withing the current CoVis team, we have nationals of Indonesia, the USA, India, Mexico, Ukraine and Brazil. We always try to be gender-neutral promoting the role of women in leadership and gender-biased professions like software development. We are happy to have two female founders, one bringing innovation to the CoVis leadership as the COO and the other bringing innovations to brand design and UX/UI.

Moving forward, CoVis wishes to be a representation of the populations that it serves bringing experts from various nations and minority groups. This is essential since disease spread differentially affects various parts of the society and local knowledge is very important to assess these impacts. Disease risk assessment requires knowing the ground realities, especially where minorities and the marginalized part of the population are involved. At CoVis we fully understand that to reach our goals of helping build a society that will be less affected by disease spread we need to include those that are most affected in addition to those who have the best expertise to address these problems.

Funding: CoVis being a social-facing solution requires initial funding to become sustainable. We started our work during the pandemic with the primary goal of bringing to society something that will help mitigate the current crisis. We are hoping that this attitude resonates with the purpose of the MIT Solve Challenge and that we can fund like-minded investors who will support our cause

Mentorship: We are a new start-up and while we are all experts in our individual fields, we will need help in making CoVis a self-sustainable company. 

Network: We will need to make our products and services be known and this can be best done with a large network of people who care about our solution. We will also need to gather more expertise as we grow and are hoping that the network will help us reach the right people.

Human Capital: We need experts in public health, ML/AI, life sciences, medicine and software development preferably from all over the world so that they are connected with ground realities in their part of the society.

Business Model: We need help with strategy development and the identification of early adopters for our solution. 

Legal or Regulatory Matters: CoVis does not have a legal team and would like to have the ability to consult legal and regulatory experts since we are dealing with individual data and also with health data at the population scale and there are different regulations in different parts of the world that we need to follow.

Product / Service Distribution: While we have a relatively clear idea as to why our solutions are useful, we need to be able to find the right audience to sell our services to. 

We already have expertise in the other fields but we are open to growth and additional help even in those areas.

Data: With vast amounts of mobility data related to locations and populations, Google can certainly help us in developing solutions for diseases that spread through contact and proximity like COVID-19. We are using data from Google in the CoVis app. For example, access to the popular times data collected by Google would allow us to very accurately predict risk at any location.

AI/ML: We can benefit greatly with AI/ML knowledge from giants like Facebook, Microsoft, Google and through academic collaborations with university partners in the USA, the UK, Singapore, Australia and China.

Human behaviour: In order to build a comprehensive risk assessment platform, we will need to understand human behaviour and their response to disease spread. The Behavioural Insights Team can help us with that.

Computational resources: We can also benefit from contributions in terms of computational resources from academic institutions or private organizations. Help with cloud services from Google and Microsoft will be highly appreciated.

The CoVis app is well fitted for this prize. The app aims at delivering real-time location-based disease risk assessment to individuals in a way that they can make informed decisions and be driven towards safer behaviour during a disease outbreak. This will not only reduce the impact of diseases on individuals in terms of sickness and additional stress but also collectively move society towards a lesser disease burden. It is known that the ongoing pandemic has disproportionately affected the minorities in the USA [1]. A sizable fraction of the cause has been the hurdle of disseminating information during the initial phase of the pandemic. This is a problem that we wish to solve. 

The CoVis app is initially being launched in the USA. The CoVis team is determined to address health inequities by making information readily accessible and relatable to all parts of the society including minorities and those who are faced with social disadvantages.

[1] A. Paul, P. Englert and M.Varga, Socio-economic disparities and COVID-19 in the USA, [Link] (under peer-review)

The CoVis platform is based on AI. The CoVis app uses disease spread forecasting to calculate the risk that is faced by individuals from the ongoing spread of COVID-19. We have built a system of the highest possible geographical granularity and continue to add countries for which we make forecasts. In the future, we will use multimodal data to make such forecasts and also use correlations between different diseases to be more accurate. 

At CoVis we believe in explainable AI and interpretable machine learning frameworks which are central areas of focus for Ayan Paul's academic research. CoVis is built on technology transfer from scientific research and will continue along this path of bringing cutting-edge science to the society built by experts who have developed these technologies. Since our goal is to understand disease spread and not only make predictions, the explainability of the statistical and mathematical framework we use is of prime importance since it will build the Human-Machine interface that will be optimal for building robust AI.

CoVis being a social-facing platform, the AI we develop will help humanity in protecting itself from diseases in the future. Not much has been done for building AI-based anomaly detection systems for monitoring disease at a large scale. This is the problem that we would like to address.