Pandemic Advanced Warning and Containment System

Problem

The economy is facing an unprecedented collapse caused by lock-downs. Every economy is a dynamic system that needs constant movement and circulation of people, goods, money. What if we could end the lock-down immediately and lift restrictions on the economy by using A.I. to process large amounts of data needed to asses risk and predict the spread of the virus?

Solution

Pandemic Advanced Warning and Containment System would use advanced mathematical models, tracking technologies and A.I. to find patterns that contribute to the virus spreading at higher rates. The system would use models based on capacities of health systems and the factors that affect the virus reproduction rate to manipulate the spread of the virus among the population.

Instead of locking down entire societies and causing massive damage to the economy and mental health of the population, we could ensure that the whole economy can operate without radical restrictions.

Hypothesis

Theoretically, it is possible to mathematically determine the breaking point of any healthcare system in the world, and to predict when this breaking point would occur.

Factors we need to consider are:

fatality rate
rate of hospitalization
rate of hospitalization in ICUs
rate of patients that require respirators
length of hospitalization
capacity of the healthcare system
individual risk factors
social and culturological factors
virus reproduction rate

Fatality rate increases with age and serious health conditions. Older population, especially over 80 years old, makes for the majority of deaths from COVID-19. If we look at the available stats, there is a very low fatality under 50 and almost none under 30. Most of these have had severe health issues and can thus be used to calculate individual risk factors. Source: vox.com

Official stats for hospitalization rates, usage of ICUs and respirators support the same conclusion: older you are, higher the risk. Sources: businessinsider.com, statista.com

Length of hospitalization varies, but CDC states that median hospital stay is 10-13 days. Source: cdc.gov

Healthcare infrastructure is the next factor to consider. Number of hospital beds per 1,000 people, ICUs, respirators, qualified staff, how much of the system is used in regular circumstances, etc. Also, it is important to know what is the expansion potential of the healthcare system.

Individual risk factors are important because they influence fatality and hospitalization rates. Fortunately, there is already a lot of information available. Sources: patient.info, cdc.gov

There are different social and cultural factors that have to be accounted for before implementation. Different countries have different customs and problems.

The reproduction rate of the virus is affected by all factors, but it also affects them as well. It is the single most important variable of the system. The first steps to reducing the rate have already been taken all over the world by educating people about responsible behavior.

Conclusion

We can manipulate the reproduction rate of the virus by educating people and providing personalized recommendations, and keeping track of those infected, isolated and quarantined.

How it works

Tracking

The system uses tracking over mobile devices to control the movement of the isolated and infected. When someone is tested positive for the virus, their contacts are "unlocked" to find which are at higher risk and thus need to be placed in self-isolation or tested.

Risk assessment and load management

Using massive amounts of data about people's health, habits, living arrangements, family members, work, personal interactions, etc., the system can create risk assessments both for individuals and groups alike, as well as specific locations, zones, public transport lines, etc. This data would be shared voluntarily and used for prediction of future positive cases to effectively manage load on the healthcare infrastructure.

Interactive mobile app

The app would use health info and other data about each user to issue personalized recommendations on how to avoid risky zones, reduce personal contacts, etc. It would also send announcements from the authorities, new discoveries about the virus, general recommendations, etc. Level of personalization depends on the amount of data each user is prepared to share with the system.

Privacy

The system is based on the presumption that people will make the responsible choice and choose to give up some of their privacy to help fight the virus. Only mandatory tracking would be of those infected, isolated and quarantined, as we believe that this does not invade their privacy.

Anonymous data verification

In the Proof-of-concept, Firebase authentication will be used, allowing Apple and Google sign-in which is already allowed by Android and Apple users.

SID - Security ID created by the user’s application
Datetime - datetime of the message created by the user’s application
UID - Unique ID created by the backend
Contacts - Bluetooth contacts nearby
CID - contact ID (BT contact)

We will use the user's smartphone to store data, but we are verifying that data is saved and the smartphone is active. Location based data will be shared only if the user allows this behavior or the user is COVID-19 positive. User’s application will ping the API every 5 seconds if the user is out of his "home" location.

On the server-side we will get the user's data, sign it with the server secret and send back the UID.

Data integrity will be secured with the chain, the previous link’s SID will be included in the new one. Server will be sending a unique ID every once in a while and for the first initialization.

Process of preparing data and creating SID on client’s smartphone: SID = sha1 ( "previous SID" + Datetime + client secret + UID)

Verifying data and sending special identifier for hashing from the server-side: UID = sha1 ( client_id + created_at + server secret)

If a user gets the positive COVID-19 result, the application will run a PUT method for each SID and will update contacts on the server-side if the user didn’t send them already.

Prediction models

To predict effect on healthcare system:

Dynamical simulations based on obtained data of interactions
Temporal dynamic network analysis

To predict chance of individual to get sick:

Standard time series models
Experimental recurrent neural models

Implementation

Tracking of the infected, quarantined and isolated is a must, as well as having a way to precisely backtrack their contacts. The risk that someone might not share every contact, either intentionally or non-intentionally, is just to great to leave it up to chance.

Everything else is voluntary. The more data the system gets, more accurate its predictions will be. In theory, if everyone decided to share everything, the system could make predictions accurate enough to fully automate the entire society. In practice, it could use common characteristics of those who shared enough data to create cohorts and provide applicable recommendations. First stage is to input all the data that is already publicly available about the virus.

Education and building trust among the population is crucial in the beginning. People need to be informed and governments need to be transparent about what they need, why they need it, and how they are going to use it. Technological methods of securing personal data would also be applied.

Future impact

We already see that impacts on society and economy this pandemic is going to have will be long-term. Terms like herd immunity, virus reproduction factor, social distancing, etc. that have been used in scientific circles are now part of mainstream communication. Robotics are being applied extensively to help medical staff, thus increasing the potential of healthcare systems. A.I. is being implemented to tackle various issues. Sources: mygreatlearning.com

If we are going to treat this as a war, and such analogies are common in the media, then every individual is a soldier. Soldiers need to be drilled, trained, and prepared. Education and social cohesion is the only way to do it. People in a certain area will be motivated to share more data if they see that it produces better results of containment in other areas where more people shared.

Investments in hospitals and healthcare infrastructure will increase. Robots will take over some jobs that carry a high risk of spreading the disease. People will see the benefits of A.I. and robotics and these will become an integral part of our daily lives. When the next new virus comes, and it will, we will be ready.