The Power of Big Data and AI in the Fight Against COVID-19

The emergence of a novel coronavirus in December 2019 has had the medical community on edge. Public health officials around the world have faced an uphill battle in treating current cases and preventing the further spread of the virus. The 2002 SARS outbreak, and later the MERS outbreak, did not spread nearly as rapidly as COVID-19 has, which has led the world to an unprecedented public health crisis.

China, where the virus first emerged, quickly used its existing big data infrastructure to contain the virus as much as possible. The success of these efforts has had major ramifications for other containment efforts around the globe. Big data and artificial intelligence experts can work together with public health officials to monitor for new cases, project the spread of the virus, and even enforce quarantines.

Epidemiology and Pandemics

The basic theory behind pandemics is simple: A disease will often spread exponentially, infecting people much faster than people recover from it. Epidemiologists can use computer modeling from existing clinical data to estimate the trajectory of the pandemic.

In theory, there are four groups of people to track:

• Susceptible individuals who have not yet caught the disease but have no immunity to it
• Infectious individuals who have caught the disease and can spread it to others
• Recovered individuals who are no longer infectious or symptomatic
• Deceased patients who must be recorded and analyzed to alter treatment methods and prevent future deaths

With COVID-19’s long incubation period, the number of exposed individuals can and should also be tracked separately from infectious individuals. However, it is a challenge to calculate and react adequately, as most areas are now experiencing community spreading, which is nearly impossible to track. Big data and data engineering could allow significant breakthroughs in the science of tracking community spread.

The “R0” Variable

One of the main variables in epidemiology is R0 or the rate of new infections. R0 indicates the rate of a disease’s spread in a given population. An R0 value of 1 indicates that a disease is spreading at a stable rate, with one patient infecting one other person. An R0 value of 2 is an exponential spread, with each patient infecting two new people.

Calculating R0 in an ongoing pandemic is not simple, as an accurate estimate requires epidemiologists to account for untested individuals. Artificial intelligence and big data about a population’s lifestyle habits can help with the creation of multiple models and account for different possible scenarios.

Unknown Variables

Unfortunately, there are many unknowns with a disease as new as COVID-19. For example, one of the unknown variables in COVID-19 is how quickly children can catch and spread the disease. So far, children seem to exhibit much milder symptoms, so the true number of infected children is challenging to estimate.

Another unknown variable is how many people will catch severe cases of COVID-19. Certain preexisting conditions such as diabetes and high blood pressure dramatically increase the chances of someone being hospitalized with severe symptoms.

However, not all people with these conditions develop severe symptoms. Modeling must be modified to estimate the number of severe cases that may occur, and these figures must be updated frequently as new data emerges. Artificial intelligence and big data provide key infrastructure for these calculations to take place.

Limits of Clinical Data

Data taken from clinical settings can have several errors, all of which must be accounted for by public health professionals. Data entry errors, incomplete patient information, and other problems can hinder public health responses. Big data could be used to correct and complete patient information when needed.

Even with improvements in COVID-19 testing kits and procedures, false positives and negatives are a recurring problem. Although some inaccurate testing results are inevitable, big data and artificial intelligence could potentially help identify trends in these false results or flag them for health officials. For example, if a hospital is turning in an unusually high number of negative test results, an AI program could notify public health officials so they can investigate.

What’s Working and What’s Not

As a pandemic progresses, even incomplete clinical data can be used to determine which treatment methods are working best. Though there are a massive number of potential variables to track, AI can calculate which variables are affecting patient outcomes and which methods are working for the average patient.

This data can further be broken down by patient demographic and preexisting conditions. For example, AI can be designed to track and propose effective treatment methods for doctors treating patients with heart disease or diabetes, two preexisting conditions that can make the prognosis for COVID-19 poor.

Big data may also find that patient outcomes are skewed based on income, zip code, or other socioeconomic factors. This data can be critical to health officials’ attempts to improve treatment practices in overcrowded urban hospitals or underfunded rural hospitals.

Proactive and Predictive Analytics

In the business world, proactive analytics refers to continuous monitoring of supply and demand, with adjustments to production and distribution as needed. In the current crisis, proactive analytics can be used to determine which areas and hospitals need more tests, ventilators, masks, and other equipment. With this information, the federal government, states, and individual companies can identify which nearby businesses need to be redirected to produce certain equipment.

Currently, the demand for tests far outstrips supply, which puts health officials in a dilemma. AI could also help determine which locations need tests most urgently, based on a complex range of criteria. A proactive analytics program could prioritize the distribution of tests to an area that is suspected of having multiple major clusters, for example.

Proactive analytics can also help health officials identify groups of high-risk patients that may need additional communications or proactive testing to keep them safe. Although there are privacy laws that cover individual patients’ data, hospitals and nursing homes may be able to reach out to individuals through a larger collaborative effort with data experts and public health officials.

Tracking Movements

COVID-19 is proving to be challenging to detect and trace, so public health officials are exploring new ways of tracking patients. One potential option is to use GPS data, which could easily be pulled from individual cell phones. While this raises a host of legal questions, all of them could be dealt with through new emergency legislation.

Tracking patient movements could also be incredibly helpful for dealing with patients who break quarantine. China has already used color-coded QR codes to act as virtual permission slips for individuals to enter and exit spaces. This technology was also used to notify people when they had been in the same space as someone newly diagnosed with COVID-19.

In some cases, countries can and should share data and surveillance information, especially for countries that share borders. This could prove critical for countries with strong economic and tourism ties, such as the U.S. and Mexico.

Drones and Facial Recognition

China has a long-standing history of using drones, facial recognition, and other cutting-edge technologies to monitor its citizens. The government’s existing facial recognition infrastructure proved useful in monitoring and enforcing quarantines during the lockdown in Wuhan.

Drones can also be used to scan residents’ temperature to detect potential new cases. The combination of temperature scanning and facial recognition provides authorities with another method for containing the outbreak. Public health authorities could follow up with individuals who potentially have COVID-19 and impose quarantine measures on them.

Obstacles to Implementation

China has proven that existing levels of technology can support monitoring a large population. However, such an undertaking requires massive resources, including server space and programmers who can continuously adjust the software to get better results. Drones and other hardware need to be periodically repaired as well.

The use of big data also hinges on people using smartphones. Some populations, especially older and rural communities, don’t use smartphones as much. Being able to track a majority of residents is far better than being able to track none and may make all the difference in densely populated areas where disease can spread rapidly.

World Health Organization Analysis

The World Health Organization (WHO) has released a report showing that China’s use of drones and facial recognition was largely effective. Since China took strict measures to keep infectious individuals quarantined in Wuhan and other major hotspots, China’s infection rate dropped sharply and there were multiple days in March 2020 without new domestic infections.

One of the key recommendations from the report is that countries with and without confirmed COVID-19 cases should immediately use surveillance programs to detect potential transmission chains. While the WHO did not explicitly endorse the level of surveillance that China used, other forms of surveillance and big data usage can still be effective in monitoring susceptible and infected populations.

Monitoring Industry & Recovery

Satellite data has also been used to monitor the recovery of industries after the initial outbreak. Factories in China have been reopening one by one after each receives clearance from local government officials. Since it can be challenging to track data through China’s local governments, satellite imaging has proven useful in determining which factories are back to work.

Public transportation and highway usage data may also prove useful in the coming months. Since commercial sales data is usually reported with a slight delay, real-time monitoring of public transportation and vehicle traffic may help public officials track and account for increasing economic activity. This can also affect how public health officials deal with future peaks, especially in areas where the outbreak has come nowhere near its peak yet.

Future Peaks and Valleys

The trajectory of COVID-19 is impossible to fully predict, especially since elected officials’ decisions will shape its path. However, those decisions can be influenced by big data testing methods designed in cooperation with public health officials. The resulting decisions’ effectiveness may also be influenced by how well big data and artificial intelligence are used to enforce the decisions and monitor their outcomes.

Removing human error and subjectivity from individual patient cases is never a simple matter, either. The tech industry and big data could potentially alter the outcome of COVID-19 and inform best practices for responding to pandemics in the future. The design and implementation of AI will continue to improve in future health crises, but the sooner data experts rise to meet this challenge, the sooner we all can learn. 

Syntelli Solutions has experience in handling patient data, analytics testing, artificial intelligence, and other data analysis technologies for predicting and improving healthcare outcomes. Our skills and knowledge make us more than ready to rise to the challenge of fighting the COVID-19 outbreak.

Contact us for more on how we can help your team turn the tide of this crisis.

SIMILAR POSTS