When COVID gets personal, look to evolutionary AI (Part 5)

My son graduated from college in March, but instead of pomp and circumstance, the graduation ceremony was a short Zoom call in which the dean of his college cried for much of it. Later, our vacation plans went down the drain due to the intense travel restrictions.

As cases and deaths rose during the months that ensued, I counted my blessings that these were my biggest tragedies in the pandemic.  But my bubble burst when I got a call from my aunt's husband, who revealed that my beloved relative had suffered a stroke, was hospitalized and wasn’t allowed visitors.

As someone who has built a career touting how artificial intelligence (AI) can be empowering, I felt helpless, not knowing how to plan for the future or what would happen next. To add insult to injury, major media outlets were publishing articles, asking the very valid question: Where is AI in the fight against the virus?

Through all of this, our work life was hit hard, too. Not being able to congregate with the team at Cognizant AI's beautiful San Francisco offices was hard, but it inspired our team to seek ways to better understand when our office could safely reopen. How did they do this? With Evolutionary AI. 

Beyond predictions, to prescriptions

Using AI and machine learning to predict the trajectory of an epidemic is not new, but these predictions often prove inaccurate in the face of interventions and other unforeseen developments. At the beginning of the coronavirus pandemic, for example, even health authorities didn't advocate the wearing of masks – they just didn't know the efficacy. As a result, a model trained just one year ago would not remain accurate once mask policies were adopted. Our Pandemic Response Challenge, launched with XPRIZE, is geared toward exactly this issue. 

Evolutionary AI, however, is adaptive: All models within the system are updated daily as new data becomes available. This adaptive quality also allows for the introduction of new data features to the system. For example, our system was up and running in May, when testing and contact tracing data were not yet available. By July, however, we received that data, and overnight, the Evolutionary AI system factored it into its predictions. The same was true later on, for masks, and by late last year, for vaccine data. 

Our system was empowering, yes, but what it told us was not very encouraging: It forecast a high rate of infections and deaths. Beyond those predictions, however, the system is also able to prescribe a schedule of policies on a regional basis that balance pandemic containment with ways to minimize economic impact. These prescriptions can and should inform policies that achieve the best balance of virus containment and economic cost.

This is how, by May 2020, it became clear to me that for California where I live, under any reasonable assumptions, workplaces and schools would remain closed for at least the following three months.

Taking it personally, responding practically

Of course, I'm writing all this from my own limited and somewhat self-centered perspective. My woes and annoyances are nothing compared with what many people near me and around the world suffer every day in this pandemic. But every individual is important, and empowering us individually to be able to make the best decisions can result in a healthier and happier society.

Imagine if we could use Evolutionary AI to help with our own personal decision making in the context of a changing world? I for one, would love to be able to know when I can visit my dear aunt, or go back to our office, or to reliably plan my next vacation.

Learn more about our Pandemic Response Challenge with XPRIZE, a $550k, four month challenge that focuses on the development of data-driven AI systems to predict COVID-19 infection rates and to prescribe intervention plans.

Take a look at the finalists, the winners and the technology behind the Pandemic Response Challenge, or see Parts 1, 2, 3, 4, 6, 7 and 8 of our blog series.