Share the Data, Not the Virus
What to do in troubled times? Typically, take action, but this crisis calls for inaction. United we sit. And we watch as the coronavirus tears through friend and foe, capitalistic and communist, Republican and Democrat. Ironically, we might be a more united world than ever before, but the world is working together less than it must to “flatten the curve.”
Right now, some of the most successful weapons we have to fight this are soap (dating to Mesopotamia); blood transfusions (critical in our response to The Spanish Flu); cloth to cover our mouths (an innovation from pre-history); TV to share the need for social distancing (invented in the 1920s). All of these technologies have certainly been improved over the decades, centuries and millenniums, but the question stands: how are we using modern inventions in the great corona conflict of 2020? To be clear I have zero doubt that scientists, researchers, doctors and many more are working untold hours to develop better tests, therapeutics, and vaccines. I am asking about IT, analytics and software — how are we using digital inventions?
We are losing the information war against the virus because it is sharing data better than we are. A virus shares all of its genetic code as it replicates itself in the host. Through small changes in genetic makeup and the process of natural selection, the virus will evolve. The good news is that the virus doesn’t appear to be rapidly mutating, but reports suggest multiple strains with varying degrees of infectivity and severity. In some ways, genetic mutations and natural selection are nature’s form of artificial intelligence.
There are many applications for AI to fight the virus from developing therapies, vaccines, predicting hot-spots, but one simple question is: how bad will it be when I get it? With over 300,000 confirmed cases, we should have enough data on similar patients to anticipate how individuals will react. Every body is different and we will all react differently, but why not use our AI tools proactively stratify patients by risk? Instead of testing and ventilating, let’s anticipate how individuals will react.
Share the data, not the virus. By combining AI and a patient’s medical history, including all of their vaccinations, procedures, diagnoses, lab results, genetic, chronic conditions and notes, we can better identify those that are the highest risk. This risk stratification could be what we need to restart the economy.
Two trillion dollars is a towering amount of cash, but I don’t think even the full faith and credit of the United States is enough to make everyone whole for the entirety of this epidemic. If we shut the economy down for too long, we could risk widespread default by private and public institutions. This will almost certainly not happen, but many of us will need to go back to work soon. Health information sharing and AI could identify those that should go back to work and those that should remain in quarantine. Whether it’s going back into the office or finding a new series or book, we must all do our part.
I think a major reason why we aren’t able to leverage AI is that we don’t have access to the data. The providers, payors, and labs control the majority of health data. There are many reasons why data sharing in health care is hard. There are few financial incentives to share information because everyone is competing for the patient/customer’s care/business. Why would Coke share its secret formula with Pepsi? There is a lot of complexity in health data and each IT system have a slightly different way of storing the information. Different stakeholders have different pieces of the puzzle — providers, insurance and labs all have semi-overlapping information but individually incomplete datasets. Last but certainly not least is regulation and security. This is very personal information and it is rightly guarded by regulation and tight security (hopefully). While there are many hurdles, we can do this.
Regulation that could prove to be very important was passed earlier last month that is going to help give patients access to their own medical information. Deidentification, safe storage, and sharing protocols can protect the patient’s private information while training AI algorithms. We must facilitate the exchange of health information to anticipate case severity and develop person-specific work and care protocols. We must be united in our fight. We must share data to predict how individuals will respond to the virus. We will be able to protect those at risk, send people back to work safely, anticipate outbreaks and allocate critical resources. To beat the virus we must think like a virus.
