I am not a health care provider, but I did marry one. It has always made me interested in the intersection of analytics and healthcare amid the analytics revolution and more so during the pandemic.
How would you feel handing over your (unlocked) phone to a complete stranger? Would it make you uncomfortable to know your phone knows everywhere you have been? Every photo where the built-in AI tags people’s faces also captures the location the picture was taken. Unless you have explicitly turned off location services all the data is there to show where you have been. This is really important to the tracing of the COVID-19 virus and how it spreads from person to person and remains a gap in contact history tracing.
Contact history is largely gathered by interviewing people with the virus and asking them to personally recall where they have been and who they were within the last few weeks. It is time-consuming and bias by our recall of places and events. Let alone the people we interact with that we do not know by name or place, like the people at the coffee shop, the grocery store or on the bus.
Privacy for One and All
Amazingly with over 1.2 million cases in the United States, there is no public data about where those people were 10 days before they were confirmed with the virus. Where did they go? Where did they catch it? Were they on the same subway car as you and others? Did they shop at the same food stores? Knowing this data is not meant to cause a panic, but with a limited number of testing available, it could help to optimize whom to test next. Optimization of a limited resource is ripe for data analytics and should be guiding the medical community toward getting ahead of the curve of infections and target those most probable to get the virus. One barrier may be how we view the privacy of our phone data in the context of how it could be used (anonymously) to benefit the many.
How can data scientists and analytics guide future testing decisions?
Early Approaches
China is taking an approach using a phone-based-app ranking your infection risk level, assigning a color code. Innovative companies such as Accent Systems created a Bluetooth wristband aimed at warning you when you are within 30 feet of a positive-COVID person. The Singapore government has launched an app called TraceTogether, which is similar to the Apple/Google COVID project. Australia’s app is called COVIDsafe and the UK has one under development too. These apps will notify you when you have been in close contact with someone who has previously reported themselves positive for the virus. The shortfall here, of course, is the approach is not backwards looking in time and since symptoms do not appear for 10+ days, it may not be as helpful as digital tracing the location history of positive patients.
How will tests be optimally allocated when the testing shortages are over?
To be clear, I support how testing is currently done. Following CDC guidance, those in priority situations (example health care workers) and those with symptoms are tested first. How will tests be optimally allocated when the testing shortages are over? This is where data science meets medical science and, with the right data perhaps those tests can be given to people most likely to be infected next.
Note: My views are my own
derekwgibson 