COVID-19 contact tracking — Solving the tough problem

Infected contact discovery is critical

A key to stamping out COVID-19 will involve tracking contacts by those who are infected –If social distancing can be maintained and when  patient testing is available. Tracking and testing can define the next group of infected people.  The process repeats: isolate, test, track, test, isolate….  This process takes lots of effort. China used about 10,000 people to track the cases in Wuhan with a population of 12 Million, about 40% more than New York City.

There is help from technology.  Tricks for discovering cellphone and computer data are used every day by computer forensic investigators.  Here’s a (blurred) map recovered from an iPhone that illustrates what can be discovered.  Each point on the map includes time which is accurate to a second and location that has GPS precision.   This sort of investigation may give useful help to pandemic contact trackers too.

tracking COVID-19 contacts

Cellphone tracking gives precise results

Finding locations

Just checking a smart phone location history is a start.  This can be done in the phone or by going to sites like MyActivity.Google.com. iPhone users can look on their phones at Settings > Privacy > Location Services > System Services > Significant Locations.  We get much more comprehensive location details in our investigations with software such as Cellebrite UFED or Magnet Forensics Axiom but it takes longer .. cellphone connections to WiFi or cell sites, navigation apps, photos or browser searches give rich location detail.

All of these are much more precise and much more dependable than having a patient guess about what was happening last Wednesday morning.

Computer users leave location clues too.  It is not unusual to have a million events in a user timeline, many with locations from email, map lookups, etc.  I use X-Ways Forensics or Sumuri Recon to mine this information from PCs and Macs.  When examining a crook’s computer the user’s intent jumps out of a forensic analysis — like in the killer’s unbelievable search for “kill spouse“.  When searching for contacts by an infected person the searches could be a Bing map search for “latte on main street”.

Big data — recovering history

Tracking power is amazing when cellphone apps and data visualization are combined.  This is seen in the Business Insider  article which shows the spread of beachgoers last month.  This was about a couple weeks before Florida’s stay-at-home order but well after the pandemic situation was known.  Of course, X-Mode Social  and Tectonix emphasize the protection of user data by anonymization and encryption… but that doesn’t protect the anonymity of someone whose home in Oklahoma is shown.

Exact data are not needed for identification of a traveler.  A classic example of this is the merging of public photos with “encrypted” public data. The encryption in this case used an easily reversed process.  The whole process, celebrity photos and comments on taxi tipping are contained in the interesting Gawkter article . Photos connected faces with taxi numbers which yielded destination information along with payment details.

These trivial examples illustrate the complexity of protecting data.  Risk calculations will need to balance the statistics of deaths with the risk of privacy loss.

Using live location data — a promise with a problem

Real time tracking apps are being developed  by Germany, Iceland, China, Singapore, Israel, South Korea, US CDC. These techniques have the possibility to warn others when they are near a known-positive person.  Imagine that everyone has cellphone Bluetooth turned on and that a central database knows which MAC addresses are associated with infected phone users.  Discovering who is nearby is just a matter of signal strength monitoring.

Complex questions on: privacy, medical records, and false alarms will need to be resolved.  That should be no problem in countries run by  authoritarian systems.  The rest of the world will need to carefully think through the implications.  Does your family want to be tracked?  Will undercover spooks like being tagged?

Despite the downside, this sort of fast tracking might be the only solution where disease spread is too fast for manual tracking.  This article highlights work on NextTrace about a community effort to manage pandemic tracking. The NextTrace effort recognizes that “Once widespread community transmission occurs, there are simply not enough resources to keep up.”

Management issues

An overview of the recent pandemic situation (with a little math) can be seen in Kevin Systrom’s blog.  The key is pretty clear, pandemic intervention can work only when follow-up is faster than the spread of infections.

Of course there have been politicians who have mocked this process during the current COVID-19 event

— In New York:  “As a matter of fact, I’m going so far that I don’t even think you can do a state-wide policy.”

— In Alabama:  “Y’all, we are not Louisiana, we are not New York state, …“Right now is not the time to order people to shelter in place.”

— In US:  “Because of all we have done, the risk to the American people remains very low”

Early COVID-19 patient isolation and testing has an astounding effect on success.  Comparing California and New York state results:

— CA: statewide shelter in place order when the number of cases hit 1000
— NY: statewide stay at home order, delayed until cases per population were 16 times higher than in CA

Systrom COVID-19 comment:  “I worry the hesitation–if only for a few days–in New York might be one of the largest public policy mistakes of our generation.”

I would like to thank:

Dr. Ted, the big data guy who really does understand trends in the world around us.
Dr. Adele, whose epidemiology and genetics expertise helped me begin to catch on to the importance of testing.
Dr. Vija, a caring and generous person who has a solid commitment to patient safety
Dr. Liqun, a physician with a breadth of wisdom and the practical skill to make important decisions.