What Did We Know And When Did We Know It? Disease Surveillance: Past, Present And Future

(Posted on Wednesday, April 22, 2020)

What can we know, when can we know it, and what can we do with what we know? As COVID-19 moves through our world wreaking destruction upon our social fabric, our lives, and our economies, these questions return to haunt me.

I have recently engaged in lengthy conversations with people whose life work is to develop the tools that allow us to understand what is happening in the world through “open source” information, that is information that can be gathered publicly via newspapers, television, and radio and ever so much more from our increasing wired world. Many of the people with whom I spoke are in commerce, using data to understand consumer trends, habits and desires. Others are from the political world, using publicly available demographic and digital footprints to predict voter behavior and to target political ads. Still others are from the world of intelligence, observing both friend and foe for hints of impending danger. Data gathering, data analysis, and clear courses of recommended action can be inferred by surveillance systems of ever increasing power and sophistication and by people who know how to use them.

My conclusion: We can predict impending epidemics and do so early on. We could have, and very likely did, observe the earliest traces of COVID-19 to accurately predict the gathering storm. Exactly when and how clearly the disaster was foreseen and by who will be, and should be, a subject of intense scrutiny. Surely we in the US and others around the world did not make good use of the information available in plain sight to protect us. My conversations have also convinced me that even now we are not using our resources anywhere near their full capacity, to identify communities in need and to predict with pinpoint accuracy where the next hot spot for infection will be, whether it be next door or halfway around the world. Such information is gold, as we know now how important nipping the epidemic in the bud is to save lives. 

And for the future. Will we learn our lesson? Will we develop the institutions both public (governmental) and private (companies and non for profit organizations) with the capabilities not only to observe but to understand global disease trends and the threat they pose? Will our leaders understand that the natural world resembles a giant machine learning device constantly generating trillions upon trillions of variants of disease pathogens to crack humanities defensive code? Will we remember that nature is the greatest terrorist of all and make full use of the means we have to defend ourselves against this relentless foe?

In what follows I summarize what I learned from these conversations.

Disturbances in the Digital Ocean

Internet connectivity has rendered transparent the behavior, perceptions and even thoughts and feelings of much of the human race.

Almost half the people in the world today have cell phones, and those without cell phones – young children and some in remote corners of the globe, are usually connected to someone who does have a cell. Directly or indirectly, virtually every human on the planet is connected digitally.

Emails, texts, internet searches, social media posts, blogs, news sites, e-commerce, video streaming, publicly available satellite images and web cams, as well as geo-positioning data from digital devices are rivers of information that flow into a vast digital ocean. When major events occur, such as civil unrest, natural calamities or presidential elections, these dynamic events plough through the digital ocean, creating bow waves, wakes, and ripples that can be observed, even when the event itself cannot be directly seen. Disease outbreaks are events with a distinctive fingerprint. They can be spotted on the digital ocean, if you know where to look, when to look, and how to look.

Bow Waves

A bow wave indicates that a nascent epidemic is poised to “go exponential” before it is obvious, even to local authorities, that something is starting to go wrong. Examples of epidemic bow waves that can be observed in most countries include:

A sudden rise in the cost of poultry.

Significant changes in mobility as observed and quantified by geo tracking of mobile phones. Such data is widely available and routinely used for targeted marketing and traffic reports. Less movement in some places (people staying home), more movement in others (people fleeing a hot zone) before it’s locked down, is data readily interpreted by those trained and watching.

Increased internet searches on medically related topics such as symptoms, treatment and availability of western and traditional remedies.

Shortages and online price increases of prescription and over-the-counter drugs, thermometers, and other medical equipment such as masks and personal protective coverings available to the public.

Changes in travel patterns to season action spots.

These are just some of the enormous sources of data that those with an eye on the clouds and an ear to the ground will understand to be the rumbling of impending storm.

Digital Wakes

Digital wakes of nascent epidemics are the faint traces detectable when a few people—usually local hospitals and government officials—know that a serious disease outbreak is imminent. Wakes can be found by observing changes in the behavior or pattern of life of individuals or groups who typically would be the first to know when a new outbreak has started. These people are the “carnies in the coal mine”. Observing their behavior constantly and in real time may seem intrusive but it is consistent with widely practiced commercial and political marketing activities using publicly available data.

The key to success is identifying in advance the individuals or groups in each locality in a position to know what is happening. These include health care workers, government officials and digital “connectors” (influencers, thought leaders with massive social networks). The most promising indicators are to be found in anonymized cell location data that is sold by mobile operators to marketers, traffic reporters, urban planners and others. When a serious outbreak has started, handset data, which can be pinpointed to specific buildings, even floors of buildings in some cases, can tell a compelling story. Here are some examples:

Workers in key disciplines and areas staying late, not going out for meals, working weekends, changing who they socialize with (spatio-temporal relationships among multiple users are possible and have been used recently for contact tracing, social distance monitoring).

Workers in key disciplines and areas changing their travel and sleeping patterns (sleep is inferred from stationary geo-positional data for five to eight hours in residential areas).

Cancellation of medical and government  conferences or absence of key attendees.

Other telling digital early warning signal include:

“Radio silence” of normally active voices on general interest (not health related) blogs and social media (due to preemptive government censorship).

Decreased traffic or pollution.

Decreased power consumption of manufacturing buildings.

Less heat in normally active factories seen from commercial satellite thermal imagery.

Empty parking lots in some places such as factories and malls, and full lots at others such as grocery stores.

Unexplained road and school closures.

A marked decrease in normal discussion about medical issues and disease, due to censorship or fear of censorship.

Websites and blogs going dark

Firings or unexplained absences of key individuals in health care or Government, cancelled leaves of first responders.

Unexplained rise or fall of certain stocks resulting from insider trading of corrupt officials.

Emotional content of postings by influencers/connectors and, in some cases, news reporters. The use of words that connote strong emotion has historically indicated that some locals are aware of big pending social disruptions.

Such a degree of data collection, surveillance and analysis may seem surprising and intrusive to many. From my many conversations, I can tell you such activities are routine, perhaps not always for health, but for intelligence and commercial purposes—certainly.

Ripples

Unlike bow waves and wakes, which occur in or near the region of outbreak, ripples are phenomena that take place at a distance from the epicenter. Examples are:

Decreased bookings for air travel out of a region or even into a region.

Uncensored rumors heard in the outside world from foreigners posted in the affected regions.

Supply chain disruptions in global industries, especially for parts or services specific to a given region.

Conversations and chats online between teleworkers who perform offshore contract work (software engineers, call center operators, customer service representatives). Local governments potentially monitor and censor such conversations, but offshore teleworkers do not always display good “communications discipline” and governments who might monitor the conversations have limited resources to detect and throttle bothersome communications.

Cover-up behavior by diplomats or commercial representatives from the affected region. Attempts to disguise, deflect or deceive regarding the severity of disease outbreaks, are–all by themselves—indicators that something is amiss. Diplomats are trained to use certain language, describe or minimize problems, and those skilled in interpreting this language can spot changes that might suggest a building problem.

Conversations across multiple officials in embassies,  consulates, and trade missions have a sameness to them that denote official “talking points” that have been circulated.

“Holes” in conversations: officials and commercial/trade representatives avoid topics that they would normally discuss.

Unexplained overseas purchase of personal protective equipment, drugs and medical equipment.

Perplexing swings in some commodities markets .

Such a degree of sophistication by trained observers may strike many as an impracticality. Believe me, it is not! When national security or commercial profits are at risk, many governments and large companies are willing to make the training and manpower investment needed— but to date, this is not the case in health.

Digital Buoys

Harnessing the predictive power of bow waves, wakes and ripples is enabled by sensors, digital buoys, that are set afloat on the digital ocean in advance of disease outbreaks to detect disease outbreaks wherever they might occur at the earliest possible moment. It is the early actions which are most effective in containing a disease threat and minimizing its impact.

For example, to sense ripples in changes in behavior of key health care workers and government officials, they must be identified in advance and their normal daily behavior recorded and analyzed. Such analysis, I am told, is commonplace for intelligence services and commercial entities that have long standing customer relationship marketing deals in place with mobile operators so that location and movement data of relevant groups can be observed and analyzed.

Similarly, analytics and tools must be developed and pre-positioned because big data analytics, machine learning, and pattern recognition artificial intelligence systems cannot be brought on line instantly. It takes time to train people to interpret and analyze a flood of data. These people will be digital first responders. Training by observing disease information flow from disease outbreaks in the recent past is essential. There is no better training set of data than COVID-19. Real time tests of predictive models are also necessary. For that, there is no time like the present.

Many of my interlocutors warned me against danger of over reliance on fully automated analysis. Forming insights from digital data is very labor-intensive, requiring the knowledge and experience of highly skilled staff. Digital collection, archival and analytic systems can be developed and pre-deployed, but humans are needed to operate and analyze these systems. They must be in place before the next epidemic. The necessary knowledge and experience cannot be acquired on the spot.

From Information to Action

Everyone I spoke with emphasized that the most important part of the equation is communication, making sure that information leads to productive action. We all know the myth of Cassandra. Her curse was to be able to see the future, to be heard, but not to be believed. Belief is the necessary precedent for action. Remember the fate of the Trojans. Cassandra correctly foresaw that there were Greeks in the gift of an outsized horse. Priam listened to her. He did not believe what he heard. He did not act and so lost his country and his life.

A robust conduit for transfer of information to decision makers is of utmost importance. So too is the credibility of the messenger. Perhaps most important of all is the selection of leaders, by whatever process a country may choose, that are receptive to hearing and acting on information in the interest of the people and the economy they serve. Yes, leaders the world over are besieged daily with an overload of data and must make decisions based on imperfect knowledge. That is why the information and analysis they receive must be of the highest quality. Both our commercial companies and intelligence services know how to plumb the depths of the digital ocean to extract the information they need. These tools must be fully deployed for disease surveillance so that our lives and economy are never again placed at such risk.

Read the article on Forbes

Originally published on Forbes (April 22, 2020)