The COVID-19 pandemic has shaken our systems to their core. With the virus still resolutely set on its death march, we have, as a global society, been left reeling. There is simply not enough testing capability anywhere in the world. We are not able to test enough people and we definitely won’t be able to test everyone. Can big-data help us predict the spread of this coronavirus? Read our new editorial piece to find out about how maps built on fever data are being used as a rough-proxy for tracking coronavirus spread and why ONiO’s continuous fever monitoring technology might prove invaluable in such a scenario.
At the time of writing this piece, there are more than 1.4 million confirmed cases of COVID-19, the world over. It feels incredibly difficult to believe that just 4 weeks ago, things were pretty “normal” for most of us. Seems surreal to remember that just last month, universities, offices and schools were open and football matches were being played.
The seriousness of this novel coronavirus outbreak and the graveness of the threat it poses to humankind have already been spoken about to no end. On the other hand, this crisis seems to be bringing out the absolute best in many people.There have been numerous displays of courage, kindness and ingenuity from people across the globe. No matter the country, people have shown up with great responsibility, awareness and courage.
The camaraderie, solidarity and “can do” spirit that is being displayed is absolutely something to take hope from. Nevertheless, the grim reality of the matter is that the virus is still at large. While panic, fear and worry will not help us in this situation, we definitely have to take a cold hard look at the facts of the matter to stand the best chance of coming through this pandemic with as little loss of life as possible.
The COVID-19 pandemic has left the frail underbellies of our healthcare systems and infrastructure brutally exposed. For all our combined might as a global civilisation, we have been shown to be woefully unprepared and under-equipped to deal with a cataclysm of such magnitude.
In the light of this global nightmare, it is painfully clear that our toolkit against a rampantly spreading global infection is woefully out of date. This pandemic has stunned healthcare systems across countries, thoroughly pulverising some of them in the process.
This crisis has brought to light just how grossly we have misallocated resources, both material and intellectual, as a global civilisation. In spite of the numerous technological tools we have at our disposal, we have been left reeling from the impact of this outbreak. Authorities around the world have been left scrambling for crumbs in an effort to contain the spread of the virus.
Those countries that have been hit the hardest, like Italy, stand as a grave example for the rest of the world. Countries that haven’t yet experienced the full fury of the outbreak would do well to learn from the calamitous outcome in countries like Italy.
As of today, the US has reported their highest daily death count due to COVID-19. This is to say that this situation is, by no means, finished. For many countries, this is only the beginning. We should all stand up and take lessons from Italy’s travails.
One of the main reasons for the calamitous outcome in Italy was that the authorities were always one step behind the virus. The virus mercilessly took advantage of the fact that authorities could never really predict where it was going next. By the time they could, it was always too late.
Containment strategies can be hopelessly ineffective once the trajectory of the epidemic is allowed to gain momentum. Today, despite having some of the strictest distancing measures in place, Italy has suffered damage that will take years to repair. The response was too haphazard when it most mattered. This is not to point fingers at any particular entity but to sound an alarm bell for the rest of us.
Our mechanisms to track the spread of the disease have just not been up to the mark. The prevalent epidemiological strategies are too labour-intensive, expensive and diffuse to help us mount a reasonable response to such a virulent infection.
“Now, we are running after it”, said a resigned Sandra Zampa, the under secretary at Italy’s health ministry, lamenting the lack of an effective early warning system. She believes that the Italian authorities did the best they could with the information they had.
This reveals a rather shocking truth. A paucity of real-time data is potentially a major culprit in this debacle. This is shocking primarily because of how powerful we have become, globally, at harnessing and processing data.
We might have been caught napping, but this pandemic has definitely shaken us out of slumber. Inspired minds all over the world have sprung forth with ideas, contributions and solutions. Numerous enterprising individuals and organisations around the world have begun to apply their collective intelligence to this global threat.
So, what is it that is missing from our current model? The simplistic answer is time, i.e responsiveness. In many countries, the present models just couldn’t provide enough data at a time when efforts at containment could have actually been effective. We needed an early warning system that would allow us to make reasonably reliable predictions about the course of the outbreak, which would then enable us to respond swiftly and incisively.
Elevated temperature is one of the most cardinal signs of infection. Any infection. Medically, it is one of the surest signs that your body is fighting a bacterial or viral invasion. The same holds true for COVID-19. Fever is likely to be the first sign of exposure to the coronavirus.
The Centers for Disease Control and Prevention (CDC) recommends taking your temperature, twice a day, if you think you have been exposed to COVID-19. As the situation stands, we should all consider ourselves potentially exposed.
For a few years now, we have employed community-based temperature tracking as a bulwark against influenza. During flu season, the dynamics of virus spread is very similar to what we are now seeing with the coronavirus pandemic. The scale of COVID-19 might be a different beast altogether, but the principle is the same.
In 2018, researchers at the University of Iowa published their findings which stated that using community-acquired data from smart thermometers that were linked to a mobile app was more effective than traditional means of flu forecasting. The researchers used data from Kinsa Health, which is a company that manufactures smart thermometers. The study used millions of temperature readings taken from over 450,000 individual devices to build a flu-prediction model. They found that the model based on the smart thermometers was much more effective than those employed by the CDC. In fact, data from CDC’s own weekly fluview tracker was more than 2 weeks behind the actual spread of the infection.
The real-time nature of this kind of data is what makes it so powerful, in being able to make reliable predictions. The existing mechanism of data collection by bodies like the CDC, believe it or not, relies on weekly reports from healthcare centres (doctor’s offices, hospitals etc) and health agencies. Moreover, all this data is manually collected. In addition to being hugely cumbersome and expensive on resources, this data is subject to human error. Here is one such example of the unreliability of manually collected data, where the British NHS was left embarrassed by an error that resulted in COVID-19 patients being given wrong test results.
This is not to point fingers or gleefully apportion blame onto the “system”. Authorities all over the world have been overwhelmed and are massively overworked. Countries all over the world are facing a devastating shortage of medical and administrative personnel. With every passing day, thousands upon thousands of such errors are bound to happen, when the system relies heavily on manpower.
Now, it should come as no surprise that an automated system for collecting data on a large scale will do better on this front. The success of Kinsa’s flu prediction strategy shows that.
Now, the same strategy is being trialled with the coronavirus outbreak. The company published a health weather map in late March to track the spread of the coronavirus. Within a day, the insights it offered proved to be invaluable - the data collected was able to reveal a reassuring trend. The fever data suggested that the social-distancing measures seemed to be working. The fever data showed that lockdown measures employed in places like the San Francisco bay area and Washington state were actually effective. The trends indicated a slowdown in the spread of COVID-19, in these places. On the flip side, and perhaps more relevantly, the system was able to forecast a devastating spike in the number of coronavirus cases in Florida.
“We frantically tried to alert authorities in the Tampa Bay area, but to no avail.” says Indebir Singh, CEO of Kinsa Inc. In early March, Florida was packed with spring breakers from all over the US and the fever data was able to detect an impending infection cluster in the area. Although efforts were made to get this information out, there wasn’t any decisive action that was taken.
”It was so frustrating,” said Ms. Nehru, who works for Kinsa. “For three days from about March 19 on, Inder was calling local government folks in Florida, The Tampa Bay Times and other papers. The government did absolutely nothing.”
This is just one example of a larger possibility. Today, we have made unimaginable advances when it comes to collecting and crunching astronomical amounts of data. Moreover, we are able to collect data at a grassroots level. This kind of a granular approach is what sets this apart from more traditional epidemiological methods.
So far, with the COVID-19 pandemic, we have been frantically lab-testing millions of people in a bid to track the spread of the disease. Needless to say, this is time-intensive and heavy on manpower. We simply do not have the capabilities to test everyone this way.
This is not to diminish the significance of lab testing or serological studies. As a long-term strategy, it is imperative that we build up our testing capabilities in an effort to make sure as many people as possible get tested. The United States, for instance, has launched a sero-survey campaign in a bid to study the spread of the SARS-CoV-2 virus. So, traditional methods of testing for antibodies, proteins etc is by no means obsolete or superfluous.
But there are practical constraints. Testing is expensive, time consuming and manpower heavy. As it is, healthcare workers have been stretched to inhuman limits by this pandemic. We just do not have enough resources to test as many people as we should. And because we are not able to test enough people, we have been perennially caught on the back foot by this novel virus.
Moreover, there is a bigger problem. We can’t use this method to inform our decisions. The lag involved is just not feasible. We need something that is more real-time. The lag between collecting data on an outbreak and using it to make swift and effective decisions can be slashed using modern connectivity technologies.
At ONiO, we are committed to making healthcare more accessible. We believe that the future of healthcare rests on how effectively we can integrate state-of-the-art data collection and connectivity technologies into our healthcare management systems.
Behind the scenes, we have been working relentlessly on our continuous temperature monitoring ecosystem which we believe could be a powerful tool against events like this coronavirus pandemic.
Think about it. Even without the pandemic, during the best of times, traditional temperature taking technology is clunky and unsophisticated. If you have ever had to record your own body temperature multiple times a day, you know how annoying it can be. And when you are faced with a situation where you need temperature data from a large population, using traditional thermometers to manually take measurements, is just not feasible.
Here’s where we think a “smarter” system to measure body temperature could be invaluable. ONiO.temp is a smart patch that you just stick onto your skin. It uses a self-powered sensor to continuously collect temperature readings from your body and transmit time-series data to the backend system. This data is anonymously and securely stored in the backend database. The user is able to access this data through their smartphone in order to gain insights into their health. But, this technology can be even more potent when employed on a wider scale.
Continuous temperature monitoring could be invaluable when used on a city-wide or nation-wide scale. Temperature trends of hundreds of thousands of people over a city or province could be studied in real-time. This would prove to be invaluable during an epidemic. As such, the example above showed just how powerful point-readings can be when there are millions of them taken over a large area. A continuous stream of temperature data that is seamlessly relayed, on a moment to moment basis, would offer exponentially more benefits to authorities and policymakers. A steady stream of accurate data can be absolutely game-changing for conducting big-data analysis of outbreaks like COVID-19.
Make no mistake, fever is by no means exclusive to coronavirus or any particular infection. As mentioned previously, analysing temperature data can’t replace lab testing as the most reliable means of studying the spread of an infection. However, it does make a huge difference if we have a robust and accurate early detection system. A warning mechanism if you will. And when this early warning system uses a continuous stream of real-time data, it is bound to offer significantly more value in terms of predictive modelling. To put it simply, the larger and more malleable the dataset, the better the outcome.
A continuous and accurate stream of fever-data could be cross-referenced with the huge amounts of epidemiological data we already have from many decades of performing health surveys on infectious diseases. We have the computing power for that. All that’s missing from the picture today is a large enough set of reliable data.
If we notice for example that there is an unusual spike in the number of fevers in a certain location, and we find out that this particular area is not a known hotbed for flu or any previously studied contagion, we will be able to direct action and resources in a timely manner. This is another massive advantage that continuous fever monitoring has. With time-series data we can splice the data in more sophisticated ways - we can analyse the data along highly specific lines; for instance, your baseline body temperature can be slightly higher than the average. We will be able to account for that when we see temperature trends. Having access to this kind of a rich data stream allows us to study the spread of a disease along extremely fine lines.
From an individual’s point of view, this allows you to gain highly personalised insights into your own health and fever response. ONiO.temp’s continuous temperature data allows us to create a “human model”, which can be used to understand the unique subtleties of each individual’s body. In a situation where access to professional medical help is scarce, this can be extremely useful.
Diseases are associated with their own unique “fever curves”, i.e patterns of temperature fluctuations that are unique to the disease. ONiO.temp could be used to chart detailed temperature trends that can allow us to separate the what from the chaff, so to speak. We will be able to hone-in on the relevant data and weed out confounding variables.
Convenience is another major advantage. We can balk all we want at people flouting measures and skimping out on taking adequate precaution. But over any reasonable sample size, this is absolutely par for the course. We need to expect this and plan for it. As humans, we are hardwired to seek convenience and avoid discomfort. A system that automatically tracks your body temperature will, 10 times out of 10, be more reliable than one that requires hundreds of thousands of people to conscientiously record their temperatures. ONiO makes things super-convenient. You don’t even necessarily have to fire the app up. Your temperature data is securely uploaded to the backend database. Only you and your healthcare provider will be able to access your fever data along with your personal information.
ONiO.temp’s data will be multivariate and the huge volume of data collected will mean that reliable predictions can be made even with relatively small populations. Moreover, in addition to body temperature, ONiO offers sensors that measure environmental parameters like room temperature, humidity, air pressure and light conditions. This confers additional robustness to any predictive modelling that is performed using the data.
Let’s look at this coronavirus outbreak. With COVID-19, high temperature and dry cough are the most telltale signs, even if the infection is mild. Even in mild cases of COVID-19, the fever is usually higher than what is seen in the case of a cold or flu. When we have access to continuous and accurate temperature readings, we can easily weed out those cases that are not pertinent to the outbreak that we are studying. And when this dataset is collected from a huge number of people, we will be able to sparse out the data that is pertinent to us, with a high degree of accuracy.
This is not just a fanciful idea. Today, we are able to produce tiny, affordable sensors that are powerful and can fit in the unlikeliest of places. Moreover, connectivity is more widespread than ever before. Thanks to technologies like energy harvesting, we can now sustainably employ millions, or even billions of sensors, in a wide range of use-cases.
ONiO.temp itself is an unobtrusive patch that you just stick onto your skin. It is affordable and more importantly, it doesn’t use any batteries. It harvests ambient RF (radio-frequency) energy to power itself. This means, the user can just stick the patch onto their skin and forget about it.
As a species, the last few weeks have given us a lot of reason to be apprehensive. However, our best hope lies in briskly taking inventory of our technological and logistical capabilities and responding with ingenuity, innovation and unorthodox thinking.