Early Detection for COVID-19 possible?
Can we catch cases before symptoms occur?
Authors: Aki Ranin MSc., Jussi Leinonen M.D. Ph.D. based on the latest, if limited, yet rapidly expanding knowledge available on COVID-19.
Your first question is probably my first question, too. Don’t we already have tests for that? Yes, but below we examine reasons why that’s not enough with the pandemic-level spread of COVID-19.
There aren’t enough tests.
Tests take time to develop, time to manufacture, and yes, they cost real money. CDC quotes $36 for the U.S. tests being distributed through Medicare clinics, while private tests available for purchase are estimated between $50 to $100. Some sources quote numbers around the world as high as $100+ for every single test, before even thinking about the costs of treatment. That’s a lot if you plan to test a billion people or more. But even if we ignore economics, we just don’t have enough. The U.S. has administered tens of thousands of tests by now when the plan was to have done millions. The rest of the world isn’t much better off. Some countries, like Finland, are already giving up on testing and adopting blanket shutdowns to slow down the virus.
Tests will be used on symptomatic cases, which is too late.
Even if we had more tests, they would almost exclusively be used on symptomatic cases. COVID-19, or rather the underlying virus behind the disease, namely SARS‐CoV-2, has a high viral count, spreads in the air, and can be asymptomatic for up to two weeks with a median value of 5 days before symptoms show up. This triple threat of factors is driving the high rates of infection. While it’s certainly good to confirm suspected symptomatic cases, it means there can be an order of magnitude of cases more out there that are asymptomatic and totally unaware that they’re shedding virus constantly and potentially infecting others.
NOTE: There are two types of testing being carried out. Immunoassays, or rapid test kits, are not a solution for early detection of COVID-19, because the test is only reliable after prolonged disease. Diagnosis with the help of rapid test kits is not feasible if symptoms have been present for less than a week. That’s pretty late in the game! In addition, the golden standard Polymerase Chain Reaction (“PCR”) is not scalable to the non-symptomatic population due to cost, delay in results, and logistics of the necessary hardware.
What should we do, then?
While our focus is on diagnostics, let’s cover all the bases for some context before we get there.
A vaccine would be great
The media is full of articles on potential drugs, treatments, and vaccines, so we won’t cover that here. The point is, that as of right now, there is no drug, treatment, and certainly no vaccine. Billions if not trillions are riding on that discovery, so it’s just a matter of time. Time, which we don’t have. It might save us from resurgence, mutations, or the next pandemic, but not this one.
Isolation, quarantine, and lockdowns
In the absence of adequate testing, governments around the world are enforcing increasingly strict isolation policies, ranging from self-enforced guidance in Singapore to full city-wide lockdowns we’re now seeing in parts of Europe and America. Without saying, confirmed cases are being quarantined either at hospitals or at home.
This scenario gets us off the red line. That is a huge decision with long-standing economic consequences. We’ve never seen anything like it. The other thing we can do is remote monitoring. Several countries such as China, Iran, Israel, and now Singapore have launched government-backed mobile apps for location-tagging of confirmed cases. These apps use either GPS, IP tracing, or Bluetooth to alert users of possible contact with confirmed cases. Hong Kong has gone as far as issuing bracelets you might typically see used for probation violations being repurposed for enforcing home quarantine for returning travelers.
Nevertheless, isolation isn’t a cure. There is a further risk of resurgence and even mutation. We need better diagnostic metrics for right now, and for whatever lies ahead.
Non-invasive diagnostics
A whole category of potential tools hasn’t been much discussed in the media. The idea is that in the absence of blood and swab kits issued to every citizen, we would look to metrics with tools that are already in every home.
While the above isolation methods will flatten the curve, as already demonstrated in places like China, Hong Kong, and Singapore, it will not eliminate the virus. We may also see further waves after the initial pandemic is solved, especially as isolation measures are lifted. Worse still, we may see mutations in the coming months and years that will put is back in square one.
Feasible candidate metrics
The ideal feasible metric is something that can be done by people with the equipment they already possess, can be done at home, and doesn’t require visiting a clinic or hospital. Let’s look at what research can tell us.
NOTE: Currently, the Healthzilla app already uses a combination of RHR and HRV trend analysis in our Stress Score algorithm, which can pick up acute and chronic changes in these markers. This is the baseline which we aim to improve upon.
Resting Heart Rate (“RHR”)
Everyone roughly knows their heart rate. You can measure it with fingers on your neck and a timer. You can also measure it using your smartphone. How? By placing your fingertip on the smartphone camera, with the flash turned on, smart algorithms can pick up the volume changes of blood in your capillary veins — that’s your pulse! The technique is called plethysmography and it’s what most wearable devices use, that’s the green light under your Fitbit and Apple Watch.
Most people would also know from subjective experience, that when you’re sick, your heart beats faster. One study of 200,000 Fitbit users found clear correlations with increased RHR and seasonal influenza (Radin 2020). In the presence of high fever, significant increases in HR and RHR are common (Karjalainen 1986). While this seems promising as a way to detect COVID-19 given fever is a common symptom, RHR seems less promising as an early warning indicator before symptoms occur.
Heart Rate Variability (“HRV”)
Your heartbeat is actually partially driven by your nervous system. This introduces minute differences in the delay between individual beats. This is called HRV. More HRV is an indication of strong control of your heartbeat by your nervous system and is a good sign. While HRV is much harder to find accurately, it can be done from the same fingertip signal as RHR.
Athletes have been using HRV for years now, to measure recovery from workouts and plan their routines. However, research has shown a clear inverse correlation with inflammatory markers and HRV, suggesting an important interaction between inflammatory pathways and the autonomic nervous system (Aeschbacher 2016). While it is hard to isolate pre and post-infection data in clinical studies of any kind, there are studies on influenza vaccinations. One such study indicates that even the mild immune response to the vaccination strain of the influenza virus can be picked up in HRV data (Perring 2012).
All-in-all, the supporting research makes HRV an ideal candidate for an early warning metric. The key question is how far ahead of symptoms appearing can we detect. If just a few hours, it makes little impact on actions taken by individuals. If we can aim for days, it can make a massive difference given the median incubation time of 5 days for SARS-CoV-2. One study on sepsis, an acute and often deadly inflammatory condition, found that significant changes in HRV could be detected up to 35 hours before a clinical diagnosis was confirmed (Ahmad 2009). This gives us some hope, especially for the prediction of severe cases of COVID-19.
Heart Rate Entropy (“HRE”)
A newer approach to analyzing heartbeat data is entropy. Researchers have found that the interactions between your heart and autonomic nervous system include non-linear components. This means there are signals that aren’t being included in RHR and HRV data, that could be used to diagnose various diseases of the heart and nervous system. Most of this recent research has focused on the entropy of HRV specifically. This area is more speculative, but also offers new opportunities for discovery.
Respiratory Rate
While you may have never thought about it much, but we all have a natural breathing rhythm. How many breaths (in and out together) you take per minute on average over a given period is your Respiratory Rate. Interestingly, and somewhat surprisingly, respiratory rate can also be measured through the same method as RHR and HRV above, from your fingertip. This is possible because breathing directly impacts your heartbeat.
The normal range is between 12 and 20 breaths per minute, so anything below or above would be considered abnormal. Those abnormalities would typically come from asthma, anxiety, pneumonia, congestive heart failure, lung disease, use of narcotics or drug overdose (Goldman 2007). Notice that pneumonia is on the list. This means that while Respiratory Rate is unlikely to predict the onset of COVID-19 symptoms, it can be an important diagnostic for those already showing symptoms. If we see an elevation or trend in respiratory rate, that could be used to trigger alerts to seek treatment urgently. A respiratory rate of above 20 per minute is associated with diagnosis of pneumonia (Gennis 1989).
Couldn’t we just tag symptoms?
Hold on, did we miss something obvious here? Rather than try all this high-tech stuff, couldn’t we just ASK people about symptoms? Yes, we absolutely should. In fact, there are many such initiatives and apps out there already.
The symptoms are quite well known at this point. Slap that on an app with some geotagging, and you have a world-wide map of the virus. So case closed? Well, not quite. Firstly, the data is highly subjective. Do you feel a little feverish, or did you actually measure? What constitutes a dry cough, how many times a day until it’s a real symptom and not just a dry throat. Subjective data is hard to analyze because of the underlying problem of… well, subjectivity. It’s a useful leading indicator of new infections if done at scale, but might also be wildly inaccurate.
However, there’s another reason it is critical data. If we measure RHR, HRV, HRE, and Respiratory Rate together, and then tag symptoms, we have a solution. This allows us to find cases where we have biological data to track your body’s immune response before and after you reported symptoms. This allows us to narrow down the predictive power of these biological indicators, and maybe even develop a predictive model! This is the holy grail we’re after.
Nailing down the time-course of COVID-19
Before we get too excited, we have to make it clear this is just a hypothesis at this point. Nobody has done proper clinical trials on COVID-19 of any kind. Those take up to a year, and it’s just been a few months. In the absence of playing it by the book, the next best thing is getting real data on COVID-19.
Most importantly, to make real use of the above biological metrics, we need to get data specific to COVID-19. It helps to know how they behave with the flu and pneumonia, but to be valid we must get data specific to the virus at hand. If we can, what can we find?
Primary and secondary viremia
The progression of any infection starts with the initial contact with the virus. From there, the virus enters the bloodstream. This is called primary viremia. The virus will travel to different organs to replicate. In the case of SARS‐CoV-2 (the virus causing the COVID-19 disease), this first base for the virus seems to be the throat. That explains sore throat as the typical first symptom reported by confirmed cases. From there, the virus replicates quickly, until it returns to the bloodstream for an all-out attack. This is secondary viremia, and how the virus ends up in the heart, lungs, and gut to cause damage.
It is very unlikely we will be able to spot the moment of infection or even primary viremia. If anything, we would see signals of a spike in immune activity from your immune system fighting the virus. It has been analyzed in related SARS and MERS-COV viruses that numerous inflammatory markers elevate in the blood, which could cause rapid changes in HRV (Huang 2020). One study from Wuhan shows similar patterns for SARS‐CoV-2 with severe cases showing double median values for inflammatory markers C-Reactive Protein and Interleukin-6 compared to less severe cases (Qin 2020).
Versus symptoms onset
One of the many dangers of SARS‐CoV-2 is the apparent high viral load before symptoms start. This means the virus count in your body is high and therefore easily passed forward, even though you don’t know you’re infected. One study shows that the peak virus count is actually at the end of the incubation period, i.e. just before symptoms start. Once the viral count rises, there is a level at which the count is high enough to transmit and infect other people, but not high enough to trigger symptoms. The study estimates this threshold at 1–2 days before symptoms occur (Anderson 2020). This is the target zone we are attempting to predict.
How much before symptoms start is the key question, and whether or not this will show up in the biological data. If we can use flu studies as a basis for our assumptions, then the hypothesis would point to a large drop in HRV up to 35 hours in advance of the first symptoms. While this is well short of the 5-day median incubation period, every hour before symptoms occur can save further infections in self-isolation!
Pneumonia
If we follow the data for people who develop symptoms, the most important thing now is to avoid getting pneumonia. As the virus does its damage in the organs incl. heart and lungs, there is an opening for bacterial infections to take hold. The most common infection has been severe pneumonia, which is driving up the mortality rate among the elderly.
There is a possibility that a combination of the metrics can help us detect the onset of pneumonia before severe breathing difficulties appear. In pneumonia research, one such predictor is the respiratory rate. Studies show that a respiratory rate above or below the normal range of 12–20 breaths per minute is an indicator of mortality risk from pneumonia. So we would want to keep an eye on that and warn both individuals as well as healthcare providers of cases outside that range for urgent treatment. This would greatly assist in prioritizing hospital beds and ventilators that are also in short supply, and lead to terrible fates in Italy.
THE ASK: Seeking partners to slow down COVID-19
We’ve already been working on this for several weeks by now. We have our existing app available for iOS. You can download it for free and use our existing early detection Stress Score algorithm. We are rolling out COVID-19 specific versions next week with consent for data sharing to contribute data to the research project. Android coming shortly after. But we also can’t do this alone. We need a coalition of partners.
Healthy and symptomatic users to volunteer on the app
To pinpoint the exact patterns of infection in different age groups, we need more data to analyze. By downloading the app and providing consent, we will be able to use your data as part of this research.
Share data with hospitals and researchers
First and foremost, we are not looking to take advantage of a dire situation financially. We are also not trying to gather and hold this data for ourselves. Quite the opposite. We hope to find hospitals and research groups who are willing to share data. We can provide biological marker data and tagged symptoms, but we also need confirmed cases. Please contact us.
Develop tools for hospitals and government
We are also willing to provide our technologies for individual hospitals and local governments to help track and manage their own local datasets. We will provide our technology without costs, and give you access to your local datasets for research and implementing better policies for control of the virus.
Partners and investors
Finally, we are also looking for other companies targeting to kill COVID-19. Whether we can distribute our technology through your platforms, integrate our technologies for better solutions, or get financial support to scale this quickly, please connect with us now so we can get this technology in the right hands!
CONTACT US
Aki Ranin is the founder of two startups, based in Singapore. Healthzilla is a health-tech company and creator of the Healthzilla health analytics app (iOS) (Android). Bambu is a Fintech company that provides digital wealth solutions for financial services companies.
References
- (Huang 2020). Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Sciencedirect.
- (Qin 2020). Dysregulation of immune response in patients with COVID-19 in Wuhan, China. Oxford Academic.
- (Radin 2020). Harnessing wearable device data to improve state-level real-time surveillance of influenza-like illness in the USA: a population-based study. Sciencedirect.
- (Karjalainen 1986). Fever and Cardiac Rhythm. Scopus.
- (Aeschbacher 2016). Heart rate, heart rate variability and inflammatory biomarkers among young and healthy adults. Annals of Medicine.
- (Perring 2012). Assessment of changes in cardiac autonomic tone resulting from inflammatory response to the influenza vaccination. Clinical Physiology and Functional Imaging. Pubmed.
- (Ahmad 2009). Continuous Multi-Parameter Heart Rate Variability Analysis Heralds Onset of Sepsis in Adults. PLOS One.
- (Anderson 2020). How will country-based mitigation measures influence the course of the COVID-19 epidemic?. The Lancet.
- (Goldman 2007). Approach to the patient with abnormal vital signs. Cecil Textbook of Medicine. 23rd edition. Chapter 7.
- (Strauss 2014). The Prognostic Significance of Respiratory Rate in Patients With Pneumonia. Pubmed.
- (Gennis 1989). Clinical Criteria for the Detection of Pneumonia in Adults: Guidelines for Ordering Chest Roentgenograms in the Emergency Department. Journal of Emergency Medicine. Pubmed.
