Important Notice

From the 5th of June 2024, elarm has been discontinued.
Full details and actions you can take to continue health monitoring are on our website here.

elarm

How elarm works

elarm works by detecting the early biometric changes using consumer-grade wearables during the inflammatory response phase or viral load. elarm then notifies the user.
Figure 1: How Elarm Works Figure 1: How Elarm Works
elarm does this by developing your personal biometric baseline and monitoring changes from your baseline over time through clinically developed artificial intelligence models to assess any anomalies based on similarity to data from observed COVID-19 viral load cases. Figure 1 shows the steps involved in the process from your wearable to the wearable apps on your phone. elarm collects your biometric data, develops derived variables and builds your biometric baseline . elarm then compares your real-time data to your baseline to detect changes that fit the viral load pattern seen in the COVID-19 positive cases 2-3 days before they were sick.

elarm in action - COVID-19 positive

The following example is from Matthieu, an elarm user in Canada, in his own words:
12/29/21: Little sweats at bedtime
12/30/21: Nostrils tender when breathing (negative RAT test)
12/31/21: Lots of clear nasal discharge
01/01/21: Runny nose, castling voice, slight tingling in throat
02/01/21: Loss of sense of smell, runny nose, castling voice, slight tingling in the throat (positive RAT coronavirus test)
Everything returned to normal a few days later:
See other examples below

elarm: the research & the science

Pre-Coronavirus, there was little research on wearables to detect viral load or potential illness. There were studies using fitness trackers to detect population changes for Influenza  , but few articles at an individual user level. Since the onset of Novel Coronavirus (SARS-CoV-2 or nCoV) in 2019, many universities and research institutes have studied wearables, fitness trackers and non-medical grade devices to detect viral load and COVID-19. The development of the elarm system started in March 2020 has been in parallel to the University research: Figure 2 below shows the effect in the body of viral load:
Figure 2: Timings of viral load and onset of symptoms Figure 2: Timings of viral load and onset of symptoms
See the Reference section for more details on elarm and wearables research.

PlusMinus AI Algorithm

The elarm system uses advanced AI (artificial intelligence) algorithms to determine your current deviation from your baseline biometrics. The PlusMinus algorithms use the data from COVID-19 positive people to develop the digital biomarkers of early biometric changes. The elarm PlusMinus algorithm identifies changes 2-3 days before someone knows they are sick. When wearable users tested positive for COVID-19, we used the period before knowing they were infected – when viral load began within the biometrics markers. When the Datamine  team first started to develop the elarm system in early 2020, we consulted widely to understand the best time to detect the viral load to enable users to take advantage of the knowledge. The clear answer was early detection - before someone knew they were sick and already infectious. Clinicians told the team that early intervention might help sick people get the help they need and isolate to reduce the risks of infecting others. Thus, the elarm system targets detecting the viral load 2-3 days before someone knows they are sick. The following graph (figure 3) shows the elarm target zone.
Figure 3: elarm targets 2-3 days before someone knows they are sick Figure 3: elarm targets 2-3 days before someone knows they are sick
The PlusMinus algorithms use a mix of raw and derived variables and use biomarkers to detect changes – see the following section for data used in elarm

What data does elarm use?

elarm builds up a picture of your wearable’s baseline or typical biometric behaviour. The system then compares your current biometrics vs that baseline. When someone has a viral load (e.g. COVID-19, Influenza), their biometrics change compared to their baseline. elarm uses the PlusMinus AI algorithms to take raw data from your wearable, e.g. heart rate, exercise, SpO2, HRV (Heart Rate Variability)  , respiratory rate, and sleep. The elarm system creates dozens of derived time series variables (e.g. exercise recovery, heart rate variability balance, heart rate variability when not available, sleep recovery). elarm builds each user’s baseline and models the likelihood of changes in the baseline fit to viral load patterns from people who have had Coronavirus. The derived variables or complex biomarkers are the key to the elarm PlusMinus algorithms. The combinations of these biomarkers enable multiple time series datasets to build up your baseline and detect changes to your current biometrics and biomarkers. The feature dimensions of the biomarkers enable the algorithms to combine different raw biometrics to enhance the sensitivity and accuracy of the algorithms. The following graph (figure 4) shows an example of the development of the derived elarm biomarkers. In this case, we look at sleep recovery: your normal sleep recovery may look like the blue line - your heart rate slowly increases until a quarter of the way through your sleep and then slowly decreases until 4am, then slightly increases. elarm builds up your baseline sleep recovery patterns.
Figure 4: An example of sleep recovery Figure 4: An example of sleep recovery
The orange line may represent last night’s sleep and shows a more prolonged and higher increase in heart rate and a continuing decrease until you wake. Note the higher final heart rate before you wake and that there is no gradual increase. Different people have different sleep patterns, and the biometric changes elarm measures are against your personal baseline. The models elarm uses can work with minimal exercise/movement and HR/ HRV (Heart Rate Variability)  data but are more accurate the more raw variables your device has, and the longer you wear your watch. See ‘How often do I need to wear my watch?’ elarm’s PlusMinus algorithms currently detect 83% of COVID-19 positive people in the system. See the accuracy section for more information.

How will I know if something has changed in my biometrics?

elarm uses the Wellness Risk Score to measure how far your current biometric measurements are from your normal biometric baseline. The Wellness Risk Score is on a scale from zero to 100. At zero, your biometric readings are close to your baseline; at 100, your biometrics are very different to your normal baseline. The higher you are on the score, the further your biometrics from your wearable is from your baseline. elarm has four levels of the Wellness Risk Score to indicate the level of deviation from baseline. They are Normal, Slightly elevated, Moderately elevated and Highly elevated. Gaps in your Wellness Risk Score indicate that the elarm system could not run the models for your biometric measurements for that period. Reasons for this include:
  • You were not wearing your device during that time
  • The system did not have enough biometric measurements for that period
  • The information has not synced from your wearable to your phone yet. You can open your wearable app (e.g. Apple Health, Fitbit or Google Fit) to check it has the information
  • Some devices (e.g. Oura Ring) do not take regular measurements, so elarm may not have enough biometric measurements for that period
  • The link between your wearable app and elarm is disconnected. You may want to try reconnecting (forget the current device and add the device again) elarm to your device (settings -> devices)

elarm Accuracy

elarm is not a test, nor is elarm is a medical device or providing medical services. elarm merely enables you to understand if your current biometrics differ from your norm. COVID-PCR test or rapid antigen tests (RAT ) are tests for SARS-CoV-2 (the virus that causes COVID-19) and have accuracy levels of 95%-99% ( Johns Hopkins Medicine  , Healthline  ) and 45-85% ( Healthline  , Cochrane Library  , National Library of Medicine  , British Medical Journal  ), respectively. Both tests have different uses in the detection of COVID-19. PCR tests are more accurate but slow to get results, costly and have system constraints (need to analyze the samples in a laboratory). Rapid Antigen Tests can be self-administered, quick to return a result, but less accurate in detecting COVID-19. elarm utilizes wearables to detect COVID-19. Verification of the elarm system using statistical cross-validation shows that elarm identifies 83% of COVID-19 positive cases. The Receiver Operating Characteristic Curve in Figure 5 below shows an example of the results from the elarm model outputs.
Figure 5: elarm Receiver Operating Characteristic Curve Figure 5: elarm Receiver Operating Characteristic Curve
elarm has a high false-positive rate and lets users know that their biometrics are different to their usual baseline - see elarm examples . As elarm is not a test, an elevated Wellness Risk Score is not a diagnosis of COVID-19. Elarm has detected different variants of Coronavirus, including Delta, Omicron and BA.2. The team continue to look at ways to improve the accuracy of identifying Coronavirus positive cases through adapting the PlusMinus AI algorithms and looking at new derived variables that indicate biometric changes. See how we would like to see the elarm system advance

How is the baseline calculated?

elarm builds your baseline from the raw variables and dozens of derived variables . Your baseline is simply the biomarker features that elarm observes when your wearable data exhibits normal behaviour or baseline characteristics. How long does it take elarm to develop a baseline? If you wear your device during the day and night, a simple baseline is possible after two to four days. elarm continues to adapt and improve your baseline over time.

elarm examples - does elarm only indicate COVID-19?

In short, elarm does not only indicate Coronavirus, and elarm is not a test for COVID-19. Research  and users have feedback on the following health issues:
  • COVID-19 (Novel Coronavirus) 1
  • Flu (Influenza) 2
  • Viral and bacterial infections 3
  • Stress and Anxiety 4,5
  • Excessive caffeine, nicotine or other stimulants 6
  • Lack of sleep, dehydration, low blood sugar 7
The elarm Wellness Risk Score is tuned to detect COVID-19. It also detects other changes (for example, users have reported stress, colds/flu, alcohol, spider bites and others). Elarm is focused on the early detection of COVID but is interested in the future tuning multiple models to different illnesses, change in body metrics (e.g. dehydration in dementia patients) or disabilities. See other uses the elarm team are working on The following are examples that elarm users have shared with the elarm team:

Influenza

The following elarm dashboard is from a user who said they had the flu symptoms and tested negative for COVID-19. Their doctors said it was most likely that they had Influenza but did not test for Influenza via a PCR test.

Sugar/Alcohol

The following user said they do not usually drink alcohol and rarely have sugar treats. The elarm graph below shows the impact of lifestyle choices on health – in the user’s words, ‘I had one too many drinks’. They reported feeling ‘under the weather for three days’ and said elarm had helped them reflect on their choices.
For regular drinkers consuming their usual amount, there is unlikely to be an increase in their Wellness Risk Score unless they increase their level of consumption.

Spider Bite

The following example is what a user experienced when a spider bit him. The user ended up having an antibiotic ointment to treat the bite.

Appendicitis

The following user ruptured their appendix and shared their elarm dashboard:

Stress

The following user was experiencing work pressure and uses elarm to monitor stress levels to reduce and take action:

Flatline

Some users have referred to the following as ‘boring mode’, but a flat line in elarm indicates that your current biometric markers are similar to your baseline. Flatline is a good thing – take a look at some of the other graphs in this section and be thankful :-)

Unvaccinated COVID-19

The following is the elarm graph for an unvaccinated COVID-19 PCR positive person. They had a PCR test two days after the onset of symptoms.

COVID-19 – Positive

The following example is from Matthieu, an elarm user in Canada, in his own words:
12/29/21: Little sweats at bedtime
12/30/21: Nostrils tender when breathing (negative RAT test)
12/31/21: Lots of clear nasal discharge
01/01/21: Runny nose, castling voice, slight tingling in throat
02/01/21: Loss of sense of smell, runny nose, castling voice, slight tingling in the throat (positive RAT coronavirus test)
Everything returned to normal a few days later:

COVID-19 – Triple vaccinated & Asymptomatic

The following graph shows a user that tested positive for COVID-19 via a Rapid Antigen Test. The user is triple-vaccinated (BioNTech, Pfizer double dose and one booster) and was asymptomatic (no COVID-19 symptoms or feeling unwell). elarm enabled the user to get tested and then take steps not to pass the virus onto others.
If you have some interesting uses of elarm to detect other ailments, then please get in touch via the chat within the help page in the elarm app.

How often do I need to wear my watch?

The more you wear your device, the better the elarm model accuracy. You need to charge your devices, but if you wear it whenever you are not charging, elarm makes the best baseline calculations and improves model accuracy. This includes through the night, while you sleep. Some users only wear their devices during the day. This means that some derived variables (e.g. sleep recovery) cannot be used.

How accurate are wearables?

elarm uses consumer-grade wearables as they are widely available (an estimated 1.1 billion connected wearable devices worldwide in 2022  ), relatively low cost and reliable. Modern wearables can accurately measure heart rate (bpm), with Apple Watch and Fitbit tests at 95% and 91% agreement with ECG, respectively link  . Oura ring produces near-perfect readings for resting heart rate (r² = 0.996) and is highly correlated with heart rate variability (r² = 0.980) when compared to a medical-grade ECG device link  . See Figure 6.
Figure 6: Fitbit Charge 2 (top) and Apple Watch 3 (bottom) Figure 6: Fitbit Charge 2 (top) and Apple Watch 3 (bottom) compared to the electrocardiogram (ECG) across 24-hours. bpm: beats per minute. Source 
More importantly, the consistency of each wearable’s reading is very high if you keep the same wearable. elarm builds a baseline with that wearable and compares your current biometrics to that baseline. When you change your wearable, elarm needs to build a new baseline for you - drop your elarm connection and reconnect with your new wearable. This is also the case when you move within the same brand (e.g. Fitbit to Fitbit) or change the model (e.g. replacing your Apple Watch 7) as it may take slightly different measurements.

How does elarm protect my information?

elarm only uses your data to deliver your results and improve your results. We care about your privacy. elarm was designed for user privacy and has the Privacy Trust Mark  . We also have ISO information security management (27001) certification. You can read all about it in our Privacy Policy and Terms & Conditions.
Privacy Trust Mark
elarm’s position is that your data is your data. It is not your doctor’s, the government’s, or wearables manufacturer and is not owned by the many health systems that store your health data. Your data is yours to own. Our ultimate vision is to have all your health and wellbeing information held in one place that you own. You choose who can access your information and how that information is used. All of your health-related information goes into your HealthVault.

How is elarm being used around the world?

elarm launched in June 2020, just three months after the World Health Organisation declared COVID-19 as a pandemic. Since then, the team has been working on increasing the ability of elarm to detect biometric changes matching COVID-19 viral load and improving the user experiance. The initial system was web-based; elarm iOS and Android Apps versions were released in September 2020. In January 2021, elarm’s offering widened to enable businesses to help protect their employees. Many companies worldwide use this service, including Newcrest Mining in Australia. In April 2021, the New Zealand Ministry of Health began using elarm for border and health workers. A survey of users found:
  • 84% enjoyed using elarm
  • 77% would recommend elarm to a friend or colleague
  • 94% found elarm easy to use
  • 67% said that elarm gave them greater confidence in health and the protection against COVID19
  • 68% had received a heightened alert, and 75% of those carefully observed their health over the following days or got a COVID test
  • 78% said that their employer should offer elarm as an employee benefit.
elarm has users in the following 117 counties:
AlbaniaAlgeriaArgentinaAustraliaAustriaBahrainBangladeshBarbadosBelarusBelgiumBrazilBruneiBulgariaBurundiCameroonCanadaCayman IslandsChadChileChinaColombiaCook IslandsCosta RicaCroatiaCuraçaoCyprusCzechiaDenmarkDjiboutiDR CongoEgyptFinlandFranceFrench PolynesiaGambiaGeorgiaGermanyGhanaGreeceGuadeloupeGuatemalaGuyanaHondurasHong KongHungaryIcelandIndiaIndonesiaIranIraqIrelandIsle of ManIsraelItalyIvory CoastJapanJerseyJordanKenyaKuwaitLatviaLebanonLuxembourgMacaoMalaysiaMaltaMauritiusMexicoMoldovaMonacoMoroccoNamibiaNepalNetherlandsNew CaledoniaNew ZealandNicaraguaNigeriaNorth MacedoniaNorwayPakistanPanamaPeruPhilippinesPolandPortugalPuerto RicoQatarRomaniaRussiaSenegalSerbiaSingaporeSlovakiaSloveniaSomaliaSouth AfricaSouth KoreaSpainSri LankaSurinameSwedenSwitzerlandSyriaTaiwanThailandTrinidad and TobagoTurkeyU.S. Virgin IslandsUkraineUnited Arab EmiratesUnited KingdomUnited StatesVenezuelaVietnamZambiaZimbabwe
elarm is also involved in clinical trials to enable wearables and the elarm system for viral load detection and other health and wellbeing applications. If you are a researcher and would like to collaborate with the elarm team, please contact us via the chat within the help page in the elarm app.

What happens to elarm after COVID-19?

The team behind elarm believe there are many uses for the elarm system in the future - see the example section for more details. The team see research and different models developed for other health and wellbeing changes to:
  • identify the early onset of illness, disease or changes in wellness
  • help understand the lifestyle or diet-related triggers to illness
  • enable people to monitor their health and wellbeing and make positive lifestyle choices
  • work with medical researchers to better understand health, illnesses and wellbeing
Specific examples are:
  • General health, wellbeing indicators and lifestyle triggers
    • Over-exercising
    • Sleep deprivation
    • Food and drinks that lead to biometrics changes
  • Technology-driven disease surveillance
  • Workplace Health & Safety – see elarm Guardian for workplace use of elarm
  • Assessment of prolonged physiological and behavioural changes associated with COVID-19 infection
  • Dehydration in dementia patients
  • Identify the early onset of Myalgic Encephalomyelitis/Chronic Fatigue Syndrome episodes
  • Monitoring of future epidemic outbreaks to get early identification and population indicators
  • Contagious diseases, e.g. Influenza, Hepatitis, Chicken Pox
  • Sepsis and reactions to infections
  • Periods
  • Menopause
  • Workplace mental health and wellbeing
  • Hyperthyroidism
  • Hypoglycaemia
  • Postural or orthostatic hypotension
  • Pregnancy
  • Stress, anxiety and panic attacks

References

Studies were undertaken by organizations such as Stanford Medicine (see “Smartwatch can detect early signs of illness”  ) show the effectiveness of wearable devices in detecting early signs of illness. This is possible due to a key characteristic of COVID-19 being high viral load (that is, the quantity of virus particles carried by an infected person) before symptoms start. Viral shedding (the release of the virus into the environment) is estimated to start 2.5 days before (and peak 14 hours before) a person starts showing outward symptoms ( link  , link  ) such as coughing and fever, and before they themselves are aware of the changes within their own body. Yet someone with COVID-19 is infectious to others during this time. During this phase of infection, heartrate, heartrate variability, skin temperature and other biometrics elevate as the body begins reacting to the virus. These biometric changes are strongly linked to viral inflammatory response biomarkers such as CRP (C-reactive protein) and leukocyte cell count link 

Other interesting articles

WVU Rockefeller Neuroscience Institute announces capability to predict COVID-19 related symptoms up to three days in advance: link  Wearable tech can spot coronavirus symptoms: link  Scripps institute’s Detect Study: link  Wearable tech can spot coronavirus symptoms before you even realize you’re sick: link  Stanford Healthcare Innovation Lab - COVID-19 Wearables Study: link  Stanford Medicine - Smartwatch can detect early signs of illness: link  Early data suggest wearables can catch some cases of Covid-19 before symptoms emerge: link