Can a smart watch test for the new crown?The Lancet: Not Reliable

On April 27th, a study published in the world’s top medical journal “The Lancet” shows that wearable devices are used in the field of monitoring new coronary pneumonia infections.It’s still in the early stages, and there has not been much progress.

In the early days of the COVID-19 outbreak, smartwatch makers and wearable technology companies wanted to use indicators such as heart rate and blood oxygen to mark such groups in advance. Even without significant progress, existing research suggests that wearables are still an effective way to track COVID-19 symptoms.

Researchers at Utrecht University in the Netherlands searched for research articles and protocols published between July 27, 2020 and 2021 on the use of wearable devices to identify COVID-19 infections. After identifying and screening 3,196 records, the researchers found Twelve articles and research protocols were analyzed.

These schemes all try to find methods from the user health data collected from Apple Apple Watch, Google’s fitbit and smart watches, wristbands, smart rings and other devices of the American smart sports bracelet Whoop. Among them, the related research on smart watches is relatively many.

The authors of the paper note that most of these studies have focused on people who have tested positive for Covid-19,Neither is a rigorous clinical trialthe existing research also does not have sufficient evidence that wearable devices can detect new coronary pneumonia infection earlier.

The paper, which will be open access May 1 in The Lancet, is titled The performance of wearable sensors in the detection of SARS-CoV-2 infection: a systematic review. Performance in Pneumonia Infection: A Systematic Review).

Paper link:Click here to view

1. There are limitations in research papers to diagnose population data

The research teams in each of the 12 studies used different methods to judge the ability of wearables to detect Covid-19 infection.

Nine of the studies usedMachine Learning Algorithmsto identify biological data to detect COVID-19 infection, including anomaly detection autoencoders, gradient boosting classifiers, and deep convolutional or gated recurrent unit neural networks.The remaining 3 studies usedStatistical Analysissuch as mixed-effects models and the Wilcoxon rank-sum test.

In addition, nine studies built models that directly compared wearable data from COVID-19 positive patients to healthy patients or COVID-19 negative controls. Eight studies considered changes in baseline parameters from presymptomatic to symptomatic infection in participants.

▲ Partial research on the use of wearable devices to detect changes in physiological parameters of COVID-19 positive individuals

Most of the current studies use retrospective data, according to the review article.small sample sizeand most authors focus only on symptoms of COVID-19,Physiological differences between COVID-19 symptoms and other illnesses such as the flu are ignored.

While there is ample evidence that physiological signals such as body temperature changes, heart rate variability and other indicators point to a new crown infection, fitbit research director Conor Heneghan told foreign media The Verge last year that the fitbit study found a correlation between flu data and new coronary pneumonia data. There is overlap, “my intuition is that it’s hard to distinguish them clearly.”

The study found that most algorithms for predicting COVID-19 from wearable data focused on symptomatic disease. Research shows that the ability of algorithms to detect pre-diagnosis infection varies widely from 14 days before symptoms to 1 day before diagnosis.will show a trend from 20% to 88%. “Accumulating evidence suggests that there is a trade-off between the model’s accuracy and its ability to identify COVID-19 infection before symptoms appear,” the paper’s authors said in the article.

The way wearables are used to identify Covid-19 infections is to flag those with symptoms earlier so they can be tested and quarantined before spreading the disease to others,The final diagnosis standard is still nucleic acid monitoring.

Another limitation of these papers is that the detection standards for patients infected with new coronary pneumonia are different. Only three papers will use PCR testing to confirm whether patients are infected with new coronary pneumonia during the research process. There are still some studies that did not use PCR testing to confirm. May lead to diagnostic bias and limit comparability between studies.

In fact, 2 of the 3 articles that introduced PCR testing were solely for healthcare professionals, a population that may have more access to PCR testing because of the demands of work during the COVID-19 outbreak.

The paper also mentions that the use of wearables as detectors for Covid-19 or other diseases would alsoThere are fairness issues.The samples analyzed in these studies haveworseEthnic Diversityso it is unclear whether these models perform equally well in non-white populations.

In addition, there are studies showing that wearables work differently and less accurately on darker skin tones, and the models evaluated in the review did not take into account women’s menstrual cycles, as body temperature is associated with different phases of the menstrual cycle. Variables also change.

2. It can recognize subtle changes of 0.2°C, and the temperature of the wrist is as stable as the forehead

Evidence on the use of wearables to detect Covid-19 suggests thatThe research is promising, but still in its early stages.

Despite these limitations, these 12 studies provide valuable insights into future research into the measurement of physiological parameters by wearable devices. For example, eight studies have shown increased heart rate to be associated with COVID-19 infection, which is consistent with flu-related population heart rate data. Likewise, changes in body temperature and heart rate present opportunities for new wearable devices equipped with temperature sensors and accelerometers.

In addition, to establish an accurate prediction model of COVID-19 in terms of respiratory rate and heart rate variability, more research is still needed to prove the existing conflicting or inconclusive conclusions. Others are cough patterns from mechano-acoustic sensors that can be used to decipher further trends associated with Covid-19 infection.

With the advancement of technology, wearable devices have more and more functions in the field of health monitoring, and subtle fluctuations in physiological parameters such as body temperature, respiratory rate, heart rate and blood oxygen saturation can be manifested by smart watches, smart rings and fitness trackers .

Notably, studies have shown that peripheral temperatures measured by wearable devices areDetects subtle temperature changes over 0.2°CAspects exhibited higher sensitivity than oral measurements, and wrist temperatures were equally stable and less susceptible to environmental influences than forehead temperatures.

As a result, more and more researchers are beginning to conduct more research on the role of wearables in the early and comprehensive detection of Covid-19 infection, unleashing the potential capabilities of wearables in monitoring personal health and predicting Covid-19.

Several studies have shown the feasibility of wearable devices in terms of monitoring one or more physiological parameters to indicate the presence of a risk of Covid-19 infection, but no more studies have yet been outlined.

Despite the limitations of existing research, wearables have the potential to be a great way to track and monitor disease. Just need to do better research to prove it and figure out the best way to use the device in these situations. Experts do believe that even a basic tool that can alert someone that they may be sick can still be useful.

3. Asymptomatic patients account for half of the transmission rate and can still spread during the incubation period

A key strategy to contain the spread of the Covid-19 outbreak is the rapid identification and tracing of infected people, according to the research paper. At present, the more authoritative diagnostic test standard for new coronary pneumonia is nucleic acid detection (RT-PCR).

Although the diagnostic time of nucleic acid testing is shortening step by step, the testing time still provides resistance to controlling the spread of the virus compared to the rapid spread of COVID-19 infection. On average, it takes six days from Covid-19 infection to onset of symptoms, while the incubation period can be as long as 18 days, the research paper said.

During the incubation period, the viral load in the patient’s upper respiratory tract increases, peaking around the onset of symptoms, and then gradually decreasing.

Many countries recommend testing the general population after symptoms appear, or days after suspected exposure, to prevent continued transmission of the virus before testing positive for COVID-19. Before people develop symptoms, however, the viral load in a person’s body is high enough to infect others.

It is still difficult to distinguish COVID-19 from other respiratory diseases based on reported symptoms alone. Many common COVID-19 symptoms such as fever and cough are similar to other flu illnesses.

Some confirmed Covid-19 patients develop unique symptoms, such as loss of smell, that are only caused by Covid-19, but such symptoms rarely appear early in the virus infection.

In addition, 20%-30% of people infected with new coronary pneumonia will remain asymptomatic. The U.S. Centers for Disease Control and Prevention reports that asymptomatic or asymptomatic confirmed patients account for half of the spread of the new coronavirus.

Therefore, in order to reduce the rate of transmission in the general population, it is critical to identify COVID-19 infection before or without symptoms.

Conclusion: Nucleic acid detection cannot be compared by analogy, but there is room for improvement in usability

Smartwatches, wristbands, rings and other wearable devices have more and more functions and categories, and their natural adaptability and convenience to the human body make more and more technology giants enter the health track. In addition, the rapid and uncontrollable spread of the new crown epidemic has also added resistance to its spread. Therefore, the development potential of wearable devices in the fields of health monitoring and new coronary pneumonia prediction has been continuously released.

But from these research analysis, we can also see that compared with nucleic acid detection,Wearables are less useful for COVID-19 detection. The paper also concludes that future research in this area should also consider how inherent differences in wearable sensor approaches, raw data processing, and algorithm development can help detect infection-related biases in physiological measurements, and how to address sources of bias.

Wearable devices have certain feasibility in predicting the symptoms of new coronary pneumonia. With the continuous deepening of research, its accuracy and practicality may be further enhanced.