- June 23, 2022
- AI Projects
“Health” trackers are connected bracelets used for many reasons, first of all, to know your heart rate, your muscle and fat mass, your skin temperature or even your stress level, and the number of steps taken per day. In a study published in the British Medical Journal, researchers have revealed an extraordinary potential: the ability to gain comprehensive insights into one’s state of health, and potentially predict Covid-19 diagnosis even before any visible symptoms manifest. Researchers base their findings on wearers of the regulated, commercially available fertility tracker bracelet codenamed “Ava”, which monitors respiratory rate, heart rate, heart rate variability, wrist skin temperature, and blood flow, as well as the quantity and quality of sleep. As the scientific team explains, it has now been several months since “attention is focused on the potential of activity trackers and smartwatches to detect all stages of COVID-19 infection, from incubation to recovery, with the aim of facilitating isolation and testing of infected persons. »
Typical symptoms of COVID-19 may take several days after infection to appear, during which time an infected person may unwittingly spread the virus. The Ministry of Health says on this subject that “ the incubation period (the period between contamination and the appearance of the first symptoms, if they appear) of Covid-19 is 3 to 5 days in general, but can extend up to 14 days. During this period, the subject may be contagious: he may carry the virus before the onset of symptoms or on the appearance of weak signals. The researchers, therefore, wanted to see if physiological changes, monitored by an activity tracker, could be used to develop a machine learning algorithm to detect COVID-19 infection before symptoms start. During the comprehensive study, a total of 1,163 individuals under the age of 51 were meticulously tracked from the onset of the pandemic up until April 2021. They were asked to wear the Ava bracelet at night, with the device recording data every 10 seconds provided they sleep at least four hours. This tool was not chosen at random: its operation is based on a machine learning algorithm in order to detect the days of ovulation, reaching an accuracy of 90%.
A difference in several physiological parameters before the onset of symptoms
The wristbands were synchronized with a companion smartphone app upon waking, which participants used to record any activities that could impair central nervous system function, such as alcohol, prescription medications, and recreational drugs, and to record any symptoms of COVID-19. Each of them regularly took rapid antibody tests for SARS-CoV-2, the virus responsible for COVID-19 infection and those who showed indicative symptoms also took an RT-PCR test. Finally, additional information such as age, sex, smoking status, blood type, number of children, exposure to family contacts or co-workers who tested positive for COVID-19, and the status of the vaccine were taken into account. It turns out that 127 people developed the infection during the study period, partly because the majority of them said they had been in contact with family members, relatives or work colleagues. Who had also tested positive for a COVID-19 infection?
Throughout the study, an1.5 million hours of physiological data were meticulously recorded, providing a robust dataset for analysis. Among the participants, 127 individuals were confirmed to have contracted Covid-19. Interestingly, 66 of these participants (accounting for 52%) had consistently worn their devices for a minimum of 29 days and were thus included in the in-depth analysis. Surveillance data revealed significant changes in five physiological indicators during the incubation, pre-symptomatic, symptomatic, and recovery periods of COVID-19 compared to baseline measures, with symptoms lasting an average of 8.5 days. The scientific team thus estimates that the use of the health tracker and the algorithm correctly identified 68% of people positive for Covid-19 two days before the onset of symptoms and that the accuracy of the tool is close to 80%. How can the usefulness of this technology for this type of screening be explained? It turns out that a beginning COVID-19 infection can be characterized by an abnormal decrease in oxygenation in the blood, a phenomenon likely to lead to variations in heart and respiratory rhythms detectable by the connected bracelet. However, the researchers concede that this study has its limitations.
First, the fact that the results were based on a small sample of people who were relatively young, therefore less likely to have severe symptoms of COVID-19, and who were not ethnically diverse. This leads him to affirm that a PCR test must remain the reference to confirm the infection. But this tracker, mobile app, and algorithm combination are currently being tested in a much larger group of 20,000 people across the Netherlands, with results expected later this year. Based on this first observation, the researchers are hopeful that it is a very promising tool for the presymptomatic or even asymptomatic detection of Covid-19. ” This easy-to-use, low-cost method for individuals to track their health and well-being during a pandemic. According to our research, the fusion of these cutting-edge devices with artificial intelligence has the potential to revolutionize personalized medicine, surpassing current boundaries. By harnessing this powerful combination, we can identify diseases even before symptoms manifest, presenting a groundbreaking opportunity to curb virus transmission and enhance healthcare outcomes “, they conclude.
- AI-Driven Agriculture
- Non-Technical Introductory AI Courses for Enthusiasts
- Microsoft’s Xiaoice AI Project
- AI-Powered Drones Outperform Human Pilots in Obstacle Course
- AI Algorithms for Game AI
- AI in Financial Services
- Stability AI Introduces Cutting-Edge ‘Stable Audio’ Model for Precise Audio Generation
- Stanford University’s Online Artificial Intelligence Lab Courses
- How AI is Revolutionizing DNA Testing
- A Fusion of Domain-Specific Assistance and Robust Language Models in an Embodied Conversational Agent
Get regular updates on data science, artificial intelligence, machine