How AI is being used to fight COVID-19Marie Donlon | July 20, 2020
Artificial intelligence (AI) is a growing presence in virtually every industry imaginable — writing songs, mimicking works of art, predicting a person’s life span and even making Oscar winner predictions.
Yet no where is its impact felt greater, arguably, than in the healthcare industry. More specifically, AI has figured heavily in the battle against COVID-19 where in the early days of the pandemic, little was understood about the virus, catching many in the healthcare industry by surprise and thus ill prepared. Luckily, AI has been applied to at least attempt to help battle the respiratory virus, be it in the shape of AI-based robots or through the design of AI models that predict those at greatest risk of being affected by the virus or the severity of the virus on specific individuals.
To reduce the spread of COVID-19 and other infectious diseases, researchers from the Korea Institute of Machinery and Materials (KIMM) built a robot capable of collecting specimens from patients' upper airways by inserting a swab into the patient’s nose or mouth.
Additionally, there are a host of other AI-driven robots designed to help battle the virus, including temperature-taking bots, suturing bots and COVID-19 information bots, all of which limit direct contact between patients and healthcare workers and thus the spread of COVID-19 and other infectious diseases.
The AI-driven disinfecting drone
A team of scientists from the National University of Ireland Galway’s Health Innovation via Engineering (HIVE) lab created a drone that emits ultraviolet (UV) light to sterilize surfaces in high traffic places.
Called the UVC Drone, the device is designed to sterilize public spaces such as restaurants, hospital wards, shopping centers and airports, from above with rays from the UVC (100 nm to 280 nm) band of UV light, emitting high frequency, short wavelength bursts of radiation imperceptible to the human eye.
Guided by AI algorithms, the drones can emit the UV light over predetermined locations and at predetermined times.
Researchers from the Chinese tech company Alibaba have created an algorithm capable of diagnosing the coronavirus in seconds based on a patient's computerized tomography (CT) scans.
Researchers from Alibaba’s Damo Academy trained an AI algorithm on a dataset of some 5,000 confirmed cases of the coronavirus. The algorithm reportedly recognizes signs of the virus in patient CT scans with a roughly 96% rate of accuracy.
The algorithm is also capable of distinguishing between typical viral pneumonia and the coronavirus, both of which present with respiratory difficulty. Additionally, the algorithm expedites the diagnostic process, identifying the virus in just 20 seconds — a process that typically takes doctors and other trained medical personnel 20 minutes to identify and diagnose following a review of patient CT scans. According to the algorithm's developers, expediting diagnostics of the virus will likely ensure immediate treatment of infected patients.
A team from King’s College London, Massachusetts General Hospital and tech company ZOE developed an AI system that determines if someone is likely to have contracted the coronavirus based on their symptoms.
The AI model was built using data from a coronavirus symptom study app that determines COVID-19 infection. According to its developers, the app compares coronavirus symptoms with test results, helping users with limited access to testing determine if they are demonstrating actual symptoms of COVID-19.
The team analyzed data gathered via the app from millions of users logging in their possible COVID-19 symptoms. Through that data, the team discovered that symptoms such as loss of taste or smell were stronger predictors of COVID-19 than fever, for instance.
As such, researchers took that data and built a math model that determines, with roughly 80% accuracy, if a user has COVID-19 based on factors such as age, sex and a combination of four key symptoms that include loss of smell or taste, severe or persistent cough, fatigue and loss of appetite.
By combining AI predictions with the app, those with the virus could be identified sooner, thereby improving both tracking and treatment efforts.
A team of researchers from North Carolina State University have developed a so-called artificial chemist, which uses AI to perform chemical reactions, thereby expediting the creation of materials, some of which may prove to successfully combat COVID-19.
Researchers from the Technical University of Denmark (DTU) are turning a wireless monitoring device into a tracking tool for coronavirus patients.
The Wireless Assessment of Respiratory and Circulatory Distress (WARD) monitoring system enables healthcare professionals to monitor coronavirus patients remotely amid a shortage of protective medical equipment and hospital beds in many parts of the world, thereby preventing the spread of the virus.
The monitoring system includes sensors worn by patients that continuously monitor patient blood pressure, oxygen saturation, respiration and heart rate. That data is then wirelessly communicated to a central computer where AI algorithms analyze and interpret the data, notifying healthcare professionals if there is a decline in the patient’s condition.
Going forward, expect to see AI to play an even greater role in the healthcare industry as it grapples with the worldwide pandemic. In the meantime, check back with Engineering360 for COVID-19 related technological developments and coverage.