Can AI tackle the growing menace of coronavirus?
Introduction
At present, the only topic that is on everyone’s minds is the coronavirus. What is ‘coronavirus’ after all? Coronavirus belongs to a large family of viruses that cause diseases ranging from a common cold to more severe illnesses such as the Middle East Respiratory Syndrome (MERS) and Severe Acute Respiratory Syndrome (SARS). The first cases of COVID-19 emerged in December 2019 from Wuhan, a city in the Hubei province of China.
Coronavirus, first discovered in 2019, is also known as COVID-19. It has not been detected in human beings before. Most of the signs associated with COVID-19 usually affect the respiratory system, which can range from fever, cough, shortness of breath, pneumonia, and in some cases, even death. In March 2020, the COVID-19 disease was declared a ‘pandemic’ by the World Health Organization (WHO).
At present, no treatments exist for COVID-19, and it will take a year or more for the vaccine to be developed. However, recent breakthroughs by technology, particularly artificial intelligence (AI) have been successful in detecting and controlling the spread of various diseases.
Could AI help intervene and possibly find a cure for coronavirus ?
How AI helps in detection and diagnosis
While AI may not deliver solutions as to how to treat an illness, there have been several instances where AI-backed technologies have assisted healthcare practitioners in containing a disease from becoming an epidemic. Nevertheless, let us first understand how AI algorithms have worked in various aspects of healthcare.
To correctly diagnose an illness, machine learning algorithms require several concrete examples of patients who have had the disease. These need to be recorded, and digitised, which would help the algorithms to observe patterns among people in the same way as doctors do. For example, a medical professional can diagnose if a patient has lung cancer or stroke based on their CT scans, MRI images or electrocardiograms. The same is in the case of algorithms, the only difference being that algorithms deduce if someone has a medical condition or nor within a matter of seconds, which is both efficient and inexpensive.
AI can detect an illness correctly only when it has information available in the form of symptoms, why the disease occurred, test results, medical scans, and doctor reports, from numerous individuals to conclude.
Besides timely detection and diagnosis, analytical processes in drug development have become more swift thanks to AI. Usually, processes involved in the invention of a new drug or medicine is an arduous and time-consuming process, owing to large amounts of data needed to find relevant patterns among patients, screening of potential compounds, and identifying suitable candidates for drug trials. Here is where AI steps in. By using machine learning to introduce a new medicine to the market, the former can quickly process the large volumes of information, identify the origin of illness and automatically discover functional target proteins.
Secondly, machine learning can forecast the suitability of a molecule through millions of fingerprints and molecular descriptions by processing through millions of potential units and filtering them down to the best options, resulting in the best possible drug design required.
At last, which is also one of the most crucial stages in drug development in clinical trials, is picking out subjects to test the drug or vaccine. Although researchers require a fair amount of time to select candidates for medicinal trials, AI can speed up the process by automatically segregating between good and bad candidates through various patterns, thus, helping scientists produce medicine in much lesser time than usual. Additionally, AI can also provide an early warning signal if a clinical trial is unable to produce the desired output, implying that researchers may have to abandon the drug development and invest in a project that yields results.
Machine learning could also help in discovering specific characteristics that would help doctors and scientists understand why every patient responds differently to the same drug by referring to patients showing similar reactions and cross-comparing their treatments. These characteristics would then assist medical professionals in understanding the needs of each patient better and accordingly, administering the appropriate medication to them.
An example in case is the research conducted by the University of Cambridge. The study used AI algorithms to identify four new types of molecules that can activate a protein that is related to the symptoms of Alzheimer’s and schizophrenia. The result was the discovery of a new drug that was twice as efficient as the one present in the market, and accelerated treatments for various mental illnesses.
Bayer, a German pharmaceutical firm, has been working with technology firms to build software that would assist in diagnosing particular complex and rare medical conditions, as well as developing drugs to treat the same.
Will AI save us from coronavirus?
In the absence of a concrete treatment or cure in place, scientists are applying technology such as interactive real-time maps and sophisticated forecasting to process large amounts of data related to COVID-19. In China, some hospitals began using robots to reduce the workload of their medical staff, warn pedestrians who were not wearing masks, and make use of navigational AI to help ambulances reach faster to the patients. Also, hospitals are utilisingInfervision, an AI-based software that can scan CT images of patients’ lungs and assist doctors in looking for signs of COVID-19.
Almost a week before the WHO warned the public about COVID-19, Blue Dot, a Toronto-based startup had identified the threat posed by it. Using AI, the startup was able to spot the virus via the data gathered from hundreds of sources such as statements made by public health organisations, health reports, air travel advisory, and the co-founder’s experience with treating SARS patients.
Similarly, HealthMap, developed by Harvard Medical School, helps in tracking infectious diseases and had also picked on signs and symptoms of the coronavirus when it first spread in Wuhan in December 2019. The process took place through building a picture of what was emerging, mining data sources, and using them to create a global map of cases that could potentially help model and predict how the virus could impact other regions based on demographics.
At Augusta University, researchers are in the process of developing an application that would help people get an assessment on whether they are carrying symptoms of coronavirus or not, at home. Once the app is in place, it would ask questions about the individual’s background such as their place of residence, demographics like gender, age, racial identity. Besides demographics, the app requires information from users about their travel history, whether they have any signs of fever, cough, shortness of breath, fatigue, headache, or pneumonia, including if they have been in contact with COVID-19 patients, or those who have travelled to regions that are virus hotbeds. With the assistance of gathered data, AI will use an algorithm to examine the details provided by an individual quickly. The app will then send the users a risk assessment report specifying the degree of risk at which they are vulnerable to catching COVID-19 and alert the nearest testing facility if a health check-up is needed. In cases where an individual is unable to the nearest healthcare centre, the app aims to provide mobile health facilities and remote testing.
One of the significant achievements of AI in tackling the growing problem of COVID-19 was through Baidu. Baidu is a Chinese multinational firm that specialises in AI and other Internet-related services. Using a Linear Fold algorithm, Baidu was able to predict the secondary structure of the RNA of COVID-19, which resulted in the reduction of the analysis time of the RNA from 55 minutes to a mere 27 seconds.
Baidu has also built an AI-powered non-contact infrared sensor system that provides users with quick temperature monitoring that can swiftly detect if a person has fever-like symptoms or not. The sensor can examine up to 200 people in a minute and is currently in use by medical practitioners in Beijing.
Another application of AI by Baidu was the collaboration with LinkingMed, a Beijing-based oncology data platform, and medical data analysis firm. With the help of LinkingMed, Baidu released China’s first open-source AI model to an analyse scans of patients with pneumonia, which in turn was powered by PaddlePaddle, Baidu’s open-source deep learning platform. This model helped the company to recognise pneumonia-related lesions within a matter of seconds while providing an assessment of the diagnosis, which specified the number, volume, and proportion of injuries a patient had.
As a result of the implementation in Hunan, a province in China, hospitals were able to identify COVID-19 in people in less than a minute, with the detection accuracy being 92 per cent.
What does the future hold for AI in healthcare?
Currently, it is not sure if AI can provide a cure for COVID-19. AI-backed technologies may encounter several challenges in diagnosing and detecting the coronavirus, since the virus is in the nascent stage in various regions and therefore, identifying patterns among patients may take time. However, governments can make use of AI and possibly find a cure for COVID-19.
Earlier, public health organisations have made use of AI to identify and prevent a disease from becoming an epidemic. An example in case is the National Biosurveillance Integration Center, a part of the US Department of Homeland Security. Through machine learning, the Center built pilot approaches to mine social data to detect flu-like symptoms and examine near real-time emergency medical services and ambulance data, therefore, providing a faster detection of abnormal illnesses.
One cannot help but acknowledge how essential AI is in healthcare. AI not only helps in a more efficient and faster way of disease diagnosis and treatment but can also provide valuable information concerning a patient through factors like travel history and if an illness can spread quickly. Lastly, AI can improve the application of current treatments and conduct several ‘what if’ analyses, which would help policymakers and governments to undertake decisions that are effective and deliver timely medical assistance to people.
About Zubi:
Zubi is an emerging technology empowering platform. Helping students take an exponential career leap by providing learning, assessments and jobs everything in a single place.
