An usual diagnostic tool

Through machine learning, researchers at Alfaisal University can scan tweets in 35 different languages to track Covid-19

Nidal Nasser, a software engineering professor at Alfaisal University, based in Saudi Arabia, has developed artificial intelligence (AI) techniques to pinpoint whether someone has Covid-19 based on their social media stream. In 2020, the highly contagious Covid-19 virus spread around the world in a matter of months, and its various mutations continue to seed new infection hot spots. At Alfaisal University, researchers are using AI to determine whether someone is infected with the virus based on their tweets.

“There are currently no regulations to do these checks,” says Dr Nasser, founder and director of the university’s Internet of Things research lab. “This is a way that we can screen their tweets and be able to tell if they are infected or have been in contact with someone who has been infected.” Dr Nasser is a professor of software engineering at the university. 

The technique uses machine-learning software, which is trained using a given data set and is then able to extract knowledge from other data sets. In the case of their Covid-19 tracker, Nasser and colleagues trained computer algorithms to identify words, phrases and context that would indicate whether a person has Covid-19 or has had it in the past. “We try to categorise the context of the words,” Dr Nasser says. “We’re not just interested in the word ‘Covid’, but also the words around it, such as ‘infected’ or ‘positive’.” The researchers’ platform allows them to mine tweets in 35 languages, including English, German and Dutch, widening their ability to identify travellers who may be infected with the novel coronavirus.

The social media Covid-detector is just one of the ways Alfaisal’s researchers are harnessing the power of AI. “AI is poised to lead the way to the next phase of the digital revolution, with the potential to disrupt every industry and make it more innovative and productive,” Dr Nasser says. This sentiment is shared by the Saudi government.

In 2019, the Saudi Data and Artificial Intelligence Authority was established. “We are living in a time of scientific innovation, unprecedented technology and unlimited growth prospects,” says the country’s crown prince, His Royal Highness Mohammed bin Salman bin Abdulaziz Al Saud. “These new technologies such as AI and the Internet of Things, if used optimally, can spare the world from many disadvantages and can bring to the world enormous benefits.”

AI and its associated industries are integral to the country’s Vision 2030, a strategy that defines the Arab nation’s future trajectory and economic growth. By 2030, the adoption of AI is expected to contribute about $135 billion (£96.29 billion) to the country’s GDP.

The field is coming into its own, and Dr Nasser and colleagues are using AI to develop other applications that could help authorities and health providers during the pandemic and beyond. One of these projects looks to assist local authorities with sanitising educational institutions. Although the pandemic rages on, governments and educators are trying to ensure that the education system is not irreparably disrupted. Unfortunately, however, educational institutions are hubs of infection and entering potentially contaminated classrooms can put cleaners at risk of contracting Covid-19.

Dr Nasser and his team have developed a way for drones to disinfect classrooms using a type of ultraviolet light known as UVC. Research has shown that UVC, which is the most energetic type of ultraviolet light, can inactivate other coronaviruses and it is thought to do the same with the new virus. If human skin is directly exposed to UVC, the light can cause severe blistering. The drone-based UVC disinfection technique would model paths for the drones, allowing them to sanitise education institutions and mitigate the spread of the coronavirus, says Dr Nasser. 

The third Covid-related project that Alfaisal’s researchers have undertaken hinges on a smart healthcare framework, which allows healthcare workers to detect Covid patients from lung X-rays and to monitor them remotely in real time. “The system uses smart sensors to collect data from medical images,” explains Dr Nasser. These sensors, which upload data to an online repository, allow healthcare workers to monitor patients’ well-being in real time. 

Crucially, researchers are also training the platform to identify Covid-19 from the lung images. Coronavirus is a respiratory disease and, in many cases, a person’s lungs become irritated and even infected. Healthcare workers and scientists can see this inflammation on X-rays. Dr Nasser and colleagues trained their platform using thousands of images of people’s lungs, meaning the “intelligent” software can identify the characteristics of Covid in lung images.

Though the world’s focus is on Covid, the potential applications of AI are vast. The same multifunctionality holds true for the social media Covid tracker, Dr Nasser says. “We are researchers and we built the platform,” he explains. “We have produced the platform and anyone can use it – the technique is plug and play.” For example, Dr Nasser explains, the social media tool could be used to track consumer sentiment: “From tweets, the business owner can get feedback.” Armed with that information, they can make strategic decisions about the product and their business.

This is the future that the world is moving towards, one in which the Internet of Things and AI can give people – from business owners and healthcare workers to entire governments – the ability to make better decisions.

Find out more about the College of Engineering at Alfaisal University.