Can Artificial Intelligence(AI) predict the next pandemic?
Today, when we talk about technology artificial intelligence (AI) is the first best thing that one can think of. What is it that artificial intelligence cannot help us with? From self-driving cars to face recognition in the phone, AI always takes a front seat when it comes to making our lives easier. In the contemporary world, it is being hypothesized that artificial intelligence can help us predict pandemics as well. Basically, you would know when your country or this world, in general, could go into lockdown mode again. This would allow us to be prepared mentally, emotionally, physically, and socially to deal with such a COVID-like situation, if it strikes again.
BlueDot
BlueDot is a company whose outbreak intelligence platform safeguards lives by assuaging subjection to infectious diseases that jeopardizes human health and security. Its main idea is to “spread knowledge faster than the diseases spread themselves”. It was apparently one of the first to sound the alarm of the onset of a possible infectious disease- COVID-19. It used an artificial intelligence-powered algorithm that analyses data from all over the planet to identify yet another outbreak. The algorithm loops a search every 15 minutes, 24 hours a day, and in over 65 different languages. When the Chinese news stations started reporting seven people getting hospitalized from a mysterious flu-like illness, within hours of that BlueDot warned health officials in 12 countries.
BlueDot discovered what would come to be known as coronavirus. It was nine other days when World Health Organization officially released a statement alerting people to the emergence of a novel coronavirus. The best part about this is that they are not using artificial intelligence to replace human intelligence.
It’s basically using AI to find needles from a haystack and presents them to its team, which is analyzed by a group of physicians and computer programmers. They do a gut check and create the reports that are sent out. Not only do they detect the diseases, but they also work towards understanding the disease and how it may disperse to different parts of the world. They also attempt to determine the potential consequences of the disease spreading. These predictive tools can also draw on air travel data to assess the risks that transit hubs might see infected people either arriving or departing.
Obstacles
Companies these days use an array of natural language processing (NLP) algorithms to keep an eye on news outlets and official healthcare reports in different languages all over the world, flagging whether they mention high-priority diseases, such as the coronavirus. The results are realistically precise.
Some companies focus on social media, as well. Taking the instance of Stratifyd, a data analytics company in Charlotte, North Carolina is developing an AI that scans posts on various social media platforms such as Instagram, Facebook, and Twitter and cross-references them with descriptions of diseases taken from sources like the National Institutes of Health, Food and Drug Administration and Administration for Community Living, which stores genome sequencing information. This clearly depicts how developed machine learning has become over the years.
However such approaches render less accurate as the epidemic progresses as it has been hard for AI to get the sort of reliable data that it needs about COVID-19. News sources and official reports offer inconsistent accounts. In previous years, we saw how much confusion there was when it came to the symptoms of the virus and how it was actually spreading from one person to another. The media dramatizes things, the authorities shrug things off. Predicting the spread of disease via hundreds of sites in dozens of countries is a far more daunting task than forecasting where a single outbreak may start spreading in the first few places.
Lack of diagnostics testing is one of the major obstacles. To know what will happen next, one needs to know what’s happening right now. To predict the future, knowledge of the present is a prerequisite. People’s behaviors, reactions, etc. shall be known. Knowledge of what is going on inside the hospitals is necessary. If government agencies do not lock away their public health data, prediction can become very easy. Therefore, this gets us to the fact that AI has to rely on data available online like news. Even the strategies to get all this data from reliable sources to be useful in this area can be controversial.
Medical records should be opened up for analysis. In this manner, AI would be able to identify people who are more at risk from the pandemic due to an underlying condition. And henceforth, the resources could be focused on the people who need them the most. This would simplify the process of prioritizing resources. Viruses do not operate within geopolitical boundaries. It is a virus. It doesn’t require a passport or visa to travel from one country to another. That is why there is a requirement of sharing health data between countries. An international agreement to release real-time data on diagnoses and hospital admissions which could be fed into global-scale machine-learning models of a pandemic could be brought into force.
But then again, different countries have different privacy regulations for medical data. Finding an agreement on international agreements takes time. Therefore, as of now, we shall make the most of the data that we possess. We have to make sure that human intelligence interprets what artificial intelligence predicts, all the while ruling out the predictions that don’t ring to be true.
Early diagnosis
As well as predicting the course of an epidemic, many hope that AI will help identify people who have been infected. AI has a proven track record here. Machine-learning models for investigating medical images can catch early signs of disease that human doctors miss. But such models require a lot of data than we think.
Alexander Selvikvåg Lundervold at the Western Norway University of Applied Sciences in Bergen, Norway, who specializes in machine learning and medical imaging, says we should expect AI to be able to detect signs of Covid-19 in patients eventually. But we are not sure if imaging is reliable. It is hard to assess the accuracy of approaches posted online as little training data is available. Most image recognition systems—including those trained on medical images—are adapted from models first trained on ImageNet, a widely used data set encompassing millions of everyday images.
Role of Social Media
Social media could be instrumental in predicting the next COVID outbreak. According to the research, social media data – anonymous information amassed from billions of accounts – presented significantly more precise insights compared to conventional location data modeling algorithms. With over 2 billion Meta users worldwide, the research concludes that social media data is a better means for envisaging viral spread when contrasted against location data from an owner’s mobile device. This is due to the abundance of user data available, even in remote regions of the world where mobile cell phone data is not obtainable. This privacy-protected data is composed and amassed through machine learning algorithms to track the mobility of users as they travel across the globe.
The technology can also identify any keywords with which infected or recovering users type to communicate their experience with the disease, making social media the new go-to source of real-time raw data regarding outbreak trends.
Role of Artificial Intelligence in Treatment
For the development of cures for diseases via artificial intelligence, data is again essential. Using generative design algorithms, which produce a vast number of probable results and then sort through them to draw attention to those that are worth looking at more closely is one of the ways for identifying possible drug candidates. SRI International is collaborating on such an AI tool, which uses deep learning to generate many novel drug candidates that scientists can then assess for efficiency. Although this could turn tables in ways that can’t even be imagined, it still may take months before a possible candidate becomes a viable treatment.
