In the midst of a global pandemic, scientists, business leaders, and politicians have turned to technology to supercharge their response, accelerating not just testing and diagnosis but social engagement. Less than six months old, the virus has infected more than 7 million people
and killed more than 400,000. Like never before, the world’s greatest minds have come together to address this once-in-a-generation challenge with the support of artificial intelligence.
AI and machine learning technology are being used to directly study the virus and its spread, speed up the testing and implementation of new treatments, expedite and improve accuracy in diagnosis, and analyze the broader public health impacts of the pandemic. Emotion AI is being used to gauge public response and inform policy decisions around health containment efforts and communication strategy with the public. Both AI and emotion AI are playing an important role in our response to COVID-19 as a society - let’s take a closer look at the specific ways in which they are doing so.
The role of AI in defending against the COVID-19 pandemic
Physicians and clinical researchers have been utilizing machine learning algorithms and AI for several years to support a range of projects. It was a natural fit to incorporate those technologies in addressing the COVID-19 pandemic when it arose in early 2020.
The most important factor in containing and responding to the virus is to know who has it and when. This has been the biggest barrier in many locations, due in part to the long gap between infection and symptoms, and the lack of a robust tracking program to identify infections and potential interactions. AI has been used to address this by helping epidemiologists better understand who has the disease, who is likely to have gotten it or get it in the future from those sick individuals, and to use those findings to ensure the proper resources are allocated where they are most needed exactly when they are needed.
Data is being captured at hundreds of points, including hospital triage, mobile device data, and search patterns to identify where people are sick, when public perception of the disease changes in a single location, and how people are generally responding to the development of infections locally. AI can much more rapidly parse large data sets captured from cameras, medical data, and mobile device data than humans and create a viable map to inform those decisions.
It goes beyond simple data analysis. Carnegie Mellon researchers, for example, are in the process of developing an app that would detect the presence of COVID-19 in someone’s voice by analyzing thousands of voice samples from sick patients. The app has not yet been implemented, but it leverages AI to identify patterns in the human voice that correlate strongly with those who are or have been sick. The initial detection of the disease in Wuhan, China was made by a machine learning algorithm that identified a spike in cases of abnormal pneumonia cases. AI has helped to map the genome and share the data around the globe, allowing dozens of companies to start work on a vaccine in record time. AI is also helping to expedite what is normally a long development cycle for a new vaccine - aiming to compress the timeline to less than two years.
The explosion in the scientific literature on COVID-19 is a testament to the global effort by scientists to find a vaccine and test new treatments, but it has also lead to a nearly impossible to sort database of research. AI is being used to produce the COVID-19 Open Research Dataset, a collection of 44,000 articles on COVID-19 and related viruses that can be accessed by natural-language-processing algorithms to extract valuable data.
A range of applications is being used to support physicians on the front line as well. Suki is an AI-powered voice assistant that has become popular with physicians to help complete their notes in real-time, identifying, and extrapolating context from the doctor’s speech to create accurate notes. The application is able to determine the intent of a doctor’s notes - both voice and written - to create a more accurate record that can then be fed into larger tracking algorithms. This is helping to identify not just confirmed, but unconfirmed or asymptomatic cases of COVID-19.
The CDC has implemented a chatbot powered by an existing Microsoft healthcare chatbot AI. The bot, Clara, is available on the CDC’s website to help individuals determine their next healthcare decision. Curai is another chatbot that helps patients evaluate symptoms and make recommendations for next steps in real-time using deep learning to evaluate COVID-19 symptoms previously identified.
How emotion AI is supporting the fight against COVID-19
Where machine learning and AI are being used to identify patterns in the spread and treatment of COVID-19, emotion AI is being used to measure the emotional and social response to the pandemic.
Cognovi Labs, a behavioral analytics company out of Ohio, is using artificial intelligence to build a Coronavirus Panic Index that actively measures conversations on Twitter, Reddit, Facebook, personal blogs, and other platforms to identify common patterns that correlate to emotions. A machine learning algorithm is then able to measure the overall “mood” of different sectors of society to identify how people are responding to the Pandemic on a scale of 0 to 100. The tool can be used to measure the response to a range of different things but has been extra effective in tracking public response to the pandemic as it has impacted nearly everyone in the country to some degree.
Cognovi Labs isn’t alone in measuring emotional response to the pandemic. Expert System, an AI company in Italy is using their natural language algorithm to measure the general sentiment from 63,000 posts on Twitter every 24 hours. The algorithm uses a set of hashtags and a controlled time of the day to evaluate the general mood of people who reference the pandemic or anything related to it.
Behavioral analytics and emotion AI have the opportunity to greatly impact the financial industry as well. Emotion AI helps banks and financial institutions provide better communication and customer interaction that can help streamline and improve the debt collection process. As the number of non-performing loans soars due to the impact of the economic crisis brought on by COVID-19, this becomes increasingly important. Rather than the typically emotional conversation between creditor and debtor, AI is helping to identify the most effective ways to diffuse a situation, provide support to those who cannot pay their loans, and chart a path to repayment that actually works.
Leveraging AI in the months to come
While most states are re-opening to some degree, we are very much still in the midst of the pandemic. For months or potentially years to come, the “new normal” will force us to rethink almost every aspect of ou lives and economy. The result is change on a scale rarely seen before, and AI will be an integral part of the solution. From supporting rapid response from health care providers to better understanding public response, and how to communicate, the right combination of AI and emotion AI is needed to meet the challenge head-on.
Rana Gujral is an entrepreneur, speaker, investor and the CEO of Behavioral Signals