How the COVID-19 Pandemic Revealed AI’s Potential in Healthcare

8

By Joel Landau

To date there have been only glimpses of how artificial intelligence might be used in the battle against the coronavirus, much less future public health crises. They are promising glimpses, certainly, but no more than that.

Consider, for instance, that on Dec. 31, 2019, an AI platform known as BlueDot—based in Toronto—took note of an unusual cluster of pneumonia cases in Wuhan, China. That was nine days before the World Health Organization issued an alert about the coronavirus. Nine crucial days. Nine days that could have made an enormous difference, had BlueDot’s warnings been heeded.

Across the world, other AI systems have provided the basis for COVID tracking and treatment, though privacy legislation has struggled to keep up with the ongoing changes. Most notably, AI was used successfully in South Korea to slow the spread of COVID-19—but at the cost of a mass collection of private data. Seoul collected data on everything from credit card transactions to CCTV footage for the sake of tracking and containing the virus. 

The country has even developed an app that warns officials about breaches of quarantine. While the privacy implications of these practices are concerning, there’s no doubt that intelligent applications of AI can aid in containment of the pandemic.

The difficulty in using AI to combat the coronavirus lies in the justification of suspending privacy for the sake of public health. While containment is important, it can be difficult to close the figurative Pandora’s box of state surveillance. 

Some countries, including the US, UK, and Germany, are exploring avenues that would allow for the use of anonymous data to combat COVID-19, effectively sidestepping this issue. However, in remote areas, anonymous tracking may not be as much of a privacy protection as intended, as it is still easy to determine the origin of the data. Many contact-tracing apps in development allow users to opt in, though this can hamstring the effectiveness of such measures.

The European Commission has released guidance for contact-tracing applications, including the use of a decentralized approach to protect sensitive data from hackers. Regardless, any tracking measure in Europe will need to comply with GDPR standards. Even with safeguards in place, a gold rush of developers creating contact tracing apps without guaranteeing the privacy or security of personal data has left governments to try to sort the wheat from the chaff.

Arguably, the worldwide threat posed by COVID warrants some suspension of privacy in order to bring the pandemic under control. But any solution needs to be transparent and have a plan to roll back surveillance after the pandemic ends. AI has the potential to inform public health decisions moving forward, but it also represents a new frontier for data which can easily infringe on privacy rights.

Contact tracing isn’t the only COVID-related application for AI. A new algorithm in China claims to diagnose new cases with 96 percent accuracy within 20 seconds. Similar diagnostic tools can alleviate time spent testing patients and help triage as fast as possible. 

Indeed, simplifying logistics in hospitals is one of the main benefits of AI—with healthcare systems often strained to the limit, finding creative ways to free up time for personnel and keep them safe is more important than ever. For instance, AI use can be as simple as using a speech-to-text software to transcribe doctor’s notes and save hospitals countless hours in paperwork or digital conversion, the latter of which has been a challenge for some time.

The value of diagnostic tools will only rise as time goes on. The shortcoming of data-based AI solutions has always been the unreliability of data and the massive amount of labor required for data scientists to distill it down into usable insights. 

However, more sophisticated methods of identifying relevant data and the iterative nature of machine learning points to AI systems becoming more efficient and effective over time. Better data models can help diagnose cases, as well as develop larger maps of how infections spread and even predict the severity of a disease. The downside to this is that AI adoption isn’t an instantaneous process—it must be taught, and data scientists that can do so are in high demand.

Robots represent another piece of the logistical puzzle that can be tackled by AI. Autonomous vehicles can help safely deliver supplies to doctors or quarantined families. Though the human touch will never be irrelevant in healthcare, some robots can even be valuable in assisting with surgery or supplementing human care for patients.

Clearly, then, the pandemic has revealed AI’s wide-ranging potential in the public-health arena. It can be used to track outbreaks, diagnose viruses, and even serve as an aid for those who treat them. There is little doubt that AI’s role in healthcare will only increase in the years to come — that it will be crucial not only in stemming the tide of future pandemics but also in improving patient outcomes in the course of everyday care.

**Joel Landau is founder and chairman of The Allure Group, a network of six New York City-based skilled nursing facilities. He has served as a member or an advisor on a number of boards and committees, including the Medicaid Managed Care Advisory Review Panel (MCCARP), NYS DOH Preventative Health and Health Services Block Grant, NYS DOH Task Force on Long Term Care Financing, and the Brooklyn Chamber of Commerce.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

16 − four =

This site uses Akismet to reduce spam. Learn how your comment data is processed.