We are living through an unprecedented crisis. During the COVID-19 pandemic, healthcare workers have emerged as frontline heroes, working overtime to protect our communities from the spread of novel coronavirus. But they aren’t immune to the anxious, uncertain atmosphere the pandemic has fostered nor, indeed, the coronavirus itself.
We need to protect the first responders and hospital staff who put their wellbeing on the line to support their communities during a crisis. To my mind, that means using every tool at our disposal to the fullest — with AI chief among those at hand.
There’s little doubt that the current situation demands a creative solution. The United States has become the center of the global pandemic; as of April 16th, the US confirmed 644,188 cases and endured 28,579 deaths. Despite efforts to flatten the curve by ordering regional shut-downs and stay-at-home orders, hospitals across the county have been all but overwhelmed by incoming cases. The impact on provider morale has, according to reporting from NPR, been similarly problematic.
“Nearly a month into the declared pandemic, some health care workers say they’re exhausted and burning out from the stress of treating a stream of critically ill patients in an increasingly overstretched health care system,” NPR reporters Will Stone and Leila Fadel recently wrote. “Many are questioning how long they can risk their own health […] In many hospitals, the pandemic has transformed emergency rooms and upended protocols and precautions that workers previously took for granted.”
Hospitals are doing all they can to keep their caregivers safe and protected, but their resources are stretched far too thin. According to reports, some hospitals in high-infection areas like New York City can only afford to give healthcare workers one N95 mask every five days. Used masks are collected, disinfected, and returned on a cycle between uses. But some frontline workers worry that, given the highly contagious nature of the disease, they may not be adequately protected.
“It can be disheartening to have that feeling of uncertainty that you are not going to be protected,” Sophia Rago, an ER nurse based in St. Louis, told reporters for NPR.
We need to shield our frontline workers as much as possible. The obvious solution would be to increase stores of personal protective equipment (PPE) and N95 masks; however, given that we face a national shortfall and harsh state-to-state bidding wars over the gear, that fix seems unlikely. What we can do to at least lessen the risk of patient-to-provider transmission is to invest in AI-powered solutions that can automate some healthcare protocols and limit the need for close contact.
“Traditional processes — those that rely on people to function in the critical path of signal processing — are constrained by the rate at which we can train, organize, and deploy human labor. Moreover, traditional processes deliver decreasing returns as they scale,” a team of digital health researchers recently wrote in an article for the Harvard Business Review.
“Digital systems can be scaled up without such constraints, at virtually infinite rates. The only theoretical bottlenecks are computing power and storage capacity — and we have plenty of both. Digital systems can keep pace with exponential growth.”
These AI-powered, digitally-facilitated solutions generally fall into two broad categories: disease containment and patient management.
Assessing AI’s Ability to Limit Disease Transmission
When it comes to limiting disease spread, the aim is to use AI tools to allocate human resources better while still protecting patients and staff. Take the screening system that was recently deployed at Tampa General Hospital in Florida, for example. This AI framework was designed by the autonomous care startup Care.ai and intended to facilitate early identification and interception of infected people before they come into contact with others. According to a report from the Wall Street Journal, the Care.ai tool taps into entryway cameras and conducts a facial thermal scan. If the system flags any feverish symptoms such as sweat or discoloration, it can notify healthcare staff and prompt immediate intervention.
Other technology companies––Microsoft, for one––have rolled out similar remote diagnostic and alert tools in facilities across the globe. Their unique capabilities vary, but their purposes are the same: to prevent the spread of infection and provide support to overworked personnel.
As representatives for Microsoft shared in a recent press release, “[AI technology] not only improves the efficiency of epidemic prevention, but it also reduces the work burden of frontline personnel so that limited human resources can be used more effectively.”
In these resource-strapped time, the aid is undoubtedly needed.
AI’s Applications for Diagnostics and Patient Management
Fighting a pandemic is a task that requires speed. Now more than ever, providers must be able to accurately and quickly identify infected patients so that they can trace and hopefully contain the viral spread. But doing so isn’t an easy order.
To borrow a quote from Forbes contributor Wendy Singer, “Analyzing test results nowadays requires skilled technicians and a lot of precious time, as much as a few days. But in our current reality, healthcare systems need to analyze thousands of results instantly, and to expose as few lab workers as possible to the virus.”
We don’t have that kind of time––and we can’t put our lab workers at undue risk. Thankfully, cutting-edge AI technologies may provide a solution. With AI, hospitals can automate some steps of the testing process, cutting down on the time and effort needed to process test results. These capabilities aren’t just hypothetical; in the weeks since the start of the pandemic, the health tech startup Diagnostics.ai has provided laboratories in the US and UK with a diagnostic tool that streamlines the testing process by automating DNA analysis.
However, the applications of AI diagnostics aren’t limited to testing alone. Some have also used artificial intelligence to support population management in overstretched hospitals. One Israeli medical-device developer, EarlySense, recently developed an AI-powered sensor that can identify which patients will most likely face complications like sepsis and respiratory failure within six to eight hours. This can give a hospital the information it needs to best allocate limited resources and staff attention.
No AI innovation — no matter how brilliant or helpful — will fix our resources shortfall. There is no question that healthcare providers need more PPE and support, or that they need it immediately. However, the benefits that AI provides to screen and patient management efforts are evident. It seems reasonable that we at least consider the weight the deployment of such tools could remove from our exhausted front-liners’ shoulders.