The health care sector has increasingly turned to artificial intelligence to aid in everything from performing surgeries to helping diagnose and predict outcomes of patient illnesses.
But as the coronavirus crisis ramps up, and hackers turn their eyes toward the health sector, experts warn these systems and the patients they support are increasingly at risk.
“Obviously any disruption or denial of service of any type of medical health technology which interrupts patient care is definitely a significant issue,” said John Riggi, the senior adviser for cybersecurity and risk at the American Hospital Association (AHA). “Worst-case scenario, life-saving medical devices may be rendered inoperable.”
AI systems have gradually been integrated into health care in the United States, often used to help speed diagnoses, such as reading X-rays, and for determining risks to patients. But with the increase in AI use comes an increase in risks to the networks they rely on.
John Frownfelter, the chief medical information officer for health care-focused AI group Jvion, told The Hill that he believes that, as it stands, hospitals are “losing the battle” against hackers.
“Cybersecurity is getting more and more sophisticated where for hospitals, it’s not their core business, but the bar is getting higher for them to match the efforts to breach them essentially,” Frownfelter said. “I think this trend is going to continue until some paradigm shift occurs.”
Jvion currently provides clinical insights for therapy through AI to about 5 to 10 percent of U.S. hospitals, according to Frownfelter. The coronavirus pandemic is providing new opportunities for companies to deploy AI resources. Jvion recently built a map that uses AI technology to analyze the communities most at risk in the United States from the spread of coronavirus, using factors such as income and employment rates.
But Frownfelter cautioned that if hackers were to target AI systems amid the coronavirus crisis, it could lead to delayed care for patients at a time when every moment counts.
“There are [some] radiologists that are much faster and more efficient at reading imaging stories,” Frownfelter said. “Radiologists can glance at it and confirm that and take a fraction of the time … if you took that away, you would have delays in getting images read and interpreted and getting back to physicians.”
AI capabilities have made a major difference in the ability to treat patients. Ray Page, a Texas-based oncologist and fellow at the American Society of Clinical Oncology, told The Hill that he uses AI to track his patients throughout their cancer treatment.
He noted that the use of these systems has been integrated into “comprehensive” treatments and was apprehensive about the implications of AI systems being compromised by hackers.
“If this gets hacked, or we lose access to those tools, then basically we are going back to the way of using clinical judgment,” Page said.
Hospitals have long been a tempting target for hackers, particularly for ransomware attacks, in which a system is locked out and hackers ask for money to unlock it. Hospitals and health care centers, where access to networks can mean the difference between life and death for patients, are seen as more likely to pay those ransoms.
There have already been worrying signs of malicious actors stepping up their game against hospitals and health care organizations during the coronavirus pandemic.
The website of the Champaign-Urbana Public Health District in Illinois, which serves about 200,000 people, was taken down by hackers for several days, while both the Department of Health and Human Services (HHS) and the World Health Organization were targeted by cyberattacks. Earlier this month, the second largest hospital in the Czech Republic was hit by a cyberattack that led to canceled surgeries and rerouted ambulances.
Riggi told The Hill that AHA, which consists of about 5,000 hospitals and health care groups, was aware of another attempted ransomware attack on an American hospital and ongoing attempts to probe the websites of HHS and state hospitals.
He noted that he would also not be surprised if a nation state such as China attempted to steal AI algorithms as part of intellectual property theft.
“Unfortunately, they are not taking a break,” Riggi said. “The bad guys are using the COVID-19 pandemic to exploit and conduct further frauds … a despicable and heinous crime.”
But concerns over hacking are not the only problem facing AI systems during the coronavirus pandemic.
The Association of American Physicians and Surgeons put out an alert earlier this month warning its members to be wary of AI-powered chatbots being used for diagnosing potential symptoms of coronavirus.
The Association of American Physicians and Surgeons detailed a report from STAT in which the reporter asked multiple chatbots for organizations, including the Centers for Disease Control and Prevention and other established medical groups, whether his symptoms pointed toward coronavirus. The chatbots all gave different answers.
“There are many unknowns,” the group warned members. “The established experts could be mistaken. Use good judgment. Help your neighbor, and don’t panic.”
On Capitol Hill, concerns over the cybersecurity of the health care sector have also not gone unnoticed.
Sen. Mark Warner (D-Va.), the vice chairman of the Senate Intelligence Committee, sent letters to top health care groups including the AHA in 2019 asking questions about what the federal government could do to counter cyberattacks on systems.
Warner told The Hill earlier this month that his concerns around the cybersecurity of hospitals were only heightened by the coronavirus pandemic.
“While we’ve seen the sector make some strides in recent months, we’re still operating from a unnecessarily low security baseline compared to other critical infrastructure sectors, and I fear any weaknesses could be magnified during a crisis such as this,” Warner said.
While it may be some time before the impact of the coronavirus pandemic across the health care sector is known, the crisis is certain to open more doors to using AI in health care. And there are ways AI could help counter broader cyber threats.
Riggi predicted that AI, while still used by hackers to create malware and seen as another route into vulnerable systems, could also speed up the identification of cybersecurity risks to hospitals.
“AI is being deployed in hospitals to identify in a dynamic process all the connected devices … and continuously monitor those medical devices for anomalous behavior that might indicate a potential cyberattack on those devices and highlight potential risk to patient care and safety,” Riggi said.
Frownfelter, of Jvion, predicted that the crisis may also lead to more dependence on AI than ever before.
“I think this is the first time that AI will be recognized as adding value,” Frownfelter said. “The season is now for AI to come into its own, and when you’ve got health systems that are overwhelmed or processes that are overwhelmed, then the power becomes more appreciated.”