How AI use cases are evolving in the time of COVID-19

As with nearly every element of the healthcare world, applications of artificial Intelligence and machine learning have been reshaped by the COVID-19 pandemic. Whole business models and strategies became outdated and entirely new approaches have replaced them.

Three industry leaders who compared notes on a recent HIMSS20 Digital presentation all agree: AI and machine learning technologies are key to responding to the coronavirus pandemic, and how they can aid patients and providers.

They will also have an additional role in helping healthcare businesses recover financially while also identifying and serving those most in need when it has ended, the group said.

Their online conversation, Reactions from the Field: Artificial Intelligence | Machine Learning  centered on how AI is being deployed right now to help care for patients, aid researchers in the hunt for therapeutics and vaccines – and how financially battered healthcare systems can use machine learning insights to help optimize their business models moving forward.

Keeping people away from disease hotspots is a priority during the pandemic and this has created a surge in the use of telehealth.

“The aim here is to keep patients safe at home if they can be at home,” said Aashima Gupta, Director of Global Healthcare Solutions at Google Cloud.

Meanwhile, “as telehealth has exploded, AI has helped reduce administrative burden,” said Peter Durlach, SVP strategy and new business development at Nuance Communications. He pointed to areas such as documentation, where automatic creation of clinical notes gives caregivers more time to spend with patients and also makes their workload less burdensome.

Helpful presentation of data is necessary for telehealth providers connecting with new patients or working to provide complicated care in a team setting. Too much information results in a bad session so AI is needed to help organize and present important information.

“You can’t just dump data,” said said Nassib Chamoun, CEO of Health Data Analytics Institute. “You have to make it legible and analyzable. Allow the clinician to rapidly focus on areas of concern.”

AI has a role in answering many other relevant questions as well, the panelists said: When there is a shortage of basic PPE like masks and ventilators, where do they get sent? Which populations are going to need the most help? How can strain be eased on physicians, who were already suffering from burnout before the pandemic struck?

Durlach said AI is being used to “keep physicians distant from the danger if they can be but to also increase the number of patients they can see, or to have some down time,” to avoid physician burnout.

By reducing the administrative burden of paperwork, which studies show doctors do more per hour than they do seeing patients, AI can automate a process that sucks physician time and saps their will, he said.

Additionally, helping patients help themselves can solve a lot of questions and pre-screen those who may need care the most.

Gupta noted that advances in conversational AI, chatbots and rapid-response virtual agents give patients self-serve options that can then be turned into actionable data for caregivers to interact with.

“The heart underneath that is natural language understanding,” she said.

Going forward, as many health systems begin to emerge from lockdown and assess the financial impact of these past few months, AI-driven analytics will be essential to driving cost efficiencies.

“The economic impact is profound,” said Chamoun. “These businesses have a 2-3% margin. They’ve lost a lot of revenue. This is an existential threat to healthcare organizations around the country.”

Because so much of a hospital’s revenue stems from the elective procedures that have been cancelled, trying to bring a major part of the business of healthcare back online will be complicated. AI can ensure that proper documentation and coding generate the highest financial returns, while also finding the most vulnerable patients who need to get care the soonest.

“There is a lot of room for automation,” said Gupta. “Well designed AI implementation” – in areas such as claims processing and assessing at risk patients – “can be automating part of the revenue cycle.”

Benjamin Harris is a Maine-based freelance writer and former new media producer for HIMSS Media.
Twitter: @BenzoHarris.
Healthcare IT News is a publication of HIMSS Media.