4 Ways Clinical AI Will Transform Healthcare in 2021

The past year has pushed the U.S. healthcare system to its limits. Hospitals filled up with COVID-19 patients, putting elective procedures and in-person visits on hold, and more than 14.6 million Americans lost their employer-sponsored health insurance.

However, the pandemic has also shown how resilient health systems can be, adapting to new technologies on the fly to continue serving patients through this disruptive time. Although telehealth is often discussed as the star technology of the COVID-19 pandemic, clinical artificial intelligence (AI) also had a watershed year.

A recent survey found 56 percent of healthcare executives accelerated their AI deployment plans in response to the pandemic. Similarly, in an Intel survey the proportion of health IT decision-makers that had deployed, or were planning to deploy AI, jumped from 45 percent early in 2020 to 84 percent after the pandemic hit. There are several reasons for this increased adoption.

AI has helped hospitals manage capacity, triage CO-19 patients, and target outreach to vulnerable patients and populations to prevent avoidable hospitalizations. Importantly, AI has also eased clinicians’ administrative burden, which worsens burnout among doctors and nurses already pushed to the brink by the pandemicand costs the US $4.6 billion annually.

The result is better patient outcomes and lower costs — critical at a time when 40 percent of acute care organizations are at risk of closing due the financial impact of the pandemic.

At Northwell Health, New York State’s largest health system, we use clinical AI from Jvion to support our transitions of care management (TCM) program. Our analysis found that patients who received interventions recommended by AI had 23.6 percent fewer readmissions than similar patients without AI interventions, saving $11,200 for each avoided readmission.

Having seen what AI can do, here are four predictions for how AI will impact healthcare in 2021:

Enabling the Shift to Value-Based Care

Under the direction of Joe Biden’s pick for Secretary of HHS, Xavier Becerra, we will almost certainly see the shift to value-based models of care accelerate. Expect a push for greater participation and more mandatory value-based payment programs. Once we reach a tipping point where 50 percent of revenue is value-based, providers’ financial performance will be directly tied to their ability to improve patient outcomes.

Prescriptive forms of clinical AI will help healthcare organizations thrive in this transition, leveraging data to guide clinical decisions for better patient outcomes.

This will be particularly valuable when elective procedures are again being delayed to clear capacity for the surge in COVID-19 patients. Healthcare utilization could dip back to the levels we saw back in April, when cancer screenings were down over 85 percent. Patients who defer necessary care now will be more likely to have worse health in the future, increasing costs for value-based care providers.

AI can be a powerful tool for identifying patients at risk of deteriorating and recommending targeted interventions tailored to their needs. Providers and payers can then proactively engage their vulnerable patients and health plan members early, providing guidance to help manage their conditions and avoid hospitalizations.

Personalized Medicine

No two patients are the same, and there’s no one-size-fits-all approach that works the same for every patient. Genetics, diet, lifestyle, and socioeconomic status are just a few of the variables that intersect to influence a person’s health. With AI, providers can tailor their care plans to patients’ unique circumstances.

At Northwell, our transitions of care teams use AI to see the external factors that put patients at greater risk of readmission after they’re released from the hospital. This includes things like whether patients can afford their medication or if they can access a pharmacy. The AI also recommends interventions our care teams can take to address these external risk factors, for example enrolling patients in mail-order pharmacy or other benefits.

As the technology advances in the future, AI will only become more powerful for personalizing medicine. By analyzing genetics and other biometrics, AI could even predict which medication will be most effective for individual patients, or inform the development of more personalized drug classes.

Preventing Suicides and Overdoses

On top of the pandemic, America is dealing with a national mental health crisis. Months of social isolation, unemployment, civil unrest, and ambient tragedy aren’t helping either. Earlier this year, the CDC found that 41 percent of Americans showed signs of anxiety, depression, substance abuse, and PTSD from the pandemic. It will take months to get the complete picture, but early data suggests 2020 will be the worst year ever for overdose deaths as well.

Behavioral health providers, already stretched thin before the pandemic, are struggling to keep up with the increased demand, and patients are tragically slipping through the cracks. However, AI can help behavioral health providers predict which patients are at risk of suicide, depression and substance abuse. With this insight, providers can reach out to vulnerable patients and offer them the help they need before it’s too late.

AI can also help non-behavioral health providers understand how mental illness increases patients’ risk for other negative outcomes. For example, a depressed patient may be less likely to exercise, eat healthy, or take their medication as prescribed. With a vast trove of data to pull from, AI can make these connections and help providers adapt their care plans accordingly.

Understanding Social Determinants of Health

The pandemic has made it painfully clear how social and economic factors impact our health. Low-income essential workers have the most exposure to the coronavirus, more chronic conditions that increase their risk for severe infection, and the least ability to cover the costs of care. To make matters worse, 50 percent of low-income adults skipped care they needed because they couldn’t afford it. These patients will go on to have worse health outcomes in the future.

AI’s ability to make connections in data can help health systems better account for these social determinants of health (SDOH) in their population health investments and individual care plans. Analyzing data from the Census and other government agencies, AI can reveal which patients live in areas with high rates of unemployment, housing instability, food insecurity, air pollution, and other risk drivers.

Understanding the impact of SDOH will be critical to mitigating the effects of the pandemic on the nation’s most vulnerable individuals. In March, Jvion released the COVID Community Vulnerability Map, which shows the communities, with resolution at the Census block level, are at greatest risk for severe cases of COVID-19 based on social, economic, and environmental factors. The map has helped target resource allocation and public health outreach to protect these vulnerable communities.

In the coming months, expect to see AI leveraging a combination of SDOH data and clinical data to help guide vaccine distribution efforts to the populations at greatest risk. And once the current surge subsides and it’s safe to resume elective procedures and other routine care, AI will help healthcare organizations lower the socioeconomic barriers to care that drive disparities in health outcomes, unlocking a more equitable healthcare system.

Dr. Zenobia Brown, M.D., is Vice President and Medical Director, Northwell Health Population Health Management. Dr. John Frownfelter, M.D., is Chief Medical Information Officer, Jvion