AdventHealth, Berg tap AI technology to reduce mortality in COVID-19 patients

AdventHealth has partnered with biotech firm Berg to gain insights on people that have tested positive for COVID-19 and reduce mortality rates from the disease.

AdventHealth, a nonprofit health system based in Orlando, Florida, has diagnosed and treated more than 25,000 patients with COVID-19 to date. With more than 250,000 Americans having died during the COVID-19 pandemic, a key reason for Berg to work with AdventHealth is to understand COVID-19 better and also help triage patients suffering from the virus, explained Niven Narain, Ph.D., co-founder, president and CEO of Berg.

Under the agreement announced Monday, the two organizations will use Berg’s proprietary artificial-intelligence-enabled Interrogative Biology platform with AdventHealth’s patient data. Narain explained that the platform processes biological patient samples on the front end and feeds that into a back-end AI analytical platform. It incorporates machine learning as well as a type of AI called a Bayesian network. With ML, data scientists generate data insights from a hypothesis, but with Bayesian AI the data generate the hypothesis. You then validate the hypothesis in the laboratory and with clinical records, Narain told Fierce Healthcare.

AdventHealth and Berg will build a patient registry biobank that will allow data scientists to find the best treatments for patients with COVID-19. The biobank will incorporate data on all patients that have undergone COVID-10 tests at AdventHealth. Data scientists will study the length of hospital or ICU stays and which medications the health system administered as well as personal medical history and patient outcomes.

Research efforts in Florida will be a model for how Berg and AdventHealth roll out similar work throughout the U.S. The Berg AI platform can provide info on why conditions such as obesity and diabetes leave people more vulnerable to COVID-19, according to Steven Smith, M.D., senior vice president and chief scientific officer at AdventHealth.

RELATED: Northwell Health research arm develops AI tool to help hospital patients avoid sleepless nights

Growing an existing relationship

AdventHealth had already been using Berg’s technology to boost outcomes and develop precision medicine for patients with nonalcoholic fatty liver disease (NAFLD) and sarcopenia, which is a reduction in skeletal mass due to aging. Data scientists from Berg and AdventHealth will collaborate in a similar way with a focus on the COVID-19 pandemic.

“We already had a relationship, and this just allowed us to magnify and grow that,” Smith told Fierce Healthcare.

While the work on NAFLD and sarcopenia was primarily focused on drug discovery and development, for the COVID-19 research, Berg and AdventHealth have developed a data lake from which they have pulled information on positive COVID-19 cases, Smith explained.

The biobank patient registry will launch in two phases: In the first phase, the organizations will release patient demographics, COVID-19 clinical information and patient medical histories. In the second phase, the organizations will incorporate data from across AdventHealth locations in multiple U.S. states. They will also analyze how chronically administered medications could be linked with a better outcome or lower probability of SARS-CoV-2 infection, the strain of coronavirus responsible for COVID-19.

RELATED: The unexpected ways AI is impacting the delivery of care, including for COVID-19

Gaining insights from AI

A desired result of the research is to generating risk engines and understand how to triage patients better, Smith noted.

“There are a few risk engines out there in the literature,” Smith said. “They perform OK—they're not great—so there's certainly the potential for advancement in that way.”

Smith’s team at AdventHealth will receive updates from Berg on the data approximately every 30 days as the second wave of the COVID-19 crisis gets underway, according to Narain.

“This is what makes this so important because the next few months are going to be presumably very difficult, and our goal of this project is to try to deliver insights and answers while this is all going on,” Narain said. “As quickly as we get it, we’ll be feeding it into the system and every 30 days or so update the insight.”

RELATED: Healthcare is ramping up AI investments during COVID. But the industry is still on the fence about Google, Amazon. Here's why

The AI data will inform doctors regarding which medications, like remdesivir and dexamethasone, work on certain populations, Narain explained.  

“It will bring together the efficiency among physicians, patients, and drug developers so that the ecosystem of information sharing becomes easier,” Narain said.

Smith added that the advantage of AI is to be able to provide key insights in a way researchers didn’t anticipate.

“The broad idea of AI layered on top of rich data is to be able to break those chains so to speak around how we think about particular illnesses and flip that upside down,” Smith said.