An AI-based system proved its mettle in screening patients in immunotherapy trials, according to Dan Gebow, chief innovation officer at Bioclinica.
Early in the COVID-19 pandemic, developers of cancer immunotherapies worried that the novel coronavirus would interfere with their clinical trial results.
In previous years, the scientific community established that immunotherapy can rarely cause interstitial lung disease (ILD). An umbrella term covering several conditions such as pulmonary fibrosis that lead to scarring of the lungs, ILD also arises in some patients with COVID-19.
The fact that ILD can arise from COVID-19 and cancer immunotherapies complicated oncology clinical trials, recalled Gebow at Bioclinica, which provides clinical trial adjudication products and services. “Patients in immunotherapy clinical trials were showing up at the emergency room with some type of lung infection,” Gebow said. “You can imagine if you’re the pharmaceutical company with a clinical trial in cancer immunotherapy, you’re worried that your treatment is going to get artificially blamed for COVID-19 infections,” he added.
Yet figuring out the cause of the lung problems remained challenging in early 2020 when PCR tests for COVID-19 were not widely available.
Dan Gebow
Traditional adjudication methods can move at a snail’s pace. “An image goes to a radiologist who reads it sometimes a month later,” Gebow said. But for oncology clinical trials involving elderly patients in the middle of a pandemic, such a lag is out of the question. “In the murky days of early COVID, it became critical to discern problems early on,” he added.
To address the challenge, Bioclinica collected chest CT images and routed them to radiologists asking for perspective on whether the lung issues were related to immunotherapy.
The company also developed an artificial intelligence–based technology in April 2020 to analyze the radiology images. “When the chest CT is uploaded to our cloud servers, we use AI to screen it really quickly to see if there’s a possibility that that patient had COVID,” Gebow recalled. If a COVID-19 infection was likely, the system automatically routed a query asking the research site to provide any records related to COVID-19.
The system also drew feedback from experts in diagnosing the emerging constellation of symptoms present in many COVID-19 patients.
Ultimately, the system proved its mettle in screening patients in immunotherapy trials, Gebow said. “Now, we’re extending [the technology out far beyond COVID,” he added.
There are countless possibilities. An AI-enabled clinical trial adjudication can, for instance, screen for patients in an Alzheimer’s trial to determine if they have a brain tumor that could be affecting their cognition.
The clinical trials space still has significant room to evolve — “especially now with artificial intelligence coming in,” Gebow said. “It opens all new horizons of things that could speed up clinical trials.”
Filed Under: clinical trials, Drug Discovery
