One of the many challenges for healthcare providers during the COVID-19 pandemic has been determining which patients are at risk for developing life-threatening medical conditions in order to monitor them more closely and, ideally, intervene sooner.

With that goal in mind, researchers at NYU Grossman School of Medicine, recently developed an AI program that used, among other data points, chest X-ray images to predict the development of life-threatening issues within four days.

According to their report, published in npj Digital Medicine, their tool was effective with 80 percent accuracy, although they noted that they only focused on chest X-ray images due to the nature of COVID-19, and thus cannot predict non-pulmonary-related complications.

“Emergency room physicians and radiologists need effective tools like our program to quickly identify those COVID-19 patients whose condition is most likely to deteriorate quickly so that health care providers can monitor them more closely and intervene earlier,” said study co-lead and assistant professor in computer engineering at NYU’s Abu Dhabi campus Farah Shamout, PhD, in a related press release.  

In addition to X-ray information, the researchers input vital signs, age, gender, and race into the program, which then used the data to assess the patient’s risk level. The program was trained using several hundred gigabytes of data gleaned from 5,224 chest X-rays taken from 2,943 seriously ill patients infected with SARS-CoV-2, the virus behind the infections.

“Additionally, our results suggest that chest X-ray images and routinely collected clinical variables contain complementary information, and that it is best to use both to predict clinical deterioration. This builds upon existing prognostic research, which typically focuses on developing risk prediction models using non-imaging variables extracted from electronic health records,” the study stated.  

Researchers then tested the predictive value of the software tool on 770 chest X-rays from 718 other patients admitted for COVID-19 through the emergency room at NYU Langone hospitals from March 3 to June 28, 2020.  For four out of five patients, the program accurately predicted which patients would go on to receive mechanical ventilation, intensive care, or would die within four days of hospital admission.

“We believe that our COVID-19 classification test represents the largest application of artificial intelligence in radiology to address some of the most urgent needs of patients and caregivers during the pandemic,” explained Yiqiu “Artie” Shen, MS, a doctoral student NYU’s Data Science Center. 

Study senior investigator Krzysztof Geras, PhD, an assistant professor in the Department of Radiology at NYU Langone, said a major advantage to machine-intelligence programs such as theirs is that its accuracy can be tracked, updated and improved with more data. 

“Our findings show the promise of data-driven AI systems in predicting the risk of deterioration for COVID-19 patients, and highlights the importance of designing multi-modal AI systems capable of processing different types of data,” the study explained, adding, “We anticipate that such tools will play an increasingly important role in supporting clinical decision-making in the future.”  

Photo by Jackie Niam/Getty Images