- Healthcare artificial intelligence (AI) can help reduce the time for processing chest x-rays and prioritize which x-rays need to be seen by an expert radiologist sooner, according to new research from the University of Warwick.
The AI system developed by researchers at the University of Warwick's WMG has been able to cut the average delay on reviewing critical chest x-rays from 11 days to 2.7 days and from 7.6 to 4.1 days for urgent chest x-rays.
The research team, led by Giovanni Montana, worked with Guy’s and St Thomas’ NHS Hospitals to extract a dataset of 470,388 anonymous adult chest x-rays acquired from 2007 to 2017 and developed a computer vision AI system that can recognize abnormalities in the x-rays and indicate how quickly the x-rays should be reported to an expert radiologist.
Researchers developed and validated a natural language processing (NLP) algorithm that can read a radiological report, understand the findings of the reporting radiologist, and infer the priority level of the exam.
The NLP system analyzed the free-text report to prioritize each radiograph as critical, urgent, nonurgent, or normal.
The team applied this algorithm to the historical x-ray exams, generating a large volume of training exams that allowed the AI system to determine which visual patterns in x-rays were predictive of their urgency.
Normal chest x-rays were detected with a positive predicted value of 73 percent and a negative predicted value of 99 percent.
“Artificial intelligence-led reporting of imaging could be a valuable tool to improve department workflow and workforce efficiency,” Montana said.
“The increasing clinical demands on radiology departments worldwide has challenged current service delivery models … It is no longer feasible for many radiology departments with their current staffing level to report all acquired plain radiographs in a timely manner, leading to large backlogs of unreported studies,” Montana noted.
“The results of this research show that alternative models of care, such as computer vision algorithms, could be used to greatly reduce delays in the process of identifying and acting on abnormal X-rays, particularly for chest radiographs which account for 40% of all diagnostic imaging performed worldwide. The application of these technologies also extends to many other imaging modalities including MRI and CT,” he added.
Healthcare artificial intelligence has had big impact on radiology so far, according to Mutaz Shegewi, research director at IDC Health Insights.
“AI holds much promise for healthcare. For example, radiology is the best example we have of the impact of AI. We are starting to see medical device manufacturers embed AI technology that allows radiologists to do their job better and more conveniently,” Shegewi told HITInfrastructure.com.
“There is also an interesting application of AI in the clinical documentation and workflow. We are seeing vendors introducing algorithms for population health. Another advanced application of AI we’ve seen recently is embedding virtual assistants to facilitate clinical documentation and workflow similar to how we would use Siri on our Apple phones to perform various functions,” Shegewi added.
To be effective, AI requires robust IT infrastructure and data standardization to ensure it can deliver the desired results.
Market research firm Frost & Sullivan forecasts that the market for AI-driven health IT applications will total $1.7 billion in 2019 and continue growing at a compound annual growth rate (CAGR) of 68.5 percent through 2022.
Application of AI platforms to healthcare workflows will result in a 10 percent to 15 percent productivity gain over the next two to three years, Frost & Sullivan estimates.
“We are seeing AI being used to convert health data from wearables and other mobile devices into actionable insights for the patient … This is going to happen massively in 2019,” Kamaljit Behera, transformational health industry analyst at Frost & Sullivan, told HITInfrastructure.com.