As the amount of healthcare data grows daily, one of the increasingly vexing issues revolves around how to determine how to prioritize diagnostic data to ensure prompt and effective care.
With that goal in mind, a recent study from the UK’s University of Warwick has shown how AI can significantly reduce the time required to ensure that abnormal chest X-rays, with serious findings, receive expert attention as soon as possible.
According to the study’s findings, which were reported in Radiology, the method can reduce the average delay from eleven days to less than three days.
For the study, British researchers teamed up with medical staff from London hospitals to get access to a dataset composed of half a million anonymized adult chest radiographs (as X-rays). Using these images, the researchers were able to develop an artificial intelligence platform for computer vision.
The new platform can recognize radiological abnormalities in the X-rays in real-time, and based on its determinations it can suggesting how quickly these exams should be examined by a radiologist.
In developing the platform, the team constructed and tested a Natural Language Processing algorithm that can read a radiological report, understand the findings described by the radiologist and determine the priority level of the exam.
The system was ‘taught’ by applying the algorithm to the historical exams, which involiced processing a large volume of data for the artificial intelligence system to understand those visual patterns in X-rays that were predictive of medical importance. With the test runs, the artificial intelligence was shown to have a 73 percent reliability.
In a synopsis of the study, lead researcher Professor Giovanni Montana, who has expertise in data science, noted that “Artificial intelligence led reporting of imaging could be a valuable tool to improve department workflow and workforce efficiency. The increasing clinical demands on radiology departments worldwide has challenged current service delivery models, particularly in publicly-funded healthcare systems. 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."
The research paves the way for other alternative models of care, such as computer vision algorithms, that can facilitate the identification of abnormal images.