Smart chest X-ray worklist prioritization using artificial intelligence: a clinical workflow simulation

Abstract

Objective: The aim is to evaluate whether smart worklist prioritization by artificial intelligence (AI) can optimize the radiology workflow and reduce report turnaround times (RTATs) for critical findings in chest radiographs (CXRs). Furthermore, we investigate a method to counteract the effect of false negative predictions by AI-resulting in an extremely and dangerously long RTAT, as CXRs are sorted to the end of the worklist.

Methods: We developed a simulation framework that models the current workflow at a university hospital by incorporating hospital-specific CXR generation rates and reporting rates and pathology distribution. Using this, we simulated the standard worklist processing "first-in, first-out" (FIFO) and compared it with a worklist prioritization based on urgency. Examination prioritization was performed by the AI, classifying eight different pathological findings ranked in descending order of urgency: pneumothorax, pleural effusion, infiltrate, congestion, atelectasis, cardiomegaly, mass, and foreign object. Furthermore, we introduced an upper limit for the maximum waiting time, after which the highest urgency is assigned to the examination.

Results: The average RTAT for all critical findings was significantly reduced in all prioritization simulations compared to the FIFO simulation (e.g., pneumothorax: 35.6 min vs. 80.1 min; p < 0.0001), while the maximum RTAT for most findings increased at the same time (e.g., pneumothorax: 1293 min vs 890 min; p < 0.0001). Our "upper limit" substantially reduced the maximum RTAT in all classes (e.g., pneumothorax: 979 min vs. 1293 min/1178 min; p < 0.0001).

Conclusion: Our simulations demonstrate that smart worklist prioritization by AI can reduce the average RTAT for critical findings in CXRs while maintaining a small maximum RTAT as FIFO.

Key points: • Development of a realistic clinical workflow simulator based on empirical data from a hospital allowed precise assessment of smart worklist prioritization using artificial intelligence. • Employing a smart worklist prioritization without a threshold for maximum waiting time runs the risk of false negative predictions of the artificial intelligence greatly increasing the report turnaround time. • Use of a state-of-the-art convolution neural network can reduce the average report turnaround time almost to the upper limit of a perfect classification algorithm (e.g., pneumothorax: 35.6 min vs. 30.4 min).

Keywords: Artificial intelligence; Radiography; Waiting lists; Workflow.

Fig. 1

Receiver operating characteristics of the artificial intelligence algorithm for all eight different findings. We show the receiver operation curve and calculated the area under the receiver operation curve (AUC)

Fig. 2

Workflow simulation. A chest X-ray (CXR) machine is constantly generating CXRs. To each CXR, zero or up to eight findings are assigned. CXRs are either sorted into the worklist chronologically (first-in, first-out; FIFO) or according to the urgency based on the prediction by artificial intelligence (PRIO). Finally, worklists are processed by a virtual radiologist

Fig. 3

Discrete distribution of chest X-ray (CXR) generation speed. The X-axis shows the day time in 24-h format and the Y-axis shows the calculated time deltas. The histogram in X- and Y-direction is shown in green

Fig. 4

Discrete distribution of chest X-ray (CXR) reporting times by radiologists. The X-axis shows the day time in 24-h format and the Y-axis shows the calculated time deltas between two CXR reports. The histogram in X- and Y-direction is shown in green

Fig. 5

Optimal operation point simulation for the artificial intelligence algorithm. To find the optimal operation point for reducing the average report turnaround time (RTAT) for critical findings, we run multiple simulations with different false positive rates between zero and one

Fig. 6

Report turnaround times (RTATs) for all eight pathological findings as well as for normal examinations on the basis of four different simulations: FIFO (first-in, first-out; green), Prio-lowFNR (false negative rate; yellow), Prio-lowFPR (false positive rate; purple), and Prio-MAXwaiting (maximum; red) with a maximum waiting time (light purple). The green triangles mark the average RTAT, while the vertical lines mark the median RTAT. On the right side, the maximum RTAT for each simulation and finding is shown