DRUGCAM®’s performance in real-life production by simulation
DRUGCAM® is a new approach to control the chemotherapy preparations with an intelligent video system designed to assist the entire drug preparation process. This real-time artificial intelligence is able to control 100% of our production. We first aimed to estimate DRUGCAM®‘s performance in real-life production and to compare it with the double human control. Furthermore, factors influencing the performance of both checking procedures were observed.
During 30 days, between 11:30 and 12:30 am, we controlled 20 different volumes contained in syringes with a method of simulation. Those controls have been conducted in real production conditions both by human visual inspection and by automated video control. Working conditions of the team have been observed and a set of information has been collected: the syringe’s model, the checking hour, the volume of product and the disturbances. A statistical analysis has been conducted to interpret results.
With 24 errors throughout the 600 volume controls, the error rate for the visual human control is 4%.
7 checked volumes were superior to the expected volume (overdosing) and 17 were inferior (underdosing). The error rate for DRUGCAM® is 0.17%. Only one error has been observed with a deviation of -1 mL. Among the disturbance factors, the type of syringe used is responsible of errors: 13 errors have been noticed with the 1 mL syringe and 8 errors with the 10 mL syringe which represent higher error rates (11% and 7%) than with the other syringes.
The « permanent » staff members of the chemotherapy preparation units present an error rate of 5.3%, more important than the « non-permanent » ones (1.8%).
Our studies justify the superiority of the DRUGCAM® system toward double human control. We witnessed the fact that the double human control could possibly be disturbed by external factors whereas DRUGCAM® is not. Using DRUGCAM® is to be considered to establish preventive measures and reduce tasks disturbance factors thanks to video analysis.