Utilizing synthetic intelligence (AI) to display screen most cancers sufferers for medical trials is efficient and environment friendly, in accordance with analysis printed within the Journal of the Nationwide Cancer Institute.
Researchers discovered that AI screening typically had excessive accuracy, sensitivity, and specificity. As well as, AI screening proved extra environment friendly than handbook screening.
The researchers reviewed 10 research of AI use within the medical trial enrollment course of. The research included 19 datasets and greater than 50,000 sufferers. The pattern sizes ranged from 96 to 48,124, and the research have been carried out in the US, Europe, and Australia.
Proceed Studying
Printed AI algorithms have been utilized in 4 research. For five research, researchers developed their very own algorithms. For 1 examine, researchers used an algorithm created for a earlier undertaking.
Throughout 13 evaluable datasets, the accuracy of AI algorithms ranged from 62.8% to 100%. Throughout 17 datasets, the sensitivity of the algorithms ranged from 46.7% to 100%.
Throughout 16 datasets, the specificity of the algorithms ranged from 59.2% to 100%. Nevertheless, all however 1 of the evaluable datasets had accuracy, sensitivity and specificity values larger than 80%.
The optimistic predictive worth ranged from 12.6% to 100% throughout 17 datasets. The unfavorable predictive worth ranged from 87.2% to 100% throughout 16 datasets.
Eight datasets encompassing a complete of 40,447 sufferers had enough knowledge for a meta-analysis. The abstract sensitivity of the AI algorithms was 90.5%, and the abstract specificity was 99.3%.
Knowledge on time financial savings have been reported for 4 research, and all 4 confirmed that AI screening was extra environment friendly than handbook screening. One examine confirmed a 78% time financial savings with AI. One other examine confirmed that AI carried out screening in 1.15 hours that may have taken 263 days if executed manually.
“AI appears to have comparable, if not superior, performance to manual screening for oncology clinical trials enrollment,” the researchers wrote. “As well, AI is efficient, requiring less time and resources to enroll patients; therefore, AI should be further investigated and implemented for this application.”
Reference
Chow R, Midroni J, Kaur J, et al. Use of synthetic intelligence for most cancers medical trial enrolment: A scientific overview and meta-analysis. J Natl Cancer Inst. Printed on-line January 23, 2023. doi:10.1093/jnci/djad013
0 Comments