Date of Award

2023

Document Type

Open Access Thesis

Degree Name

MS in Physician Assistant Studies (PA)

Department

Physician Assistant Studies

Abstract

Background: This systematic review examines how the use of artificial intelligence compares to conventional methods in the early detection and accuracy of diagnosing lung and breast cancer. Methods: A comprehensive systematic review was conducted using Google Scholar, ScienceDirect, MDPI Journals, PubMed, JAMA Network, The Lancet Digital Health, Frontiers, Journal of Patient Safety, Thorax, NPJ Breast Cancer, BMC, and Nature Medicine. The inclusion criteria were artificial intelligence models or components of artificial intelligence detecting or classifying breast or lung cancer and articles published within the last five years. The study excluded articles that did not include either breast or lung cancer. The results were compiled into a table based on the key data gathered, such as accuracy, specificity, sensitivity, or P-value. Results: A total of 15 studies were reviewed, eight of the articles were on breast cancer, and seven of the articles were on lung cancer. Each study showed an improvement in their results of accuracy, specificity, and sensitivity. One article gave a confidence score of 63% and two other articles gave a significant P-value < 0.05.

Discussion: The limitations or bias include a small sample size of radiologists, and possible bias due to authors following their own model. Overall, the articles showed improvement, but many articles utilized subsets of artificial intelligence, which mainly consisted of machine learning or deep learning models. Therefore, more research needs to be done, but the current research available has shown a promising future of the use of artificial intelligence.

Identifier

SC 11.PAS.2023.Luu.B

Included in

Oncology Commons

COinS