Andy Li
Pronouns: He/Him
Research Mentor(s): Deanna Marriott
Research Mentor School/College/Department: / Nursing
Program:
Authors:
Session: Session 5: 2:40 pm – 3:30 pm
Poster: 6
Abstract
The process of analyzing clinical papers and extracting their data sets is a rather prolonged and redundant task. The aim of the project is to construct an Artificial Intelligence (AI) that effectively analyzes clinical papers and extracts their data sets. Clinical Papers were selected on the following criteria: the paper was published after 2005; the study contained a minimum of 20 individuals who had diabetes mellitus and underwent Laparoscopic Sleeve Gastrectomy (LSG) or Roux-en-Y Gastric Bypass (RYGB). Extraction of the data sets from the Clinical papers were manually conducted and by way of utilizing Microsoft co-pilot’s AI. The primary outcome of interest from the clinical papers was the remission of diabetes mellitus. The extractions performed by the AI were compared to human extractors for 14 data set categories.The comparison of the extraction indicates that the AI performed well on data that were qualitatively measured whereas the AI did not perform as well on quantitative data sets. Data such as inclusion and exclusion criteria were consistently extracted by the AI. Numerical data such as age was not consistently extracted by the AI. Values such as standard deviation, and median of the data would sometimes be falsely created by the AI. Based on the results of the project, AI’s can be reliably used for extracting data that does not require further extrapolations. Conversely, data that requires further extrapolation is more challenging for the AI. Further research will be needed to best determine the effectiveness of AI in data extraction. The development of AI will lead to the eventual goal of creating a shortcut in extracting data, thus, making a time consuming task a quick one
