[2312.02959] Detecting algorithmic bias in medical AI-models


Obtain a PDF of the paper titled Detecting algorithmic bias in medical AI-models, by Jeffrey Smith and three different authors

Obtain PDF
HTML (experimental)

Summary:With the rising prevalence of machine studying and synthetic intelligence-based medical resolution help methods, it’s equally necessary to make sure that these methods present affected person outcomes in a good and equitable style. This paper presents an revolutionary framework for detecting areas of algorithmic bias in medical-AI resolution help methods. Our method effectively identifies potential biases in medical-AI fashions, particularly within the context of sepsis prediction, by using the Classification and Regression Timber (CART) algorithm. We confirm our methodology by conducting a collection of artificial knowledge experiments, showcasing its skill to estimate areas of bias in managed settings exactly. The effectiveness of the idea is additional validated by experiments utilizing digital medical information from Grady Memorial Hospital in Atlanta, Georgia. These checks display the sensible implementation of our technique in a medical surroundings, the place it might perform as a significant instrument for guaranteeing equity and fairness in AI-based medical choices.

Submission historical past

From: Jeffrey Smith Jr. [view email]
Tue, 5 Dec 2023 18:47:34 UTC (8,698 KB)
Wed, 6 Dec 2023 20:57:39 UTC (8,698 KB)
Wed, 28 Feb 2024 17:40:32 UTC (8,843 KB)
Thu, 29 Feb 2024 13:30:59 UTC (8,843 KB)

Supply hyperlink


Please enter your comment!
Please enter your name here