February 2nd, 2023 – 16:00 CET
Registration link: https://shre.ink/WideHealthSeminarsSeries
Registration is free. Please share with your networks.
Date: February 2nd, 2023, 16:00 CET (15:00 in Portugal)
Speaker: Bojana Velichkovska
Title: Vital Signs as source of racial bias
Abstract: The exponential growth of artificial intelligence has increased its application in tackling complex clinical challenges and will likely progress into creating game-changing approaches that will assist healthcare processes all around the world. With that impact scale, it is important that the built algorithms be robust, reliable, and unbiased. However, there have been numerous studies reporting indisputable evidence of bias in healthcare providers’ attitudes, medical ML applications, and even in medical datasets used for research. With all of the above being a fact, we wanted to investigate whether bias can be introduced from sources which are considered neutral, more specifically vital signs. In this talk, I will give an overview of the challenges of applied machine learning in a medical setting, with a major focus on the presence of racial bias in actively used machine learning algorithms. Then, I will present our approach in predicting the self-declared ethno-race of patients admitted in the ICU based on their vital signs. Namely, I will discuss the definition of the problem, the methodology used in addressing it, and the results which were obtained. Finally, I will discuss the results and present the conclusions we obtained.
Short bio: Bojana Velichkovska is a Teaching and Research Assistant at the Faculty of Electrical Engineering and Information Technologies (FEEIT), UKIM. She graduated from FEEIT in 2018 after defending her thesis titled “Object Recognition in a 3D Point Cloud based on Machine Learning”. She obtained her master’s degree in Computer Networks – Internet of Things in 2020, after defending her thesis titled “Pneumothorax Identification in Chest X-rays”. Bojana is currently pursuing a PhD in Electrical Engineering and Information Technologies. Her research currently focuses on the application of machine learning algorithms in investigations of ethno-racial bias in ICU data.
The EU-funded WideHealth project aims to conduct research on pervasive eHealth and establish a sustainable network of research and dissemination across Europe.