An important target for machine learning research is obtaining unbiased results, which require addressing bias that might be present in the data as well as the methodology. This is of utmost importance in medical applications of machine learning, where trained models should be unbiased so as to result in systems that are widely applicable, reliable and fair. Since bias can sometimes be introduced through the data itself, in this paper we investigate the presence of ethnoracial bias in patients' clinical data. We focus primarily on vital signs and demographic information and classify patient ethnoraces in subsets of two from the three ethnoracial groups (African Americans, Caucasians, and Hispanics). Our results show that ethnorace can be identified in two out of three patients, setting the initial base for further investigation of the complex issue of ehtnoracial bias.
Background: racial bias has been shown to be present in clinical data, affecting patients unfairly based on their race, ethnicity and socio-economic status. This problem has the potential to be significantly exacerbated in the light of Artificial Intelligence-aided clinical decision making. We sought to investigate whether bias can be introduced from sources that are considered neutral with respect to ethnicity and race and consequently routinely used in modelling, specifically vital signs. Methods: to perform our analysis, we extracted vital signs from 49,610 admissions from a cohort of adult patients during the first 24 hours after the admission to the Intensive Care Units (ICU), derived from multi-centre eICU-CRD database and single-centre MIMIC-III database, spanning over 208 hospitals and 335 ICUs. Using heart rate, SaO2, respiratory rate, systolic, diastolic, and mean blood pressure, we develop machine learning models based on Logistic Regression and eXtreme Gradient Boosting and investigate their performance in predicting patients' self-reported race. To balance the dataset between the three ethno-races considered in our study, we use a matching cohort based on age, gender, and admission diagnosis. Findings: standard machine learning models, derived solely on six vital signs can be used to predict patients' self-reported race with AUC of 75%. Our findings hold under diverse patient populations, derived from multiple hospitals and intensive care units. We also show that oxygen saturation is a highly predictive variable, even when measured through methods other than pulse oximetry, namely arterial blood gas analysis, suggesting that addressing bias in routinely collected clinical variables will be challenging. Interpretation: our finding that machine learning models can predict self-reported race using solely vital signs creates a significant risk in clinical decision making, further exacerbating racial inequalities, with highly challenging mitigation measures. Funding: The funders had no role in the design of this study.
With the rapid expansion and widespread adoption of the Internet of Things (IoT), maintaining secure connections among active devices can be challenging. Since IoT devices are limited in power and storage, they cannot perform complex tasks, which makes them vulnerable to different types of attacks. Given the volume of data generated daily, detecting anomalous behavior can be demanding. However, machine learning (ML) algorithms have proven successful in extracting complex patterns from big data, which has led to active applications in IoT.In this paper, we perform a comprehensive analysis, including 4 ML algorithms and 3 neural networks (NNs), and propose a pipeline which analyzes the influence data reduction (loss) has on the performance of these algorithms. We use random undersampling as a data reduction technique, which simulates reduced network traffic data. The pipeline investigates several degrees of data loss. The results show that models trained on the original data distribution obtain accuracy that verges on 100%. XGBoost performs best from the classic ML algorithms. From the deep learning models, the 2-layered NN provides excellent results and has sufficient depth for practical application. On the other hand, when the models are trained on the undersampled data, there is a decrease in performance, most notably in the case of NNs. The most prominent change is seen in the 4-layered NN, where the model trained on the original dataset detects attacks with a success of 93.53%, whereas the model trained on the maximally reduced data has a success of only 39.39%.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.