On (assessing) the fairness of risk score models
Authors:
Eike Petersen,
Melanie Ganz,
Sune Hannibal Holm,
Aasa Feragen
Abstract:
Recent work on algorithmic fairness has largely focused on the fairness of discrete decisions, or classifications. While such decisions are often based on risk score models, the fairness of the risk models themselves has received considerably less attention. Risk models are of interest for a number of reasons, including the fact that they communicate uncertainty about the potential outcomes to use…
▽ More
Recent work on algorithmic fairness has largely focused on the fairness of discrete decisions, or classifications. While such decisions are often based on risk score models, the fairness of the risk models themselves has received considerably less attention. Risk models are of interest for a number of reasons, including the fact that they communicate uncertainty about the potential outcomes to users, thus representing a way to enable meaningful human oversight. Here, we address fairness desiderata for risk score models. We identify the provision of similar epistemic value to different groups as a key desideratum for risk score fairness. Further, we address how to assess the fairness of risk score models quantitatively, including a discussion of metric choices and meaningful statistical comparisons between groups. In this context, we also introduce a novel calibration error metric that is less sample size-biased than previously proposed metrics, enabling meaningful comparisons between groups of different sizes. We illustrate our methodology - which is widely applicable in many other settings - in two case studies, one in recidivism risk prediction, and one in risk of major depressive disorder (MDD) prediction.
△ Less
Submitted 22 February, 2023; v1 submitted 17 February, 2023;
originally announced February 2023.
Sorting of particles suspended in whole blood
Authors:
Stefan H. Holm,
Zunmin Zhang,
Jason P. Beech,
Gerhard Gompper,
Dmitry A. Fedosov,
Jonas O. Tegenfeldt
Abstract:
An important step in diagnostics is the isolation of specific cells and microorganisms of interest from blood. Since such bioparticles are often present at very low concentrations, throughput needs to be as high as possible. In addition, to ensure simplicity, a minimum of sample preparation is important. Therefore, sorting schemes that function for whole blood are highly desirable. Deterministic l…
▽ More
An important step in diagnostics is the isolation of specific cells and microorganisms of interest from blood. Since such bioparticles are often present at very low concentrations, throughput needs to be as high as possible. In addition, to ensure simplicity, a minimum of sample preparation is important. Therefore, sorting schemes that function for whole blood are highly desirable. Deterministic lateral displacement (DLD) has proven to be very precise and versatile in terms of a wide range of sorting parameters. To better understand how DLD performs for blood as the hematocrit increases, we have performed measurements and simulations for spherical particles in the micrometer range moving through DLD arrays for different flow velocities and hematocrits ranging from pure buffer to whole blood. We find that the separation function of the DLD array is sustained, even though blood cells introduce a shift in the trajectories and a significant dispersion for particles that are close to the critical size in the device. Simulations qualitatively replicate our experimental observations and help us identify fundamental mechanisms for the effect of hematocrit on the performance of the DLD device.
△ Less
Submitted 11 January, 2019;
originally announced January 2019.
