Julia Wagner | Searcharxiv

arXiv2026

A User-Centric Analysis of Explainability in AI-Based Medical Image Diagnosis

In recent years, AI systems in the medical domain have advanced significantly. However, despite outperforming humans, they are rarely used in practice since it is often not clear how they make their decisions. Optimal explanation and visualization of the decision process are often lacking. Therefore, we conducted a comparative user-centric analysis of the latest state-of-the-art textual, visual and multimodal explainable artificial intelligence (XAI) methods for medical image diagnosis. Our survey of 33 physicians showed that 88% agree that it is important that AI explains the diagnosis -- 64% even strongly agree. A combination of bounding box and report is rated better than the other tested XAI methods in the evaluated aspects understandability, completeness, speed, and applicability. We even tested the potential negative impact of false AI-based medical image diagnoses and found that 50% of the participants trusted false AI diagnoses over all tested XAI methods.

cs.HC↗

arXiv2020

Self-Tuning State Estimation for Adaptive Truss Structures Using Strain Gauges and Camera-Based Position Measurements

In the context of control of smart structures, we present an approach for state estimation of adaptive buildings with active load-bearing elements. For obtaining information on structural deformation, a system composed of a digital camera and optical emitters affixed to selected nodal points is introduced as a complement to conventional strain gauge sensors. Sensor fusion for this novel combination of sensors is carried out using a Kalman filter that operates on a reduced-order structure model obtained by modal analysis. Signal delay caused by image processing is compensated for by an out-of-sequence measurement update which provides for a flexible and modular estimation algorithm. Since the camera system is very precise, a self-tuning algorithm that adjusts model along with observer parameters is introduced to reduce discrepancy between system dynamic model and actual structural behavior. We further employ optimal sensor placement to limit the number of sensors to be placed on a given structure and examine the impact on estimation accuracy. A laboratory scale model of an adaptive high-rise with actuated columns and diagonal bracings is used for experimental demonstration of the proposed estimation scheme.

eess.SP↗