340880 research outputs found
Sort by
Public attitudes towards police use of AI-driven face recognition technology
No full textThis study examined public attitudes toward police use of AI-driven facial recognition technology (FRT) for face detection, identification, verification, tracking, kinship verification, and masked perpetrator recognition. In a scenario-based survey with N = 507 participants, we investigated how perceptions of trust, fairness, accuracy, and support for specific FRT applications were influenced by general AI knowledge, trust in law enforcement, and application type. Masked face identification and kinship verification consistently received the lowest trust, fairness, accuracy, and support ratings, while face verification gathered the highest levels of acceptance. Contrary to expectations, deeper general AI knowledge was linked with decreased trust and support for FRTs in policing contexts. This suggests that technological literacy enhanced critical awareness of algorithmic limitations and ethical concerns. Participants expressed significant concerns about algorithmic bias, privacy implications, and surveillance capabilities. Trust in law enforcement emerged as the strongest predictor of FRT acceptance, indicating that acceptance of AI is embedded in broader socio-political relationships rather than determined by technological concerns alone. These findings contribute to our understanding of the social embeddedness of AI technologies and emphasize the need for governance frameworks that address not only technical performance but also institutional accountability and transparency in algorithmic systems deployed within law enforcement contexts
Connecting online criminal behavior with machine learning:Using authorship attribution to analyze and link potential online traffickers
No full text
Spatiotemporal determinants of stretch-activated channel-induced re-entry in ventricular tissue:An in-silico study
No full textStretch-activated ion channels (SACs) mediate mechano-electric feedback in cardiomyocytes by coupling mechanical and electrical activity. While SACs activation can induce pro-arrhythmic effects at the cellular level, its impact on tissue-level arrhythmias remains poorly understood. Particularly unclear are the specific stretch characteristics that promote arrhythmogenesis, a knowledge gap largely due to limited experimental control over these parameters. We investigated how SACs activation affects excitation-wave propagation in simulated ventricular tissue and identified parameters promoting arrhythmias, with relevance to commotio cordis, in which a chest impact can trigger ventricular arrhythmias and sudden cardiac death. Our approach employed a validated human ventricular action potential model incorporating three types of SACs (non-selective, potassium-selective, and calcium-selective) applied to a two-dimensional tissue framework. Through systematic multiparameter analysis, we examined the effects of stretch stimulus parameters (amplitude, duration, timing), spatial characteristics (area, location, gradient), and tissue properties (size, conduction velocity). Our simulations revealed that re-entry arises from interactions between stretch-induced depolarization waves and repolarization tails of preceding excitation waves. Acute supra-threshold stretch (i.e., stretch able to trigger an action potential) initiated re-entries with increased likelihood when path lengths were longer and when stretched regions were closer to non-conducting borders oriented perpendicular to the line of block. Furthermore, stretch amplitude gradients attenuated pro-arrhythmic effects, while sustained sub-threshold stretch either reduced conduction velocity or caused conduction block. This in silico analysis demonstrates that tissue-level proarrhythmic effects of stretch depend on complex interactions between stretch stimulus characteristics, spatial parameters, and tissue properties
Explainable AI for automatic heart disease diagnosis using 3DFMMecg features:A novel ECG-based approach
No full textThe electrocardiogram (ECG), a gold standard in cardiac diagnostics, is increasingly combined with Artificial Intelligence (AI) methods to enhance its clinical utility. However, many recent studies have prioritised performance over clinical interpretability by focusing on Deep Learning (DL) techniques, which offer limited explainability since they do not directly correlate with clinical features. This lack of transparency causes mistrust among physicians and hinders adoption in daily clinical practice. We developed novel, self-explainable features from the 3DFMMecg model parametrisation, enabling highly accurate and clinically interpretable ML classifiers for cardiovascular pathology diagnosis from 12-lead ECG signals. We evaluated our approach on PTB-XL+, a widely used dataset of annotated ECG recordings. Our framework outperforms existing feature-based methods in four out of six classification tasks, achieving macro-AUC between 0.88 and 0.95 and weighted macro-AUC between 0.90 and 0.95, comparable to and in some tasks surpassing DL approaches. We further show that the model maintains high diagnostic accuracy when using only three of the standard twelve ECG leads, with less than 6% loss in performance, enabling deployment in mobile, wearable, or resource-constrained environments. Feature importance analyses using SHapley Additive exPlanations (SHAP) confirm strong alignment between model predictions and established clinical markers, such as QRS width and T-wave amplitude parameters. These results underscore the potential of 3DFMMecg-based pipelines toward reliable, transparent, and accessible ECG-based diagnostic systems
Patient-Centred Explainability in IVF Outcome Prediction
No full textThis paper evaluates the user interface of an in vitro fertility (IVF) outcome prediction tool, focussing on its understandability for patients or potential patients. We analyse four years of anonymous patient feedback, followed by a user survey and interviews to quantify trust and understandability. Results highlight a lay user's need for prediction model explainability beyond the model feature space. We identify user concerns about data shifts and model exclusions that impact trust. The results call attention to the shortcomings of current practices in explainable AI research and design and the need for explainability beyond model feature space and epistemic assumptions, particularly in high-stakes healthcare contexts where users gather extensive information and develop complex mental models. To address these challenges, we propose a dialogue-based interface and explore user expectations for personalised explanations
Let's (not) escalate this! Leadership and communication in a group contest
No full textEconomic and social situations where groups have to compete are ubiquitous. Such group contests create both a coordination problem within and between groups. Introducing leaders may help to mitigate these coordination problems, but little is known about the effect of leadership in group contests. In a group contest experiment, we compare two types of leadership-leading-by-example and transactional leadership-and also investigate the effect of communication between leaders under both leadership styles. We find that the introduction of leaders mostly increases contest investment. Transactional leaders increase followers' investment through the allocation of a relatively larger share of the prize to followers who have invested more. Communication between leaders decreases contest investments when there is leading-by-example but not when there is transactional leadership. Overall, leaders do not mitigate the over-investment problem in group contests
Nuances in the memory undermining effects of EMDR and imagery rescripting
No full textWe reviewed the evidence on the memory undermining ef(EMDR) and imagery rescripting. Both therapies appear to undermine memory quality by making memories less vivid and emotionally-negative. Also, while eye movements used in EMDR seem to increase spontaneous false memories, they do not increase the susceptibility to suggestion. Inconsistent findings have emerged on the effects of imagery rescripting on false memory generation. Furthermore, a substantial number of clinicians who use EMDR strongly believe in the notion of repressed memory and EMDR has been associated with the occurrence of recovered memories. The belief in repressed memory might encourage suggestive therapeutic techniques, thereby increasing the risk of false memory creation. Overall, nuance is required on potential memory undermining effects of EMDR and imagery rescripting