Search Results for « CoALA »

Abstract Motivation Machine learning (ML) models have become increasingly complex, often functioning as black boxes that limit our understanding of the features contributing to ML predictions. Common explainable AI (XAI) methods such as SHapley Additive exPlanations (SHAP) focus on feature importance but fall short in identifying interactions among features and informing targeted, personalized interventions. Counterfactuals are hypothetical events where specific variables are altered to cause a change in outcome. These causal statements can be applied to AI models to identify actionable interventions for different subjects in a population. Results We propose the framework Counterfactual Optimization for Actionable interpretabiLity in AI (COALA). COALA interprets models by identifying optimal counterfactuals for each subject, which are defined as actionable changes that lead to the most positive change in predicted outcome. When applied to a gradient boosted tree model trained on the National Health and Nutrition Examination Survey (NHANES) dataset, COALA identifies different profiles of optimal counterfactuals across subjects. Features that remain constrained were able to predict the optimal counterfactual changes for a subject at 85.4%, revealing specific features that drive what a subject’s optimal counterfactual is. Availability and Implementation Code for COALA implementation, synthetic data, models trained on synthetic data, and code to replicate results and figures are available at https://github.com/brt-solo/COALA . The NHANES 2017–2018 dataset is publicly available from the National Center for Health Statistics (NCHS). The Framingham Heart Study dataset used in this study is a publicly available, Framingham-derived dataset distributed through the Massachusetts Institute of Technology OpenCourseWare (MIT OCW) repository.

Abstract Background/Introduction Atrial fibrillation (AF) can be asymptomatic and it is associated with stroke and heart failure which can present as first symptoms [1,2]. There is increasing evidence of benefits of screening for AF to prevent ischemic stroke by initiating anticoagulation therapy in affected individuals [3]. There are several devices used for screening of AF [4] and one of them is Coala Heart Monitor (CHM), which is a handheld device, connected by blue-tooth to a smartphone for registration of the rhythm from the chest and the thumbs. The aim with this this study was to validate the accuracy of CHM for detecting atrial tachyarrhythmia (ATA) (AF, atrial flutter or atrial tachycardia) compared to 12-lead-ECG. Methods Eligible participants were ≥18 years of age, planned for a 12-lead-ECG, with a history of ATA. Exclusion criteria were; pacemaker or implantable cardioverter-defibrillator (ICD)-treatment, ongoing arrhythmia that was not ATA or if the patient was unwilling or unable to provide informed consent. One hundred and eleven 12-lead-ECG were collected and immediately after a 30 + 30 second thumb-and chest-registration from CHM where collected. 12-lead-ECG independently interpreted by two electrophysiologist was used as the reference standard (100% coherence). In one patient neither chest nor thumb-registration could be interpreted due to bad quality, and this was not included in the analysis. Eight thumb-registrations and three chest-registrations could not be interpreted and in these cases the other registration could be interpreted. In one case there was disagreement between interpreter one and two, and in this case a third interpreter was involved. Results Mean age was 65.2±12.1 years and 28.8% were female. Prevalence of ATA was 36.4%. The sensitivity for ATA was 100% (95% CI, 96.3% - 100%) and the specificity was 100% (95% CI, 96.3% - 100%). Conclusion CHM provides excellent sensitivity and specificity for ATA compared to 12-lead-ECG. Funding Acknowledgement Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Swedish Research Council

The occurrence of congestion has an extremely deleterious impact on the performance of Wireless Sensor Networks (WSNs). This article presents a novel protocol, named COALA (COngestion ALleviation and Avoidance), which aims to act both proactively, in order to avoid the creation of congestion in WSNs, and reactively, so as to mitigate the diffusion of upcoming congestion through alternative path routing. Its operation is based on the utilization of an accumulative cost function, which considers both static and dynamic metrics in order to send data through the paths that are less probable to be congested. COALA is validated through simulation tests, which exhibit its ability to achieve remarkable reduction of loss ratios, transmission delays and energy dissipation. Moreover, the appropriate adjustment of the weighting of the accumulative cost function enables the algorithm to adapt to the performance criteria of individual case scenarios.

This article describes a computational system for the linguistic analysis of language acquisition data (COALA). The system is a combined AI and database tool which allows the user to form highly complex queries about morphosyntactic and semantic structures contained in large sets of data. COALA identifies those sentences that meet the linguistic criteria defined by the user. It allows the user to freely define such linguistic contexts and to step through the sentences identified by the system. COALA then rapidly displays those sentences in their original discourse context. Additionally, COALA can perform statistical analyses in response to structural linguistic queries. This article contains (1) a discussion of the computational approach taken in the design of COALA, (2) a description of the functionality of the system, and (3) a reflection on the validity of the analytical categories contained therein.