Weight-of-evidence through shrinkage and spline binning for interpretable nonlinear classification

2021 ◽  
pp. 108160
Author(s):  
Jakob Raymaekers ◽  
Wouter Verbeke ◽  
Tim Verdonck
2010 ◽  
Vol 5 (1) ◽  
pp. 104
Author(s):  
Daniel S Menees ◽  
Eric R Bates ◽  
◽  

Coronary artery disease (CAD) affects millions of US citizens. As the population ages, an increasing number of people with CAD are undergoing non-cardiac surgery and face significant peri-operative cardiac morbidity and mortality. Risk-prediction models can be used to help identify those patients at increased risk of peri-operative cardiovascular complications. Risk-reduction strategies utilising pharmacotherapy with beta blockade and statins have shown the most promise. Importantly, the benefit of prophylactic coronary revascularisation has not been demonstrated. The weight of evidence suggests reserving either percutaneous or surgical revascularisation in the pre-operative setting for those patients who would otherwise meet independent revascularisation criteria.


2017 ◽  
Vol 13 (9) ◽  
pp. 6480-6488 ◽  
Author(s):  
A.D. Jeyarani ◽  
Reena Daphne ◽  
Solomon Roach

The main contribution of this paper has been to introduce nonlinear classification techniques to extract more information from the PCG signal. Especially, Artificial Neural Network classification techniques have been used to reconstruct the underlying system’s state space based on the measured PCG signal. This processing step provides a geometrical interpretation of the dynamics of the signal, whose structure can be utilized for both system characterization and classification as well as for signal processing tasks such as detection and prediction.


2001 ◽  
Vol 36 (2) ◽  
pp. 319-330 ◽  
Author(s):  
Mark Servos ◽  
Don Bennie ◽  
Kent Burnison ◽  
Philippa Cureton ◽  
Nicol Davidson ◽  
...  

Abstract A number of biological responses and multigenerational effects, mediated through the disruption of endocrine systems, have been observed in biota exposed to relatively low concentrations of environmental contaminants. These types of responses need to be considered within a weight of evidence approach in our risk assessment and risk management frameworks. However, including endocrine responses in an environmental risk assessment introduces a number of uncertainties that must be considered. A risk assessment of nonylphenol and nonylphenol polyethoxylates (NP/NPE) is used as a case study to demonstrate the sources and magnitude of some of the uncertainties associated with using endocrine disruption as an assessment endpoint. Even with this relatively well studied group of substances, there are substantial knowledge gaps which contribute to the overall uncertainties, limiting the interpretation within the risk assessment. The uncertainty of extrapolating from in vitro or biochemical responses to higher levels of organization or across species is not well understood. The endocrine system is very complex and chemicals can interact or interfere with the normal function of endocrine systems in a number of ways (e.g., receptors, hormones) which may or may not result in an adverse responses in the whole organism. Using endocrine responses can lead to different conclusions than traditional endpoints due to a variety of factors, such as differences in relative potencies of chemicals for specific endpoints (e.g., receptor binding versus chronic toxicity). The uncertainties can also be considerably larger and the desirability of using endocrine endpoints should be carefully evaluated. Endocrine disruption is a mode of action and not a functional endpoint and this needs to be considered carefully in the problem formulation stage and the interpretation of the weight of evidence.


2021 ◽  
Vol 275 ◽  
pp. 116575
Author(s):  
Annachiara Codello ◽  
Sandra L. McLellan ◽  
Peter Steinberg ◽  
Jaimie Potts ◽  
Peter Scanes ◽  
...  

2008 ◽  
Vol 54 (1) ◽  
pp. 36-46
Author(s):  
Katherine Manaras Smith ◽  
William S. Keeton ◽  
Therese M. Donovan ◽  
Brian Mitchell

Abstract We explored the role of stand-level forest structure and spatial extent of forest sampling in models of avian occurrence in northern hardwood-conifer forests for two species: black-throated blue warbler (Dendroica caerulescens) and ovenbird (Seiurus aurocapillus). We estimated site occupancy from point counts at 20 sites and characterized the forest structure at these sites at three spatial extents (0.2, 3.0, and 12.0 ha). Weight of evidence was greatest for habitat models using forest stand structure at the 12.0-ha extent and diminished only slightly at the 3.0-ha extent, a scale that was slightly larger than the average territory size of both species. Habitat models characterized at the 0.2-ha extent had low support, yet are the closest in design to those used in many of the habitat studies we reviewed. These results suggest that the role of stand-level vegetation may have been underestimated in the past, which will be of interest to land managers who use habitat models to assess the suitability of habitat for species of concern.


Sign in / Sign up

Export Citation Format

Share Document