ObjectivesLung ultrasound (LUS) is a portable, low-cost respiratory imaging tool but is challenged by user dependence and lack of diagnostic specificity. It is unknown whether the advantages of LUS implementation could be paired with deep learning (DL) techniques to match or exceed human-level, diagnostic specificity among similar appearing, pathological LUS images.DesignA convolutional neural network (CNN) was trained on LUS images with B lines of different aetiologies. CNN diagnostic performance, as validated using a 10% data holdback set, was compared with surveyed LUS-competent physicians.SettingTwo tertiary Canadian hospitals.Participants612 LUS videos (121 381 frames) of B lines from 243 distinct patients with either (1) COVID-19 (COVID), non-COVID acute respiratory distress syndrome (NCOVID) or (3) hydrostatic pulmonary edema (HPE).ResultsThe trained CNN performance on the independent dataset showed an ability to discriminate between COVID (area under the receiver operating characteristic curve (AUC) 1.0), NCOVID (AUC 0.934) and HPE (AUC 1.0) pathologies. This was significantly better than physician ability (AUCs of 0.697, 0.704, 0.967 for the COVID, NCOVID and HPE classes, respectively), p<0.01.ConclusionsA DL model can distinguish similar appearing LUS pathology, including COVID-19, that cannot be distinguished by humans. The performance gap between humans and the model suggests that subvisible biomarkers within ultrasound images could exist and multicentre research is merited.
ObjectivesLung ultrasound (LUS) is a portable, low cost respiratory imaging tool but is challenged by user dependence and lack of diagnostic specificity. It is unknown whether the advantages of LUS implementation could be paired with deep learning techniques to match or exceed human-level, diagnostic specificity among similar appearing, pathological LUS images.DesignA convolutional neural network was trained on LUS images with B lines of different etiologies. CNN diagnostic performance, as validated using a 10% data holdback set was compared to surveyed LUS-competent physicians.SettingTwo tertiary Canadian hospitals.Participants600 LUS videos (121,381 frames) of B lines from 243 distinct patients with either 1) COVID-19, Non-COVID acute respiratory distress syndrome (NCOVID) and 3) Hydrostatic pulmonary edema (HPE).ResultsThe trained CNN performance on the independent dataset showed an ability to discriminate between COVID (AUC 1.0), NCOVID (AUC 0.934) and HPE (AUC 1.0) pathologies. This was significantly better than physician ability (AUCs of 0.697, 0.704, 0.967 for the COVID, NCOVID and HPE classes, respectively), p < 0.01.ConclusionsA deep learning model can distinguish similar appearing LUS pathology, including COVID-19, that cannot be distinguished by humans. The performance gap between humans and the model suggests that subvisible biomarkers within ultrasound images could exist and multi-center research is merited.
Daily, fine-scale spatially explicit wildland fire occurrence prediction (FOP) models can inform fire management decisions. Many different data-driven modelling methods have been used for FOP. Several studies use multiple modelling methods to develop a set of candidate models for the same region, which are then compared against one another to choose a final model. We demonstrate that the methodologies often used for evaluating and comparing FOP models may lead to selecting a model that is ineffective for operational use. With an emphasis on spatially and temporally explicit FOP modelling for daily fire management operations, we outline and discuss several guidelines for evaluating and comparing data-driven FOP models, including choosing a testing dataset, choosing metrics for model evaluation, using temporal and spatial visualisations to assess model performance, recognising the variability in performance metrics, and collaborating with end users to ensure models meet their operational needs. A case study for human-caused FOP in a provincial fire control zone in the Lac La Biche region of Alberta, Canada, using data from 1996 to 2016 demonstrates the importance of following the suggested guidelines. Our findings indicate that many machine learning FOP models in the historical literature are not well suited for fire management operations.
Key message We used clustering to construct fuel classes from fuel inventory data based on three stand attributes relevant to crown fire behaviour: surface fuel load (SFL), canopy base height (CBH) and canopy bulk density (CBD). Resulting fuel classes explained more of the stand-to-stand variability in predicted crown fire behaviour than fuel types of the Canadian Forest Fire Behaviour Prediction (FBP) System. Context Wildfire behaviour is partly determined by stand structure and composition. Fuel characterization is essential for predicting fire behaviour and managing vegetation. Currently, categorical fuel types based on associations with major forested or open vegetated landcovers are used nationally in Canada for fire research and management applications. Aim To provide an alternative description of selected forest fuels in Alberta, Canada, using direct classification in which fuel categories are constructed from data using analytical methods. Methods Fuel inventory data for 476 stands were used to construct fuel classes with clustering. Potential crown fire behaviour was modelled for resulting fuel class clusters (FCCs) and FCCs were compared with assigned FBP System fuel types. Tree-based modelling was used to identify stand characteristics most influential on FCC membership. Fuel treatment effects on FCC and modelled crown fire behaviour were explored for the FCC most susceptible to crown fire. Results Four FCCs were identified: Red (low SFL, low CBH, low CBD); Green (high SFL, low-moderate CBH, low CBD); Blue (low SFL, high CBH, low-moderate CBD); and Black (low SFL, moderate CBH, high CBD). Stand density of live conifers and FBP System fuel type were the most important variables influencing FCC membership; however, FCCs did not align directly with assigned FBP System fuel types. Fuel reduction treatments in the Black FCC were effective at shifting the stand to a less flammable FCC. Conclusion FCCs explained more of the stand-to-stand variability in predicted crown fire behaviour than assigned FBP System fuel types, which suggests FCCs could be used to improve fire behaviour predictions and aid fire managers in prioritizing areas for fuel treatments. Future technological and remote sensing advances could enable mapping FCCs across large regions.
Wildland fire occurrence prediction (FOP) modelling supports fire management decisions, such as suppression resource pre-positioning and the routeing of detection patrols. Common empirical modelling methods for FOP include both model-based (statistical modelling) and algorithmic-based (machine learning) approaches. However, it was recently shown that many machine learning models in FOP literature are not suitable for fire management operations because of overprediction if not properly calibrated to output true probabilities. We present methods for properly calibrating statistical and machine learning models for fine-scale, spatially explicit daily FOP followed by a case-study comparison of human-caused FOP modelling in the Lac La Biche region of Alberta, Canada, using data from 1996 to 2016. Calibrated bagged classification trees, random forests, neural networks, logistic regression models and logistic generalised additive models (GAMs) are compared in order to assess the pros and cons of these approaches when properly calibrated. Results suggest that logistic GAMs can have similar performance to machine learning models for FOP. Hence, we advocate that the pros and cons of different modelling approaches should be discussed with fire management practitioners when determining which models to use operationally because statistical methods are commonly viewed as more interpretable than machine learning methods.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.