Top Model Research Articles

A deep learning approach for generating intracranial pressure waveforms from extracranial signals routinely measured in the intensive care unit

Intracranial pressure (ICP) is commonly monitored to guide treatment in patients with serious brain disorders such as traumatic brain injury and stroke. Established methods to assess ICP are resource intensive and highly invasive. We hypothesized that ICP waveforms can be computed noninvasively from three extracranial physiological waveforms routinely acquired in the Intensive Care Unit (ICU): arterial blood pressure (ABP), photoplethysmography (PPG), and electrocardiography (ECG). We evaluated over 600 h of high-frequency (125 Hz) simultaneously acquired ICP, ABP, ECG, and PPG waveform data in 10 patients admitted to the ICU with critical brain disorders. The data were segmented in non-overlapping 10-s windows, and ABP, ECG, and PPG waveforms were used to train deep learning (DL) models to re-create concurrent ICP. The predictive performance of six different DL models was evaluated in single- and multi-patient iterations. The mean average error (MAE) ± SD of the best-performing models was 1.34 ± 0.59 mmHg in the single-patient and 5.10 ± 0.11 mmHg in the multi-patient analysis. Ablation analysis was conducted to compare contributions from single physiologic sources and demonstrated statistically indistinguishable performances across the top DL models for each waveform (MAE±SD 6.33 ± 0.73, 6.65 ± 0.96, and 7.30 ± 1.28 mmHg, respectively, for ECG, PPG, and ABP; p = 0.42). Results support the preliminary feasibility and accuracy of DL-enabled continuous noninvasive ICP waveform computation using extracranial physiological waveforms. With refinement and further validation, this method could represent a safer and more accessible alternative to invasive ICP, enabling assessment and treatment in low-resource settings.

Assessing and Optimizing Large Language Models on Spondyloarthritis Multi-Choice Question Answering: Protocol for Enhancement and Assessment.

Spondyloarthritis (SpA), a chronic inflammatory disorder, predominantly impacts the sacroiliac joints and spine, significantly escalating the risk of disability. SpA's complexity, as evidenced by its diverse clinical presentations and symptoms that often mimic other diseases, presents substantial challenges in its accurate diagnosis and differentiation. This complexity becomes even more pronounced in nonspecialist health care environments due to limited resources, resulting in delayed referrals, increased misdiagnosis rates, and exacerbated disability outcomes for patients with SpA. The emergence of large language models (LLMs) in medical diagnostics introduces a revolutionary potential to overcome these diagnostic hurdles. Despite recent advancements in artificial intelligence and LLMs demonstrating effectiveness in diagnosing and treating various diseases, their application in SpA remains underdeveloped. Currently, there is a notable absence of SpA-specific LLMs and an established benchmark for assessing the performance of such models in this particular field. Our objective is to develop a foundational medical model, creating a comprehensive evaluation benchmark tailored to the essential medical knowledge of SpA and its unique diagnostic and treatment protocols. The model, post-pretraining, will be subject to further enhancement through supervised fine-tuning. It is projected to significantly aid physicians in SpA diagnosis and treatment, especially in settings with limited access to specialized care. Furthermore, this initiative is poised to promote early and accurate SpA detection at the primary care level, thereby diminishing the risks associated with delayed or incorrect diagnoses. A rigorous benchmark, comprising 222 meticulously formulated multiple-choice questions on SpA, will be established and developed. These questions will be extensively revised to ensure their suitability for accurately evaluating LLMs' performance in real-world diagnostic and therapeutic scenarios. Our methodology involves selecting and refining top foundational models using public data sets. The best-performing model in our benchmark will undergo further training. Subsequently, more than 80,000 real-world inpatient and outpatient cases from hospitals will enhance LLM training, incorporating techniques such as supervised fine-tuning and low-rank adaptation. We will rigorously assess the models' generated responses for accuracy and evaluate their reasoning processes using the metrics of fluency, relevance, completeness, and medical proficiency. Development of the model is progressing, with significant enhancements anticipated by early 2024. The benchmark, along with the results of evaluations, is expected to be released in the second quarter of 2024. Our trained model aims to capitalize on the capabilities of LLMs in analyzing complex clinical data, thereby enabling precise detection, diagnosis, and treatment of SpA. This innovation is anticipated to play a vital role in diminishing the disabilities arising from delayed or incorrect SpA diagnoses. By promoting this model across diverse health care settings, we anticipate a significant improvement in SpA management, culminating in enhanced patient outcomes and a reduced overall burden of the disease. DERR1-10.2196/57001.

Diverse task-driven modeling of macaque V4 reveals functional specialization towards semantic tasks.

Responses to natural stimuli in area V4-a mid-level area of the visual ventral stream-are well predicted by features from convolutional neural networks (CNNs) trained on image classification. This result has been taken as evidence for the functional role of V4 in object classification. However, we currently do not know if and to what extent V4 plays a role in solving other computational objectives. Here, we investigated normative accounts of V4 (and V1 for comparison) by predicting macaque single-neuron responses to natural images from the representations extracted by 23 CNNs trained on different computer vision tasks including semantic, geometric, 2D, and 3D types of tasks. We found that V4 was best predicted by semantic classification features and exhibited high task selectivity, while the choice of task was less consequential to V1 performance. Consistent with traditional characterizations of V4 function that show its high-dimensional tuning to various 2D and 3D stimulus directions, we found that diverse non-semantic tasks explained aspects of V4 function that are not captured by individual semantic tasks. Nevertheless, jointly considering the features of a pair of semantic classification tasks was sufficient to yield one of our top V4 models, solidifying V4's main functional role in semantic processing and suggesting that V4's selectivity to 2D or 3D stimulus properties found by electrophysiologists can result from semantic functional goals.

A survey of generative AI for de novo drug design: new frontiers in molecule and protein generation.

Artificial intelligence (AI)-driven methods can vastly improve the historically costly drug design process, with various generative models already in widespread use. Generative models for de novo drug design, in particular, focus on the creation of novel biological compounds entirely from scratch, representing a promising future direction. Rapid development in the field, combined with the inherent complexity of the drug design process, creates a difficult landscape for new researchers to enter. In this survey, we organize de novo drug design into two overarching themes: small molecule and protein generation. Within each theme, we identify a variety of subtasks and applications, highlighting important datasets, benchmarks, and model architectures and comparing the performance of top models. We take a broad approach to AI-driven drug design, allowing for both micro-level comparisons of various methods within each subtask and macro-level observations across different fields. We discuss parallel challenges and approaches between the two applications and highlight future directions for AI-driven de novo drug design as a whole. An organized repository of all covered sources is available at https://github.com/gersteinlab/GenAI4Drug.

Metabolomic prediction of severe maternal and newborn complications in preeclampsia

IntroductionPreeclampsia (PreE) remains a major source of maternal and newborn complications. Prenatal prediction of these complications could significantly improve pregnancy management.ObjectivesUsing metabolomic analysis we investigated the prenatal prediction of maternal and newborn complications in early and late PreE and investigated the pathogenesis of such complications.MethodsSerum samples from 76 cases of PreE (36 early-onset and 40 late-onset), and 40 unaffected controls were collected. Direct Injection Liquid Chromatography–Mass Spectrometry combined with Nuclear Magnetic Resonance (NMR) spectroscopy was performed. Logistic regression analysis was used to generate models for prediction of adverse maternal and neonatal outcomes in patients with PreE. Metabolite set enrichment analysis (MSEA) was used to identify the most dysregulated metabolites and pathways in PreE.ResultsForty-three metabolites were significantly altered (p < 0.05) in PreE cases with maternal complications and 162 metabolites were altered in PreE cases with newborn adverse outcomes. The top metabolite prediction model achieved an area under the receiver operating characteristic curve (AUC) = 0.806 (0.660–0.952) for predicting adverse maternal outcomes in early-onset PreE, while the AUC for late-onset PreE was 0.843 (0.712–0.974). For the prediction of adverse newborn outcomes, regression models achieved an AUC = 0.828 (0.674–0.982) in early-onset PreE and 0.911 (0.828–0.994) in late-onset PreE. Profound alterations of lipid metabolism were associated with adverse outcomes.ConclusionPrenatal metabolomic markers achieved robust prediction, superior to conventional markers for the prediction of adverse maternal and newborn outcomes in patients with PreE. We report for the first-time the prediction and metabolomic basis of adverse maternal and newborn outcomes in patients with PreE.

Mapping the Intersubunit Interdomain FMN-Heme Interactions in Neuronal Nitric Oxide Synthase by Targeted Quantitative Cross-Linking Mass Spectrometry.

Nitric oxide synthase (NOS) in mammals is a family of multidomain proteins in which interdomain electron transfer (IET) is controlled by domain-domain interactions. Calmodulin (CaM) binds to the canonical CaM-binding site in the linker region between the FMN and heme domains of NOS and allows tethered FMN domain motions, enabling an intersubunit FMN-heme IET in the output state for NO production. Our previous cross-linking mass spectrometric (XL MS) results demonstrated site-specific protein dynamics in the CaM-responsive regions of rat neuronal NOS (nNOS) reductase construct, a monomeric protein [Jiang et al., Biochemistry, 2023, 62, 2232-2237]. In this work, we have extended our combined approach of XL MS structural mapping and AlphaFold structural prediction to examine the homodimeric nNOS oxygenase/FMN (oxyFMN) construct, an established model of the NOS output state. We employed parallel reaction monitoring (PRM) based quantitative XL MS (qXL MS) to assess the CaM-induced changes in interdomain dynamics and interactions. Intersubunit cross-links were identified by mapping the cross-links onto top AlphaFold structural models, which was complemented by comparing their relative abundances in the cross-linked dimeric and monomeric bands. Furthermore, contrasting the CaM-free and CaM-bound nNOS samples shows that CaM enables the formation of the intersubunit FMN-heme docking complex and that CaM binding induces extensive, allosteric conformational changes across the NOS regions. Moreover, the observed cross-links sites specifically respond to changes in ionic strength. This indicates that interdomain salt bridges are responsible for stabilizing and orienting the output state for efficient FMN-heme IET. Taken together, our targeted qXL MS results have revealed that CaM and ionic strength modulate specific dynamic changes in the CaM/FMN/heme complexes, particularly in the context of intersubunit interdomain FMN-heme interactions.

Broad-scale acoustic monitoring of koala populations suggests metapopulation stability, but varying bellow rate, in the face of major disturbances and climate extremes.

Population trends are lacking for most threatened species, especially those that are cryptic and difficult to survey. Recent developments in passive acoustics and semi-automated call recognition provide a cost-effective option to systematically monitor populations of vocal species. We assessed recent trends for the koala Phascolarctos cinereus, an iconic marsupial, based on 7 years of acoustic monitoring across 224 forested sites. The study period overlapped with a severe drought and extensive megafires in 2019 followed by 2 years of extremely high rainfall. Dynamic occupancy modelling with a range of covariates at multiple landscape scales found that initial occupancy was related to elevation (-ve), NDVI (+ve) and previous selective harvesting (16-30-year age class; weakly +ve). Extinction probability increased with the extent of high-severity fire. Colonisation probability was related to a range of factors, with the top model showing a decrease with increasing lagged annual rainfall. However, the null model was also supported, suggesting weak associations for colonisation. Using these relationships, koala occupancy was found to be high and stable over the study period. Fire did not influence regional trends because koalas often persisted with low- to moderate-severity fire and because high-severity fire was limited to 11% of their habitat. In contrast, bellow rate varied across years, being initially low and declining immediately after the 2019 fires, with the driver of these changes unclear. Neither timber harvesting nor low-severity fire influenced koala occupancy or bellow rate. Given the extensive area of koala habitat in the region, our results point to the presence of a large population in these public forests, and in recent years, stable occupancy (albeit with site-scale reductions in density with high-severity fire). Ongoing monitoring is crucial for tracking future changes, especially with predictions of more frequent, severe forest fires due to climate change.

Research on optimization of basic rail top bending prediction model

Since the basic rail of the switch needs to have a certain bending angle when the train changes direction, top bending is an important link in the production process of the basic rail. The three-point pressure top bending method is simple, flexible and widely used. In this study, the traditional three-point pressure bending is optimized, the influence of the pick width in the model is considered, a corresponding rebound model is established, and the model is applied to the pressure bending process of the basic rail. The bilinear strengthening model of the material was used to construct the bending moment expressions at different positions during the top bending process, and the relationship between the load and bending deflection in the elastic stage and elastic-plastic stage was obtained. The final top bending prediction model was obtained by combining the load-deflection model in the bending stage and the rebound stage. The correctness of the theoretical mathematical model was verified by establishing finite element simulations, and the theoretical calculation results were compared with the experimental results. The results showed that the top bend prediction optimization model established in this study had high feasibility and met the machining accuracy requirements.

Machine learning-based multi-performance prediction and analysis of Earth-Air Heat Exchanger

The Earth-Air Heat Exchanger (EAHE) plays a crucial role in global renewable energy utilization. Despite various EAHE models, a machine learning-based multi-performance prediction model is lacking. This study introduces a framework based on diverse machine learning models for precise prediction and comprehensive analysis of EAHE's multiple performance indicators. Initially, the theoretical EAHE model is established and validated. Six feature variables and four target variables are input to generate a dataset for machine learning. Six machine learning models are then selected for training, optimization, and validation. Subsequently, model interpretation and analysis are conducted using SHapley additive exPlanations and Partial dependence plots. Results indicate the optimized-extreme gradient boosting model performs the best, with a determination coefficient of 0.98, root mean squared error of 22.98, and mean absolute error of 8.88, identified as the top multi-performance prediction model for EAHE. Notably, pipe diameter, air velocity, and ground type significantly influence EAHE performance. Complex correlations and mutual exclusions exist among target variables, with a strong negative correlation (−0.59) between the coefficient of performance and payback time, highlighting the challenge of simultaneous improvement. This study presents a valuable framework for predicting and analyzing EAHE performance, crucial for promoting its widespread application and enhancing renewable energy utilization.

Top management team social capital and business model innovation: moderating role of business ecosystem health

PurposeThis study explores the influence of the Top Management Team (TMT) social capital on business model innovation in business ecosystems.Design/methodology/approachThis study examines the impact of internal and external TMT social capital on enterprises’ business model innovation, explores the relationship between internal and external TMT social capital, and investigates how business ecosystem health moderates the relationship between external TMT social capital and enterprises’ business model innovation. These hypotheses are proposed and tested using a hierarchical regression analysis with data from 168 Chinese firms.FindingsFirst, both internal and external TMT social capital exert a significantly positive influence on an enterprise’s business model innovation. Second, internal TMT social capital positively contributes to the development of external TMT social capital, affecting business model innovation. Finally, the moderating effect of business ecosystem health on the relationship between external TMT social capital and business model innovation depends on the dimensions. Specifically, the productivity of the business ecosystem negatively moderates this relationship, whereas the niche creation capability of the business ecosystem has a positive moderating effect.Originality/valueThese findings enrich prior research on business model innovation within the business ecosystem, thoroughly exploring the critical role of TMT social capital. This study reveals the diverse impacts of internal and external TMT social capital on business model innovation and the intricate relationship between these elements. Furthermore, it emphasizes that the success of enterprise’s business model innovation within a business ecosystem depends on the alignment and adaptation to dynamic ecosystem conditions. By presenting these insights, this study provides valuable practical implications for enterprises aiming to cultivate social capital within business ecosystem to facilitate business model innovation.

Going beyond Top EFT

We present a new way to interpret Top Standard Model measurements going beyond the SMEFT framework. Instead of the usual paradigm in Top EFT, where the main effects come from tails in momenta distributions, we propose an interpretation in terms of new physics which only shows up at loop-level. The effects of these new states, which can be lighter than required within the SMEFT, appear as distinctive structures at high momenta, but may be suppressed at the tails of distributions. As an illustration of this phenomena, we present the explicit case of a UV model with a Z\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$ \\mathcal{Z} $$\\end{document}2 symmetry, including a Dark Matter candidate and a top-partner. This simple UV model reproduces the main features of this class of signatures, particularly a momentum-dependent form factor with more structure than the SMEFT. As the new states can be lighter than in SMEFT, we explore the interplay between the reinterpretation of direct searches for colored states and Dark Matter, and Top measurements, made by ATLAS and CMS in the differential tt¯\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$ t\\overline{t} $$\\end{document} final state. We also compare our method with what one would expect using the SMEFT reinterpretation, finding that using the full loop information provides a better discriminating power.

Exploration of Girls’ Role Models: Are there Female STEM Role Models in Sight?

This study explores the presence and impact of female role models in Science, Engineering, Technology and Mathematics (STEM) education due to the underrepresentation of girls in STEM fields. The objective is to assess girls’ role models, particularly the role they play in their interest and in a career choice in STEM.A survey was distributed through an online panel and 546 valid answers from girls aged 14-20 in Germany were received. The analysis links to social cognitive theory and revealed that immediate family members (50,1%), musicians, actors and artists (12,4%), and media influencers (9,6%) were the top role model categories. Male role models dominated in teachers, entrepreneurs, and friends, while female role models were prominent in the family category. Female scientists and entrepreneurs who could serve as STEM role models, were found to be underrepresented. The study also showed that girls with STEM occupational role models in their immediate surroundings were more likely to have an interest in STEM subjects.

The Influence of The Reality Show 'Indonesia’s Next Top Model' on Beauty Perception Among Instagram Followers

Reality shows serve as entertainment platforms for society, conveying implicit messages and portrayals that influence how audiences perceive and follow what is presented. Audiences utilize Instagram and YouTube as mediums to engage with the frequency, attention, duration, and appreciation of the reality show "Indonesia’s Next Top Model." Instagram, in particular, serves as an interactive tool through which viewers express comments and appreciation for the show. This study aims to investigate whether there is an influence of reality show content on the perception of female beauty, focusing on the viewers of "Indonesia’s Next Top Model" among the followers of Instagram account @intm_nettv. The research employs a quantitative method with an explanatory research design. The unit of analysis comprises individuals who are followers of the Instagram account @intm_nettv. Data collection is conducted using probability sampling, obtaining a sample of 100 respondents who watch "Indonesia’s Next Top Model." The findings reveal a small influence of 8.2% of the reality show on the perception of female beauty among the followers of Instagram account @intm_nettv. This minimal impact is attributed to the followers' ability to distinguish between the beauty of models and that of other women. Various factors can affect the perception of female beauty, particularly reality show content as entertainment for its audience. Therefore, future research is encouraged to further explore the concept of reality show content and its influence on the perception of female beauty.

An Interpretable Machine Learning Approach to Predict Survival Outcomes in Spinal and Sacropelvic Chordomas.

Retrospective, population-based cohort study. This study aimed to develop machine learning (ML) models to predict five-year and 10-year mortality in spinal and sacropelvic chordoma patients and integrate them into a web application for enhanced prognostication. Past research has uncovered factors influencing survival in spinal chordoma patients. While identifying individual predictors is important, personalized survival predictions are equally vital. Though prior efforts have resulted in nomograms aiming to serve this purpose, they cannot capture complex interactions within data and rely on statistical assumptions that may not fit real-world data. Adult spinal and sacropelvic chordoma patients were identified from the National Cancer Database. Sociodemographic, clinicopathologic, diagnostic, and treatment-related variables were utilized as predictive features. Five supervised ML algorithms (TabPFN, CatBoost, XGBoost, LightGBM, and Random Forest) were implemented to predict mortality at five and 10 years postdiagnosis. Model performance was primarily evaluated using the area under the receiver operating characteristic (AUROC). SHapley Additive exPlanations (SHAP) values and partial dependence plots provided feature importance and interpretability. The top models were integrated into a web application. From the NCDB, 1206 adult patients diagnosed with histologically confirmed spinal and sacropelvic chordomas were retrieved for the five-year mortality outcome [423 (35.1%) with five-year mortality] and 801 patients for the 10-year mortality outcome [588 (73.4%) with 10-year mortality]. Top-performing models for both of the outcomes were the models created with the CatBoost algorithm. The CatBoost model for five-year mortality predictions displayed a mean AUROC of 0.801, and the CatBoost model predicting 10-year mortality yielded a mean AUROC of 0.814. This study developed ML models that can accurately predict five-year to 10-year survival probabilities in spinal chordoma patients. Integrating these interpretable, personalized prognostic models into a web application provides quantitative survival estimates for a given patient. The local interpretability enables transparency into how predictions are influenced. Further external validation is warranted to support generalizability and clinical utility.

Designing high-efficiency organic semi-conductors for organic photodetectors assisted by machine learning and property prediction

Quantum chemical calculations have been widely utilized to design and explore organic semi-conductors for various photovoltaics applications. Photovoltaic devices such as photodetectors work on the principle of photo-electric-effect and widely used for optical communications. In recent years, machine learning has become famous in many research fields. In present study, we have predicted reorganization energy of organic semi-conductors with the aid of machine learning. Statistical analysis is used to select best molecular descriptors (features). About 10 machine learning models are tested for selecting the best one. Among various tested models, gradient forest regressor and random forest regressor are the top models. By using our new designing approach, easy to synthesize building blocks are connected to design new organic semiconductors. Their reorganization energies are predicted using fast machine learning method. For this purpose, already trained and hyper parameter optimized gradient boosting regressor is used. It is generally accepted that to obtain higher charge mobility molecules should have lower reorganization energy. Ten organic semi-conductors with lower reorganization energy are selected for density functional calculations and their energy levels are calculated. The chemical similarity analysis using clustering and heatmap is also done for selected organic semi-conductors. Synthetic accessibility score is calculated and majority of semi-conductors are easy to synthesize. These molecules are potential candidates for device fabrication. The proposed python-based pipeline has ability to select the molecules for organic photodetectors in fast and cheaper way.

"Next top" mouse models advancing CTCL research.

This review systematically describes the application of in vivo mouse models in studying cutaneous T-cell lymphoma (CTCL), a complex hematological neoplasm. It highlights the diverse research approaches essential for understanding CTCL's intricate pathogenesis and evaluating potential treatments. The review categorizes various mouse models, including xenograft, syngeneic transplantation, and genetically engineered mouse models (GEMMs), emphasizing their contributions to understanding tumor-host interactions, gene functions, and studies on drug efficacy in CTCL. It acknowledges the limitations of these models, particularly in fully replicating human immune responses and early stages of CTCL. The review also highlights novel developments focusing on the potential of skin-targeted GEMMs in studying natural skin lymphoma progression and interactions with the immune system from onset. In conclusion, a balanced understanding of these models' strengths and weaknesses are essential for accelerating the deciphering of CTCL pathogenesis and developing treatment methods. The GEMMs engineered to target specifically skin-homing CD4+ T cells can be the next top mouse models that pave the way for exploring the effects of CTCL-related genes.

Prediction of pediatric peanut oral food challenge outcomes using machine learning

Clinical testing, including food-specific skin and serum IgE level tests, provides limited accuracy to predict food allergy. Confirmatory oral food challenges (OFCs) are often required, but the associated risks, cost, and logistic difficulties comprise a barrier to proper diagnosis. We sought to utilize advanced machine learning methodologies to integrate clinical variables associated with peanut allergy to create a predictive model for OFCs to improve predictive performance over that of purely statistical methods. Machine learning was applied to the Learning Early about Peanut Allergy (LEAP) study of 463 peanut OFCs and associated clinical variables. Patient-wise cross-validation was used to create ensemble models that were evaluated on holdout test sets. These models were further evaluated by using 2 additional peanut allergy OFC cohorts: the IMPACT study cohort and a local University of Michigan cohort. In the LEAP data set, the ensemble models achieved a maximum mean area under the curve of 0.997, with a sensitivity and specificity of 0.994 and 1.00, respectively. In the combined validation data sets, the top ensemble model achieved a maximum area under the curve of 0.871, with a sensitivity and specificity of 0.763 and 0.980, respectively. Machine learning models for predicting peanut OFC results have the potential to accurately predict OFC outcomes, potentially minimizing the need for OFCs while increasing confidence in food allergy diagnoses.

Broadscale and fine‐scale variables predict the occurrence of a stream‐breading bufonid: Habitat use by the Arizona toad (Anaxyrus microscaphus)

AbstractArizona toad (Anaxyrus microscaphus) is a stream‐breeding bufonid of conservation concern in Arizona, New Mexico, Utah, and Nevada. We determined the occupancy and habitat use of the Arizona toad throughout its range. We surveyed 500‐m reaches along perennial and intermittent streams during the summers of 2021 and 2022 (n = 232) in Arizona. We recorded the presence of all toad life stages, focusing on larval stages. We related toad occupancy to broadscale environmental variables, including measures of bioclimate, habitat heterogeneity, solar radiation, and topography. We collected fine‐scale variables to summarize vegetation cover and substrate within plots (n = 53). We applied multiple occupancy models. Single‐species model results found low toad occupancy with high detection. Two principal component analyses (PCA) were run on broadscale and fine‐scale variables to reduce the number of variables included in the models. Toad occupancy was best predicted by top models with bioclimatic components; occupancy decreases with extremely hot temperatures and less precipitation. A logistic regression related toad presence to fine‐scale components with top models describing riparian complexity and algae. Arizona toads were selected for areas with foliar canopy cover, shallow water, algae cover, and pebble substrates. Arizona toad is an uncommon species and maintaining riparian forests of complex habitats with shallow and side‐channel flow will be important for toad conservation. Implications of less water cause riparian habitat to dry, which ultimately harms aquatic life.

A tapestry of habitats: exploring abundance and habitat preferences of the Northern Red Muntjac (Muntiacus vaginalis) across the Central Himalayan landscape

Abstract Understanding the occupancy and density estimate of Northern Red Muntjac (Muntiacus vaginalis) at a landscape scale is crucial due to their significant ecological impact on ecosystems, particularly as prey for large predators. This study employed a single-season site occupancy model to assess the occupancy and density of Northern Red Muntjac using camera traps in the east Sikkim and Darjeeling hills Central Himalayan landscape from 2018 to 2020. A total of 513 photo captures of Northern Red Muntjac were obtained from 3150 trap nights with a naive occupancy estimate of 0.64 for the Central Himalayan landscape. The top model revealed positive influences of broadleaved forest (β = 0.93 ± 0.26), mixed forest (β = 0.6 ± 0.47), and slope (β = 0.2 ± 0.24) on the occupancy probability, while elevation (β = −0.08 ± 0.24) and ruggedness (β = −0.53 ± 0.23) had negative influences. Similarly, broadleaved forest (β = 1.3 ± 0.24), mixed forest (β = 0.74 ± 0.39), and ruggedness (β = 0.62 ± 0.25) positively influenced the detection probability, whereas elevation (β = −0.3 ± 0.22) and slope (β = −1.13 ± 1.21) had negative effects. The study estimated a total abundance of 169.23 ± 20.77 individuals with a density of 0.33 per square kilometre in the landscape. This study is a pioneering effort and provides a foundation for future investigations into the occupancy and density estimate of Northern Red Muntjac in the Central Himalayas.

An explainable long short-term memory network for surgical site infection identification

BackgroundCurrently, surgical site infection surveillance relies on labor-intensive manual chart review. Recently suggested solutions involve machine learning to identify surgical site infections directly from the medical record. Deep learning is a form of machine learning that has historically performed better than traditional methods while being harder to interpret. We propose a deep learning model, a long short-term memory network, for the identification of surgical site infection from the medical record with an attention layer for explainability. MethodsWe retrieved structured data and clinical notes from the University of Utah Health System’s electronic health care record for operative events randomly selected for manual chart review from January 2016 to June 2021. Surgical site infection occurring within 30 days of surgery was determined according to the National Surgical Quality Improvement Program definition. We trained the long short-term memory model along with traditional machine learning models for comparison. We calculated several performance metrics from a holdout test set and performed additional analyses to understand the performance of the long short-term memory, including an explainability analysis. ResultsSurgical site infection was present in 4.7% of the total 9,185 operative events. The area under the receiver operating characteristic curve and sensitivity of the long short-term memory was higher (area under the receiver operating characteristic curve: 0.954, sensitivity: 0.920) compared to the top traditional model (area under the receiver operating characteristic curve: 0.937, sensitivity: 0.736). The top 5 features of the long short-term memory included 2 procedure codes and 3 laboratory values. ConclusionSurgical site infection surveillance is vital for the reduction of surgical site infection rates. Our explainable long short-term memory achieved a comparable area under the receiver operating characteristic curve and greater sensitivity when compared to traditional machine learning methods. With explainable deep learning, automated surgical site infection surveillance could replace burdensome manual chart review processes.