Comparison Of Models Research Articles

A Visualization Tool to Study Dyadic Caregiving Health Profiles.

Guided by a life course perspective and fundamental cause theory, this study aims to visualize co-trajectories of health between partners and examine how changes in one spouse's cognitive status can cohesively impact the health of the other spouse along three dimensions (functional, mental, and cognitive). Drawing longitudinal data from the Health and Retirement Study 2000-2016 (N = 3582), we measure women's health profiles by functional limitation (physical health), depression (mental health), and cognitive function (cognitive health). We use multivariate linear mixed models to summarize these paths in the same visual representation. The approach provides a visualization tool that depicts data and model in the same spatial representation allowing assessment of model fit and comparison. This study advances the traditional life course studies by representing underlying processes as a multidimensional time vector of health outcomes. The described approach provides a blueprint for studying complex health profiles or trajectories.

Optimizing recombinant antibody fragment production: A comparison of artificial intelligence and statistical modeling.

Maximizing the recombinant protein yield necessitates optimizing the production medium. This can be done using a variety of methods, including the conventional "one-factor-at-a-time" approach and more recent statistical and mathematical methods such as artificial neural network (ANN), genetic algorithm, etc. Every approach has advantages and disadvantages of its own, yet even when a technique has flaws, it is nevertheless used to get the best results. Here, one categorical variable and four numerical parameters, including post-induction time, inducer concentration, post-induction temperature, and pre-induction cell density, were optimized using the 232 experimental assays of the central composite design. The direct and indirect effects of factors on the yield of anti-epithelial cell adhesion molecule extracellular domain fragment antibody were examined using statistical methods. The analysis of variance results indicate that the response surface methodology (RSM) model is effective in predicting the amount of produced single-chain fragment variable (p-value = 0.0001 and R2 = 0.905). For ANN modeling, the evaluation using normalized root mean square error (NRMSE) and R2 values shows a good fit (R2 = 0.942) and accurate predictions (NRMSE = 0.145). The analysis of error parameters and R2 of a dataset, which contained 30 data points randomly selected from the complete dataset, showed that the ANN model had a higher R2 value (0.968) compared to the RSM model (0.932). Furthermore, the ANN model demonstrated stronger predictive ability with a lower NRMSE (0.048vs. 0.064). Induction at the cell density of 0.7 and an isopropyl β-D-1-thiogalactopyranoside concentration of 0.6mM for 32 h at 30°C in BW25113 was the ideal culture condition leading to the protein yield of 259.51mg/L. Under the optimum conditions, the output values predicted by the ANN model (259.83mg/L) were more in line with the experimental data (259.51mg/L) than the RSM (276.13mg/L) expected value. This outcome demonstrated that the ANN model outperforms the RSM in terms of prediction accuracy.

Revisiting the mechanism of β-O-4 bond cleavage during acidolysis of lignin part 10: reactions of C6–C2-type non-phenolic syringyl model compounds and comparison of the reactions with those of the guaiacyl analogues

A C6–C2-type non-phenolic syringyl lignin model compound (I, 2-(2-methoxyphenoxy)-1-(3,4,5-trimethoxyphenyl)ethanol) as well as its derivatives was acidolyzed in aqueous 82vol% 1,4-dioxane containing 0.2 mol/L HBr, HCl, or H2SO4 at 85ºC. The results were compared with those on the guaiacyl analogues (V as well as its derivatives) obtained in our previous reports. The acidolysis of compound I was slower than that of compound V, which is in accordance with our previous result on the formation rates of the benzyl cation intermediates (BC) from a syringyl compound and its guaiacyl analogue. The enol ether-type compound was primarily produced in the acidolysis of compound I, similar to that of compound V. The acidolysis using HBr or HCl was faster than that using H2SO4, indicating the participation of Br¯ or Cl¯, respectively, in the acidolysis. It was suggested that Br¯ (as well as Cl¯) adds to the cation center of the BC to afford the adduct in the acidolysis of compound I, similar to that of compound V, and hence, this adduct formation is a major bypass of the common route. The bypass activity in the acidolysis of compound I was about half in that of compound V.

Comparison of numerical calculations and ALOHA modeling in consequence assessment of chlorine gas emissions from ethylene dichloride reactors.

Incidents involving chemical storage tanks in the petrochemical industry are significant events with severe consequences. Within the petrochemical industry, EDC is a sector that produces ethylene dichloride through the reaction of chlorine and ethylene. The present research was conducted to evaluate the consequences of chlorine gas released from the EDC reactor in a petrochemical industry in southern Iran. Data regarding reactor specifications were obtained from the factory's technical office, while climatic data was acquired from the Meteorological Organization. The consequences of chlorine gas release from the reactor were assessed in four predefined scenarios using numerical calculation methods and modeling with the ALOHA software. The numerical calculation method involved thermodynamic fluid path analysis, discharge coefficient calculations, and wind speed impact analysis. The hazard radius was determined based on the ERPG1-2-3 index. Results showed that in the scenario of chlorine gas release from EDC reactors, according to the ALOHA model, an increase in wind speed from 3 to 7m/h led to an expanded dispersion radius. At a radius of 700m from the reactor, the maximum outdoor concentration reached 3.12ppm, decreasing to 2.27ppm at 800m and further to 1.53ppm at 1000m. The comparison of numerical calculations and modeling using the ALOHA software indicates the desirable conformity of the results with each other. The R2 coefficient for evaluating the conformity of the results was 0.9964, indicating the desired efficiency of the model in evaluating the consequences of the release of toxic gasses from the EDC tank. The results of this research can be useful in designing the site and emergency response plan.

Forecaster evaluation as model comparison.

Forecaster evaluation as model comparison.

"Leap before you look": Conditions that suppress explicit, knowledge-based learning during visuomotor adaptation.

When learning a novel visuomotor mapping (e.g., mirror writing), accuracy can improve quickly through explicit, knowledge-based learning (e.g., aim left to go right), but after practice, implicit or procedural learning takes over, producing fast, natural movements. This procedural learning occurs automatically, whereas it has recently been found that knowledge-based learning can be suppressed by the gradual introduction of the novel mapping when participants must make fast movements and visuomotor perturbations are small (e.g., 30° rotations). We explored the range of task instructions, perturbation parameters, and feedback that preclude or encourage this suppression. Using a reaching task with a rotation between screen position and movement direction, we found that knowledge-based learning could be suppressed even for an extreme 90° rotation, but only if it was introduced gradually and only under instructions to move quickly. If the rotation was introduced abruptly or if instructions emphasized accuracy over speed, knowledge-based learning occurred. A second experiment indicated that knowledge-based learning always occurred in the absence of continuous motion feedback, evidenced by the time course of learning, the aftereffects of learning when the rotation was abruptly removed, and the outcome of formal model comparison between a dual-state (procedural and knowledge-based) versus a single-state (procedural only) learning model of the data. A third experiment replicated the findings and verified that the knowledge-based component of the dual-state model corresponded to explicit aiming, whereas the procedural component was slow to unlearn. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Predictive ecological niche model for Cinnamomum parthenoxylon (Jack) Meisn. (Lauraceae) from Last Glacial Maximum to future in Vietnam

Cinnamomum parthenoxylon (Jack) Meisn. is a tree in genus Cinnamomum that has been facing global threats due to forest degradation and habitat fragmentation. Many recent studies aim to describe habitats and assess population and species genetic diversity for species conservation by expanding afforestation models for this species. Understanding their current and future potential distribution plays a major role in guiding conservation efforts. Using five modern machine-learning algorithms available on Google Earth Engine helped us evaluate suitable habitats for the species. The results revealed that Random Forest (RF) had the highest accuracy for model comparison, outperforming Support Vector Machine (SVM), Classification and Regression Trees (CART), Gradient Boosting Decision Tree (GBDT) and Maximum Entropy (MaxEnt). The results also showed that the extremely suitable ecological areas for the species are mostly distributed in northern Vietnam, followed by the North Central Coast and the Central Highlands. Elevation, Temperature Annual Range and Mean Diurnal Range were the three most important parameters affecting the potential distribution of C. parthenoxylon. Evaluation of the impact of climate on its distribution under different climate scenarios in the past (Last Glacial Maximum and Mid-Holocene), in the present (Worldclim) and in the future (using four climate change scenarios: ACCESS, MIROC6, EC-Earth3-Veg and MRI-ESM2-0) revealed that of C. parthenoxylon would likely expand to the northeast, while a large area of central Vietnam will gradually lose its adaptive capacity by 2100.

Artificial neural networks for model identification and parameter estimation in computational cognitive models.

Computational cognitive models have been used extensively to formalize cognitive processes. Model parameters offer a simple way to quantify individual differences in how humans process information. Similarly, model comparison allows researchers to identify which theories, embedded in different models, provide the best accounts of the data. Cognitive modeling uses statistical tools to quantitatively relate models to data that often rely on computing/estimating the likelihood of the data under the model. However, this likelihood is computationally intractable for a substantial number of models. These relevant models may embody reasonable theories of cognition, but are often under-explored due to the limited range of tools available to relate them to data. We contribute to filling this gap in a simple way using artificial neural networks (ANNs) to map data directly onto model identity and parameters, bypassing the likelihood estimation. We test our instantiation of an ANN as a cognitive model fitting tool on classes of cognitive models with strong inter-trial dependencies (such as reinforcement learning models), which offer unique challenges to most methods. We show that we can adequately perform both parameter estimation and model identification using our ANN approach, including for models that cannot be fit using traditional likelihood-based methods. We further discuss our work in the context of the ongoing research leveraging simulation-based approaches to parameter estimation and model identification, and how these approaches broaden the class of cognitive models researchers can quantitatively investigate.

Evaluation of a digital Cognitive Behavioural Therapy programme for prison residents with histories of substance misuse (“Breaking Free”): Comparison of delivery models employed in the United Kingdom and the United States

Substance misuse is a criminogenic risk factor—however, substance-involved prison residents may have limited access to support and services that might be able to help them overcome their substance misuse. Research around the efficacy of the “Breaking Free” digital Cognitive Behavioural Therapy programme for substance-involved prison residents demonstrates that this digital technology may overcome barriers to support. Breaking Free is the first digital programme for substance misuse that has been implemented in prisons. In the United Kingdom (U.K.) the programme is delivered via secure desktop and laptop computers in staff-facilitated group sessions, and in the United States (U.S.) is delivered via individual secure tablets that prison residents can access independently, without staff support. This study compares U.K. and U.S. prison residents who have engaged with Breaking Free on a number of baseline measures (sociodemographic factors, types of problem substances reported, clinical complexity, and criminogenic risk factors). The delivery models employed in these two jurisdictions are compared in relation to their associations with prison residents’ engagement with the programme and the clinical outcomes experienced. Differences between U.K. and U.S. prison residents in baseline clinical and psychosocial factors were identified, including greater clinical complexity in the U.K. group. U.K. prison residents, who were supported by staff when using Breaking Free, demonstrated higher levels of engagement with the programme than U.S. prison residents who were not supported by staff. The different delivery models employed in the United Kingdom and United States appeared to be associated with group differences in clinical outcomes. Guidance for the optimal implementation of Breaking Free, based on findings from this study, is provided. Further analyses will explore how the different approaches to the delivery of Breaking Free might interact with the differences in prison resident characteristics identified in this study (sociodemographic, clinical and criminogenic), in order to influence prison resident engagement with the programme and clinical outcomes.

Next‐generation energy‐efficient optical networking: DQ‐RGK algorithm for dynamic quality of service and adaptive resource allocation

SummaryIn green optical networking, designing an adaptive energy‐saving scheme plays a vital role, in optimizing energy consumption by dynamically adjusting resources based on network traffic and environmental conditions, to a more sustainable and efficient optical communication infrastructure. Traditional methods in optical networking face challenges such as static resource allocation, limited adaptability, inefficient power usage, environmental insensitivity, and scalability issues. Therefore this article proposed a novel method named Dynamic Quality of Service based Random update Genghis Khan (DQ‐RGK) algorithm, the proposed model can tackle the abovementioned complexities. In this study, cluster head dynamic placement is utilized to optimize the network's performance by adapting the placement of cluster heads to the current topology, load distribution, and energy levels in the network nodes. Additionally, Dynamic Quality of Service (QoS) is employed to respond dynamically to changes in network conditions, adapting to varying traffic patterns and resource availability. In this work, the Genghis Khan Shark optimization with a random update strategy is implemented for hyperparameter optimization to enhance the performance of the DQ‐RGK method. The DQ‐RGK adjusts the parameters of QoS in real‐time, and this ensures that network resources based on the requirements changed and priorities of applications, which ultimately optimizes performance and enhances user experience. By dynamically assigning and reallocating resources based on the current demand the algorithm enhances overall network efficiency and reduces energy consumption. Then, this work analyzes the experimental results, where some evaluation measures estimate the DQ‐RGK method's performance. Routing efficiency, latency, scalability, spectral efficiency, Packet Delivery Ratio, throughput, network lifetime, energy consumption, jitter, and energy consumption are the measures employed by the DQ‐RGK model. In The results, other routing models that do not provide efficiency are utilized, a comparison of these other routing models is represented in results. The overall DQ‐RGK model's effectiveness is represented in the experimental results and its effectiveness is greater among other methods.

Comparison of machine learning and deep learning models for the assessment of rondo wine grape quality with a hyperspectral camera

Hyperspectral images provide rich spectral/spatial data and have shown remarkable performance in precision viticulture. Standardized data-processing methods are necessary to reduce the dimensionality and to identify powerful wavelengths. Toward this goal, we evaluated and compared the performance and novel-wavelength-identification ability of five renowned machine learning models: the linear models Ridge and LASSO, a 1D + 2D convolutional neural network (1D +2D CNN) non-linear model, a gradient boosting decision tree (GBDT) using XGBoost as an ensemble model, and an explainable boosting machine (EBM) followed by support vector regression (SVR) as a hybrid model. The model evaluations were conducted using leave-one-out cross-validation (LOOCV) as we sought to clarify the best-fitted machine learning model. Our results demonstrated that Ridge, LASSO, showed better performance with relatively high accuracy but were weak as a wavelength identifier. GDBT-XGBoost showed considerable prediction power and wavelength identification. EBM-SVM emerged as the most powerful model, demonstrating exceptional stability and clear wavelength classification even for destructive measurements under varying environmental stresses across Rondo's growth stages. The combined approach of 1D + 2D CNN algorithms was advantageous to handle the dynamic shapes of spectral curve and horizontal shift of the wavelengths obtained from outdoor data acquisition, and notably, it showed the highest accuracy to predict the brix and pH of wine grapes for both indoor and outdoor sensings. But the combined effects of 1D and 2D CNN algorithms were difficult to clarify the importance of spectral features for the brix and pH prediction. The integrated machine learning models with dimensionality reduction, and proper image acquisition can increase the model's accuracy. The common absorption peaks were observed in the near-infrared region of 700 nm and 900 nm. Those wavelengths should be considered for the development of low-cost sensing platforms with fewer bands. Wavelengths over 900 nm are also important to develop outdoor sensing platforms.

AN examination of linear factor models in U.K. stock returns in the presence of dynamic trading

AbstractThis study uses the approach of Ferson and Siegel, Rev Financ Stud 22:2735–2758 (2009), and Ferson, Siegel and Wang, J Financ Quant Anal, forthcoming, (2024) to examine the unconditional mean–variance efficiency, in the presence of conditioning information (UMV), of ten linear factor models in U.K. stock returns. The study finds that the UMV efficiency of all the multifactor models is strongly rejected in U.K. stock returns in two different sets of test assets. This rejection is mainly driven by allowing dynamic trading in the test assets and factors. The optimal use of conditioning information also has a significant impact in relative model comparison tests. In relative model comparison tests based on UMV efficiency, the best performing model is the eight-factor model of Chib and Zeng, J Bus Econ Stat 38:771–783 (2020) model.

Machine Learning-Based Predictive Models for Patients with Venous Thromboembolism: A Systematic Review.

Venous thromboembolism (VTE) is a chronic disorder with a significant health and economic burden. Several VTE-specific clinical prediction models (CPMs) have been used to assist physicians in decision-making but have several limitations. This systematic review explores if machine learning (ML) can enhance CPMs by analyzing extensive patient data derived from electronic health records. We aimed to explore ML-CPMs' applications in VTE for risk stratification, outcome prediction, diagnosis, and treatment. Three databases were searched: PubMed, Google Scholar, and IEEE electronic library. Inclusion criteria focused on studies using structured data, excluding non-English publications, studies on non-humans, and certain data types such as natural language processing and image processing. Studies involving pregnant women, cancer patients, and children were also excluded. After excluding irrelevant studies, a total of 77 studies were included. Most studies report that ML-CPMs outperformed traditional CPMs in terms of receiver operating area under the curve in the four clinical domains that were explored. However, the majority of the studies were retrospective, monocentric, and lacked detailed model architecture description and external validation, which are essential for quality audit. This review identified research gaps and highlighted challenges related to standardized reporting, reproducibility, and model comparison. ML-CPMs show promise in improving risk assessment and individualized treatment recommendations in VTE. Apparently, there is an urgent need for standardized reporting and methodology for ML models, external validation, prospective and real-world data studies, as well as interventional studies to evaluate the impact of artificial intelligence in VTE.

A face-to-face comparison of the BBB cell models hCMEC/D3 and hBMEC for their applicability to adenoviral expression of transporters.

The blood-brain barrier (BBB) is a structure mainly formed by brain capillary endothelial cells (BCEC) whose role is to regulate the exchange of compounds between the blood and the brain. In this process efflux and uptake transporters play a key role. Aim of this study was to compare the two previously established cell lines hCMEC/D3 and hBMEC as BBB cell models for the application of an adenoviral system to transiently express OATP2B1 and Pgp. Comparison of hCMEC/D3 and hBMEC mRNA and protein levels of BBB markers showed a unique expression pattern for each cell line. While showing similar expression of the efflux transporter BCRP, transferrin receptor (TFRC) and of the tight junctions proteins Occludin and ZO-1, hCMEC/D3 displayed higher levels of the endothelial marker PECAM1, VE-cadherin, Von Willebrand Factor (VWF) and of the efflux transporter Pgp. Moreover, measuring integrity of the monolayer by determining the Trans-Endothelial Electrical Resistance (TEER), electrical capacitance (CCl), and inulin apparent permeability coefficient (Papp) revealed higher TEER and lower CCl for hBMEC but comparable Papp in the two cell lines. Following adenoviral infection, enhanced OATP2B1 and Pgp expression and functionality could be observed only in hBMEC. Importantly, the adenoviral expression system did not affect expression of BBB markers and permeability in both cell lines. Taken together, our results provide first evidence that hBMEC is an applicable human BBB cell model in which adenoviral infection can be used to transiently express and investigate transporters of interest.

Comparison of time-domain and frequency-domain models for vibration prediction of a rail transit viaduct paved with floating slab track

Noise from rail transit viaducts often has a negative effect on surrounding buildings and nearby residents. To reduce the bridge vibration and radiated noise, floating slab tracks have been widely adopted in rail transit viaducts. To calculate the vibration of coupled track-bridge systems, both time-domain and frequency-domain methods can be applied although their relative performance is not very clear. This study aims to compare the two approaches and then provide some guidelines in choosing the appropriate method for predicting the vibration. The principle and methodology of the two calculation methods are firstly introduced, and the models for the train, floating slab track and bridge are developed using similar parameters. To further improve the efficiency and accuracy of the frequency-domain method, a half period method and a weighted average method are proposed to calculate the average responses of a certain point on the rail, floating slab and bridge to approximate the response during the pass-by. An indirect method is used to estimate the wheel–rail combined roughness from the measured rail acceleration when the train is running on the bridge, and compared with the direct measurement rail roughness. For the vibration of the bridge and floating slab, comparisons are made between results obtained from the time-domain and frequency-domain methods with different train speeds, and between results from experiment and simulation with different roughness. It is found that the wheel–rail contact forces and other structural responses from the two approaches show good consistency in the frequency region above 20 Hz. It is shown that a good approximation of the response during the pass-by can be obtained using the frequency-domain method by adopting the half period average or a simplified weighted average of two wheel position cases. When using the measured roughness and estimated roughness, the simulated vibration of the floating slab and bridge match well with the experimental results, and the indirect method can be used to modify direct measurement roughness in lower frequency regions. In the vibration prediction, the rail vibration or roughness, and other parameters in the vehicle–track–bridge system, should be measured accurately to ensure high reliability.

Analyzing the Impact of Geometrophysical Modeling on Highway Design Speeds: A Comparative Study for Mexico’s Case

This manuscript presents an examination of the impact of geometrical and physical parameters on highway design speeds, critical for traffic safety and efficiency. Originating from a classical dynamics discussion in an undergraduate automotive technology engineering class, an exploration of the consequences of different geometrophysical considerations on a vehicle’s dynamics over pavement surfaces is developed. Considering various analytical models, an assessment of their principles and the significance of geometric and physical concepts involved in the problem is made, such as plane of motion and trajectory curvature radius, on safe (non slippage) operational speeds. The subsequent comparative study shows that one of the most accepted models in highway design regulations in México, when used as reference, yields percentage error differences respect to others of 0≲%EMax≲5, as well as a consistent trend for relatively underestimating safe highway operational speeds. A discussion of the immediate implications of these findings, emphasizing the necessity of experimental studies to validate theoretical predictions, is presented. This work contributes to the field by providing a detailed comparison of analytical models under a general applied science perspective, suggesting modifications to current highway design practices in México based on geometrophysical insights. In summary, this work’s main aim is to shed light on the intricacies of determining safe design speeds from an applied sciences point of view, while also calling for a reevaluation of the existing guidelines to enhance highway design and safety.

Approaching Temporal Dynamics in the Dimension-Level Associations Between Career Adaptability/Ambivalence and Internalizing Symptoms Among Chinese Adolescents Throughout Their High Middle School Years.

Despite the well-established associations between adolescents' internalizing symptoms and career development, it still remains unclear whether adolescents' internalizing symptoms are precursors or consequences of their career adaptability/ambivalence. Subtler nuance inherent within such association also await to be revealed, because internalizing symptoms and career development have been primarily treated as broad constructs, despite the multifaceted nature of both. To narrow such gaps, this study examined the potentially dynamic associations among career adaptability, career ambivalence, and internalizing symptoms using three-wave longitudinal data. The study collected data from 3196 Chinese adolescents (52.72% girls, mean age = 15.56 years, SD = 0.58) at Wave 1, with 2820 (attrition rate = 11.76%) participating in Wave 2 and 2568 (attrition rate = 8.93%) in Wave 3. The measurement invariance suggested that there were no significant differences across both waves and genders. This study approached associations at both broader construct levels and subtler dimension levels. Results of cross-lagged path models at broader construct levels demonstrated a unidirectional association between internalizing symptoms and career adaptability. Results of models at subtler dimension levels indicated a series of transactional links over time between career adaptability dimensions/ambivalence and depressive symptoms in particular. Career adaptability dimensions and career ambivalence predicted later anxiety symptoms rather than the reverse. Group model comparisons showed no difference across waves and genders. These findings shed light on the dynamic nature of the associations during adolescence between career adaptability/ambivalence and internalizing symptoms, particularly at subtler dimensional levels, which should be considered in relevant clinical and educational practices.

Estimation of mixed-gas selectivity of membranes using the linear friction-based model

The development of analytical models for predicting the transport properties of membranes in the separation of multicomponent mixtures is highly relevant because (1) measurements of permeability and sorption of gas mixtures in membrane materials are laborious and time-consuming, and (2) estimation of transport properties based on pure gas measurements is desirable, however, rather unreliable and can lead to serious errors. In this paper, within the framework of the well-known frictional-coefficient formalism, we present nonlinear equations for the fluxes of gas mixture components through the membrane, which take into account the diffusional coupling of the fluxes compared to independent gas transport. As a result, an analytical equation for the mixed-gas selectivity was obtained. This equation has an implicit form, which is not quite convenient for practical application, since it has to be solved numerically. A simple explicit expression for mixed gas selectivity was derived using the so-called linear transport model, which is a linearized version of the original nonlinear flux equations. It is important to emphasize that a comparison of the linear model with the formally more rigorous nonlinear model showed only minor quantitative differences. The input parameters for the model are the single gas transport parameters. Testing the linear model using experimental data for the separation of hydrocarbons (n-butane/methane, n-butane/i-butane, propylene/propane) revealed an acceptable fit. A comparison of the model with recent experimental data on the separation of CO2-containing mixtures using microporous polymers is also presented. It is found that the calculated CO2/N2 mixed gas selectivity is in good agreement with the experimental data, while only a semi-quantitative agreement is observed for the CO2/CH4 separation. Besides, it is shown that the model developed for binary gas separation can be extended to ternary gas mixtures.

Comparison of deep learning models to traditional Cox regression in predicting survival of colon cancer: Based on the SEER database.

In this study, a deep learning algorithm was used to predict the survival rate of colon cancer (CC) patients, and compared its performance with traditional Cox regression. In this population-based cohort study, we used the characteristics of patients diagnosed with CC between 2010 and 2015 from the Surveillance, Epidemiology and End Results (SEER) database. The population was randomized into a training set (n=10596, 70%) and a test set (n=4536, 30%). Brier scores, area under the (AUC) receiver operating characteristic curve and calibration curves were used to compare the performance of the three most popular deep learning models, namely, artificial neural networks (ANN), deep neural networks (DNN), and long-short term memory (LSTM) neural networks with Cox proportional hazard (CPH) model. In the independent test set, the Brier values of ANN, DNN, LSTM and CPH were 0.155, 0.149, 0.148, and 0.170, respectively. The AUC values were 0.906 (95% confidence interval [CI] 0.897-0.916), 0.908 (95% CI 0.899-0.918), 0.910 (95% CI 0.901-0.919), and 0.793 (95% CI 0.769-0.816), respectively. Deep learning showed superior promising results than CPH in predicting CC specific survival. Deep learning showed potential advantages over traditional CPH models in terms of prognostic assessment and treatment recommendations. LSTM exhibited optimal predictive accuracy and has the ability to provide reliable information on individual survival and treatment recommendations for CC patients.

A comparison of global flood models using Sentinel-1 and a change detection approach

AbstractAdvances in numerical algorithms, improvement of computational power and progress in remote sensing have led to the development of global flood models (GFMs), which promise to be a useful tool for large-scale flood risk management. However, performance and reliability of GFMs, especially in data-scarce regions, is still uncertain, as they are difficult to validate. Here we aim at contributing to develop alternative, more flexible, and consistent methods for GFM validation by applying a change detection analysis on synthetic aperture radar (CD-SAR) imagery obtained from the Sentinel-1 imagery, on a cloud-based geospatial analysis platform. The study addresses two main objectives. First, to validate four widely adopted GFMs with flood maps generated through the proposed CD-SAR approach. This exercise was conducted for eight different large river basins on four continents, to account for a diverse range of hydro-climatic environments. Second, to compare CD-SAR-derived flood maps with those obtained from alternative remote sensing sources. These comparative results offer valuable insights into the reliability of CD-SAR data as a validation tool, more specifically how it stacks up against flood maps generated by other remote sensing techniques.