Similar Predictions Research Articles

Series2vec: similarity-based self-supervised representation learning for time series classification

We argue that time series analysis is fundamentally different in nature to either vision or natural language processing with respect to the forms of meaningful self-supervised learning tasks that can be defined. Motivated by this insight, we introduce a novel approach called Series2Vec for self-supervised representation learning. Unlike the state-of-the-art methods in time series which rely on hand-crafted data augmentation, Series2Vec is trained by predicting the similarity between two series in both temporal and spectral domains through a self-supervised task. By leveraging the similarity prediction task, which has inherent meaning for a wide range of time series analysis tasks, Series2Vec eliminates the need for hand-crafted data augmentation. To further enforce the network to learn similar representations for similar time series, we propose a novel approach that applies order-invariant attention to each representation within the batch during training. Our evaluation of Series2Vec on nine large real-world datasets, along with the UCR/UEA archive, shows enhanced performance compared to current state-of-the-art self-supervised techniques for time series. Additionally, our extensive experiments show that Series2Vec performs comparably with fully supervised training and offers high efficiency in datasets with limited-labeled data. Finally, we show that the fusion of Series2Vec with other representation learning models leads to enhanced performance for time series classification. Code and models are open-source at https://github.com/Navidfoumani/Series2Vec

Rock glaciers as proxy for machine learning based debris‐covered glacier mapping of Kinnaur District, Himachal Pradesh

AbstractThis research introduces an innovative approach by utilising rock glaciers (RGs) as a proxy for mapping debris‐covered glaciers (DCGs). This approach focuses on the interconnected nature of glaciers, DCGs and RGs in a continuum where DCGs can transform into RGs over time due to various processes. This study utilises six machine learning models—logistic regression (LR), support vector machine (SVM), K‐nearest neighbour (KNN), Naïve Bayes (NB), decision tree (DT) and random forest (RF)—combined with multispectral satellite data (Sentinel‐2 and Landsat 8) and topographical data derived from ALOS PALSAR DEM. Performance metrics such as accuracy, area under the curve (AUC) score, precision, recall and F1‐score were evaluated to assess model performance. This detailed mapping provides a precise estimation of the extent of DCGs in the Kinnaur district. The estimated DCG areas revealed intriguing variation across models, with RF (9.71%), KNN (9.67%) and NB (9.41%) yielding similar predictions. SVM (11.61%) projected a slightly larger DCG area, whereas DT (5.54%) and LR (25.55%) provided contrasting results. Validation against high‐resolution satellite images, Google Earth images and glacier inventories confirmed the accuracy and reliability of our approach. Based on our findings for our specific study, the most effective method for mapping DCGs is RF, followed by KNN, NB, DT and SVM. The combination of machine learning models and RG data presents a novel and promising approach to remote sensing‐based DCG mapping, with potential applications for other regions and broader environmental studies.

Multi-parameter similarity model-based lightweight design of battery electric vehicle body-in-white

The bending stiffness, torsional stiffness, and low-order modal frequencies of the body-in-white (BIW) are crucial factors in structural lightweight design. This paper introduces a BIW optimization approach for electric vehicles (BEVs) that integrates comprehensive sensitivity analysis and multi-parameter similarity models. The aim is to efficiently balance the BIW’s stiffness performance with minimal weight, while also managing potential manufacturing cost increases from design modifications. Firstly, the validity of the computational model is demonstrated by the BIW bending and torsion stiffness test and modal test. On this basis, the integrated sensitivity evaluation indexes of the structural member thickness parameters to the stiffness, weight and low-order modal frequency response of the BIW were calculated. The sheet metal thickness parameters of 17 structural members were preferred as design variables, and a multi-parameter similarity prediction model for BIW weight reduction design was constructed. Minimum weight, maximum stiffness and higher modal frequency of the BIW are defined as the optimization objectives. The optimal solution was finally solved using a multi-objective optimization algorithm. The results indicate a 1.582% reduction in weight of the BIW compared to the pre-optimization BIW. This paper delves into the intricate interplay between the parameters of the BIW’s structural components using simulation optimization techniques alongside experimental validation. This study not only proposes a new idea for the problem of how to efficiently and accurately determine the best optimization object from a large number of structural components in the development of vehicle structural safety, but also provides a research basis for automobile manufacturers to implement further weight reduction and optimization work on the BIW at the later stage of BEV development.

Spatiotemporal distribution and inter-media transfer of polycyclic aromatic hydrocarbons in Shanghai, China: Historical patterns and future trends

Polycyclic Aromatic Hydrocarbons (PAHs) represent pervasive pollutants, posing health risks in urban environments. It is essential to comprehend the spatiotemporal distributions, composition profiles, and inter-media transfer processes of PAHs in various environmental compartments, influenced by both natural changes and anthropogenic activities. This study integrates historical and future spatiotemporally changing environmental parameters, including climate data, GDP, population data, land-use types, and hydrological variables, into the Multimedia Urban Model (MUM). This integration enables the simulation of spatiotemporal distributions and inter-media transfer fluxes of PAHs among six different media from the 2010s to the 2100s under two distinct Shared Socio-economic Pathways (SSP) scenarios in the megacity of Shanghai, China. The MUM model, featuring diverse gridded parameters, effectively captures PAH concentrations and movement across environmental compartments. Results indicate a decreasing trend in PAHs concentrations in the 2100s compared to the 2010s, with PAH concentrations in water, sediment, vegetation, and organic film covering impermeable surfaces under the SSP3–7.0 scenario higher than those of the SSP1–2.6 scenario. Low molecular weight PAHs dominate in the sediment, water, and air, whereas high molecular weight PAHs prevail in the organic film, vegetation, and soil. Sediment and soil serve as the predominant sinks for PAHs. The primary transport processes for PAH movement include air-film, air-soil, film-water, soil-air, and water-air. Almost all transfer fluxes exhibit a declining trend in future periods except for the air-film transport pathway. The principal input and removal routes for PAHs in Shanghai involve the advection of air and water. The study provides essential insights into the environmental behavior of PAHs and informs targeted pollution control in Shanghai. Additionally, it serves as a technical reference for similar pollution prediction research.

Integrating intrinsic information: A novel open set domain adaptation network for cross-domain fault diagnosis with multiple unknown faults

Existing popular domain adaptation approaches typically assume that the source and target domains share the same label set. However, in industrial scenarios, the equipment may encounter unknown fault modes because of the harsh operating environment, which limits the application of existing diagnostic methods. To address the above problem, an intrinsic information-guided open set domain adaptation network is proposed for cross-domain fault diagnosis with unknown faults. First, a similarity-based discrimination framework is constructed to enhance the robustness of the model for unknown samples, which learns the similarity between samples and fault prototypes to enhance the classification performance. Then, a multi-information integrated weighting module is designed to quantify the transferability of samples through enhanced domain similarity information learning and prediction uncertainty information learning methods. Additionally, a self-supervised neighborhood clustering learning method is constructed, which enables the model to learn structural information about the target domain and encourages the target samples to cluster closely for better separability. Finally, the weighted open set adversarial training framework effectively facilitates diagnostic knowledge transfer and unknown fault recognition. Comprehensive experimental results on two datasets demonstrate the effectiveness of the proposed method in addressing the open set cross-domain diagnosis problem, which achieves promising performance over the comparison methods.

Code Similarity Prediction Model for Industrial Management Features Based on Graph Neural Networks.

The code of industrial management software typically features few system API calls and a high number of customized variables and structures. This makes the similarity of such codes difficult to compute using text features or traditional neural network methods. In this paper, we propose an FSPS-GNN model, which is based on graph neural networks (GNNs), to address this problem. The model categorizes code features into two types, outer graph and inner graph, and conducts training and prediction with four stages-feature embedding, feature enhancement, feature fusion, and similarity prediction. Moreover, differently structured GNNs were used in the embedding and enhancement stages, respectively, to increase the interaction of code features. Experiments with code from three open-source projects demonstrate that the model achieves an average precision of 87.57% and an F0.5 Score of 89.12%. Compared to existing similarity-computation models based on GNNs, this model exhibits a Mean Squared Error (MSE) that is approximately 0.0041 to 0.0266 lower and an F0.5 Score that is 3.3259% to 6.4392% higher. It broadens the application scope of GNNs and offers additional insights for the study of code-similarity issues.

Improving river water quality prediction with hybrid machine learning and temporal analysis

River systems provide multiple ecosystem services to society globally, but these are already degraded or threatened in many areas of the world due to water quality issues linked to diffuse and point-source pollutant inputs. Water quality evaluation is essential to develop remediation and management strategies. Computational tools such as machine learning based predictive models have been developed to improve monitoring network capabilities. The model's performance is reduced when datasets composed of reductant information are used for training, on the other hand, the selection of most representative and variable water quality scenarios could result in higher precision. This study analyzed historical water quality behavior in the Santiago River, Mexico, to identify the most variable and representative data available to train machine learning models (Adaptive Neuro Fuzzy Inference System – ANFIS, Artificial Neural Network – ANN, and Support Vector Machine - SVM). Thirteen monitoring sites were clustered according to their water quality variability from 2009 to 2022. Subsequently, a Time Series Analysis (TSA) was used to select the most representative monitoring station from each cluster. Data for 6/13 monitoring sites were retained for the Best Training Subset (BTS) used to train restricted models that performed with similar (ANN and SMV) or higher (ANFIS) prediction accuracy (in terms of RMSE, MAE, MSE and R2) for both training and testing. This study provides evidence of water quality data containing redundant information that is not useful to improve machine learning model performance, in turn leading to overtraining. Combined analytical approaches can maximize the representativeness and variability of data selected for machine learning applications, leading to improved prediction.

Fluid-to-fluid similarity and CFD predictions of surrogate fluid as a replacement for R11 refrigerant used in a subcooled flow boiling analysis: Validation against reference experimental data

In this study, we present the possibility of substituting the primary fluid Freon 11 (CCl3F, R11) used in the experimental thermohydraulic test facility which offers the advantage to obtain phase change at relatively low heat fluxes and pressure if compared to water working fluid. Nevertheless, R11 is a CFC (Chlorofluorocarbures) part of the gases that contribute to the degradation of the ozone layer.For this purpose, four working fluids (non-CFC) are judiciously selected to replace the R11, namely the R123, R134a, R143a and CO2, and this through multidimensional Computational Fluid Dynamics (CFD) modeling of steady state flow boiling phenomenon. The calculation model covers a wide range of mass fluxes (435 – 1443 kg/m2s) and heat fluxes (1.2– 36 W/cm2); it predicts the heat transfer rates and transitions points for each flow regime. In this research, the experiment's data were compared to the outcomes of Ansys-CFX computations, using a steady Reynolds-Averaged Navier-Stokes (Steady-RANS) model, to investigate the best replacement refrigerant for R11. The turbulence effects on the mean flow and near wall are taken into consideration using the Shear Stress Transport (SST k-ω) turbulence model. The results from the four fluids were transposed to their water equivalent and compared to those deduced from the experimentation and correlations available in the literature to select the appropriate substitution fluid. The results qualitative and quantitative for vertical and horizontal subcooled boiling flow at the CHF location show that, the most effective surrogate fluid is R123 (HCFC), is still widely used in the refrigeration and air-conditioning industry, then R134a (HFC), then R143a (HFC).

Prediction of immune-related adverse events in urological cancer during checkpoint inhibitor immunotherapy through immunohistochemical analysis of tumorous LCP1/ADPGK.

e14699 Background: The expanded utilization of immune checkpoint inhibitors (ICIs) have successfully improved the outcome of cancer patients, yet the concomitant immune-related adverse events (irAEs) can lead to fulminant unfavorable consequences. Currently, a comprehensive approach to predict irAEs is lacking. Some relevant biomarkers derived from database analyses are still in the early stages of investigation and lack robust validation in clinical samples, including LCP1/ADPGK. Herein, we validated the performance of LCP1 and ADPGK in a retrospective patient cohort with advanced urological cancers undergoing ICIs treatment, while comparing the predictive efficacy between different algorithmic models. Methods: We retrospectively collected clinical data of patients with renal cell carcinoma (RCC), upper urinary tract epithelial carcinoma (UTUC), bladder cancer (BC) receiving anti-PD-1/PD-L1 treatment from 2020 to 2023 in Peking University First Hospital. Patients were categorized into irAEs and non-irAEs groups, with irAE assessments conducted by two expert urological oncologists. Immunohistochemistry (IHC) was performed on 5-µm-thick pre-treated FFPE tumor tissue sections. Results: A total of 112 patients were included in the study, comprising 60 with RCC, 21 with UTUC, and 31 with BC. Among them, 51 (45.5%) patients experienced irAEs, with 9 exhibiting severe irAEs of grade III-IV. Immunohistochemical analysis revealed higher expression levels of LCP1 and ADPGK in patients with irAEs compared to those without irAEs. The area under the receiver-operating characteristic curve (AUC) for predicting irAEs using LCP1 and ADPGK individually was 0.82 (p < 0.0001, 95%CI = 0.74-0.90) and 0.86 (p < 0.0001, 95%CI = 0.80-0.93), respectively. The combination of LCP1 and ADPGK demonstrated an improved AUC of 0.89 (p < 0.0001, 95%CI = 0.83-0.95). Among different algorithms, the bivariate linear-regression and the geometric mean model show similar prediction effects. Notably, the combined LCP1+ADPGK model successfully predicted irAEs in RCC (AUC = 0.89, p < 0.0001, 95%CI = 0.81-0.98), UTUC (AUC = 0.81, p = 0.0063, 95%CI = 0.63-0.99), and BC (AUC = 0.88, p = 0.0004, 95%CI = 0.75-1.00). The expression of LCP1/ADPGK was not associated with the type of cancers. Conclusions: The bivariate linear-regression model of LCP1 and ADPGK could accurately predict irAEs in patients with urological cancers, which may allow us to improve the risk-benefit balance for individuals considered for ICIs therapy.

Evaluating the contribution of methanotrophy kinetics to uncertainty in the soil methane sink

The oxidation of atmospheric methane by soil microbes is an important natural sink for a potent greenhouse gas. However, estimates of the current and future soil methane sink are highly uncertain. Here we assessed the extent to which methanotrophy enzyme kinetics contribute to uncertainty in projections of the soil methane sink. We generated a comprehensive compilation of methanotrophy kinetic data from modern environments and assessed the patterns in kinetic parameters present in natural samples. Our compiled data enabled us to quantify the global soil methane sink through two idealized calculations comparing first-order and Michaelis–Menten models of kinetics. We show that these two kinetic models diverge only under high atmospheric CH4 scenarios, where first-order rate constants slightly overestimate the soil methane sink size, but produce similar predictions at modern atmospheric concentrations. Our compilation also shows that the kinetics of methanotrophy in natural soil samples is highly variable—both the V max (oxidation rate at saturation) and KM (half-saturation constant) in natural samples span over six orders of magnitude. However, accounting for the correlation we observe between V max and KM reduces the range of calculated uptake rates by as much as 96%. Additionally, our results indicate that variation in enzyme kinetics introduces a similar magnitude of variation in the calculated soil methane sink as temperature sensitivity. Systematic sampling of methanotroph kinetic parameters at multiple spatial scales should therefore be a key objective for closing the budget on the global soil methane sink.

Improving Molecule-Metal Surface Reaction Networks Using the Meta-Generalized Gradient Approximation: CO2 Hydrogenation.

Density functional theory is widely used to gain insights into molecule-metal surface reaction networks, which is important for a better understanding of catalysis. However, it is well-known that generalized gradient approximation (GGA) density functionals (DFs), most often used for the study of reaction networks, struggle to correctly describe both gas-phase molecules and metal surfaces. Also, GGA DFs typically underestimate reaction barriers due to an underestimation of the self-interaction energy. Screened hybrid GGA DFs have been shown to reduce this problem but are currently intractable for wide usage. In this work, we use a more affordable meta-GGA (mGGA) DF in combination with a nonlocal correlation DF for the first time to study and gain new insights into a catalytically important surface reaction network, namely, CO2 hydrogenation on Cu. We show that the mGGA DF used, namely, rMS-RPBEl-rVV10, outperforms typical GGA DFs by providing similar or better predictions for metals and molecules, as well as molecule-metal surface adsorption and activation energies. Hence, it is a better choice for constructing molecule-metal surface reaction networks.

Validation of the CREST model and comparison with SCAI shock classification for the prediction of circulatory death in resuscitated out-of-hospital cardiac arrest.

We validated the CREST model, a 5 variable score for stratifying the risk of circulatory aetiology death (CED) following out-of-hospital cardiac arrest (OHCA) and compared its discrimination with the SCAI shock classification. Circulatory aetiology death occurs in approximately a third of patients admitted after resuscitated OHCA. There is an urgent need for improved stratification of the patient with OHCA on arrival to a cardiac arrest centre to improve patient selection for invasive interventions. The CREST model and SCAI shock classification were applied to a dual-centre registry of 723 patients with cardiac aetiology OHCA, both with and without ST-elevation myocardial infarction (STEMI), between May 2012 and December 2020. The primary endpoint was a 30-day CED. Of 509 patients included (62.3 years, 75.4% male), 125 patients had CREST = 0 (24.5%), 162 had CREST = 1 (31.8%), 140 had CREST = 2 (27.5%), 75 had CREST = 3 (14.7%), 7 had a CREST of 4 (1.4%), and no patients had CREST = 5. Circulatory aetiology death was observed in 91 (17.9%) patients at 30 days [STEMI: 51/289 (17.6%); non-STEMI (NSTEMI): 40/220 (18.2%)]. For the total population, and both NSTEMI and STEMI subpopulations, an increasing CREST score was associated with increasing CED (all P < 0.001). The CREST score and SCAI classification had similar discrimination for the total population [area under the receiver operating curve (AUC) = 0.72/calibration slope = 0.95], NSTEMI cohort (AUC = 0.75/calibration slope = 0.940), and STEMI cohort (AUC = 0.69 and calibration slope = 0.925). Area under the receiver operating curve meta-analyses demonstrated no significant differences between the two classifications. The CREST model and SCAI shock classification show similar prediction results for the development of CED after OHCA.

Leveraging a Small Dataset to Predict Nonlinear Global Loads in Irregular Waves

In this work, a hybrid machine learning method, which uses ML strategies to model high-order force components within a low-order equation of motion, is considered in the context of the global wave-induced loads of a ship in irregular waves. It is shown that the method can make predictions in a range of wave conditions even when the training data set only includes a single seaway. The proposed method offers a data-leveraging technique which may be useful in the design space, where a small data set derived from a high-fidelity source can be leveraged to make similar fidelity predictions in a larger number of wave conditions.

Fluid and kinetic studies of tokamak disruptions using Bayesian optimization

When simulating runaway electron dynamics in tokamak disruptions, fluid models with lower numerical cost are often preferred to more accurate kinetic models. The aim of this work is to compare fluid and kinetic simulations of a large variety of different disruption scenarios in ITER. We consider both non-activated and activated scenarios; for the latter, we derive and implement kinetic sources for the Compton scattering and tritium beta decay runaway electron generation mechanisms in our simulation tool Dream (Hoppe et al., Comput. Phys. Commun., vol. 268, 2021, 108098). To achieve a diverse set of disruption scenarios, Bayesian optimization is used to explore a range of massive material injection densities for deuterium and neon. The cost function is designed to distinguish between successful and unsuccessful disruption mitigation based on the runaway current, current quench time and transported fraction of the heat loss. In the non-activated scenarios, we find that fluid and kinetic disruption simulations can have significantly different runaway electron dynamics, due to an overestimation of the runaway seed by the fluid model. The primary cause of this is that the fluid hot-tail generation model neglects superthermal electron transport losses during the thermal quench. In the activated scenarios, the fluid and kinetic models give similar predictions, which can be explained by the significant influence of the activated sources on the runaway dynamics and the seed.

Estimating lags in a kraft mill.

In pulp mills, lags obscure the effect of upstream operations on downstream measurements. Here, we estimate lags in a Canadian pulp mill using autoregressive exogenous (ARX) models. First, we show that ARX models can approximate lags in a process simulation that resembles the liquor storage tanks in pulp mills, a major source of lag inthe kraft recovery cycle. Then, we use ARX models to estimate the lagged effect of a change in species pulped on as-fired liquor heating value, viscosity, and boiling point rise. Additionally, we compare the predictions of the ARX models to autoregressive (AR) models and a persistence model. The estimated lags between a change in species and heating value (49 h) and boiling point rise (41 h) agree with a detailed simulation of the mill and are close to estimated hydraulic residence times, suggesting that the liquor tanks exhibit imperfect mixing. A lagged effect of species change on viscosity could not be identified. ARX and AR models produce similar predictions that are slightly better than those of a persistence model. Finally, we show that process measurements upstream of units characterized by large residence times will likely provide little benefit to prediction accuracy.

Predictive Value of Inflammatory Markers NLR, PLR, APRI, SII, and Liver Function Tests in Systemic Inflammatory Response Syndrome Detection in Full-Term Newborns.

Systemic Inflammatory Response Syndrome (SIRS) is associated with significant morbidity and mortality in full-term newborns. This study aimed to evaluate the predictive value of the Neutrophil-to-Lymphocyte Ratio (NLR), Derived Neutrophil-to-Lymphocyte Ratio (dNLR), Platelet-to-Lymphocyte Ratio (PLR), Neutrophil, Lymphocyte, and Platelet Ratio (NLPR), AST-to-Platelet Ratio Index (APRI), and Systemic Immune-Inflammation Index (SII) in identifying the risk for SIRS development in full-term newborns. Conducted between January 2023 and January 2024, this observational cohort study compared full-term newborns diagnosed with SIRS with newborns without SIRS, measuring the inflammatory markers within the first day of life and three days post-birth. The study included 229 newborns, 81 with SIRS and 148 controls without SIRS. Statistically significant differences were observed in NLR (3.81 vs. 2.20, p < 0.0001), PLR (68.12 vs. 52.30, p < 0.0001), and liver enzymes (AST 40.96 U/L vs. 31.58 U/L, ALT 34.66 U/L vs. 22.46 U/L, both p < 0.0001) between the groups. The NLPR demonstrated substantial diagnostic value, with a sensitivity of 78.36% and specificity of 83.52% at 72 h (p < 0.0001). Regression analysis highlighted that the NLPR and SII were strongly predictive of SIRS, with the NLPR showing over three-times higher SIRS risk (HR 3.29, p < 0.0001) and SII indicating nearly 3.5 times the risk (HR 3.47, p < 0.0001). The NLPR, APRI, and SII showed similar prediction values to CRP levels measured on the first and third days of life (HR 3.16). Inflammatory markers like NLR, PLR, and systemic indices such as NLPR and SII, alongside liver function tests, are significant predictors of SIRS in full-term newborns. These findings support the integration of these markers into routine neonatal care, allowing for early identification and potentially improved management of newborns at risk for SIRS, thereby enhancing clinical outcomes.

Contrastive BiLSTM-enabled Health Representation Learning for Remaining Useful Life Prediction

Remaining useful life (RUL) prediction is of vital significance in prognostics health management tasks. Due to powerful learning capabilities, deep learning methods, particularly long short-term memory (LSTM) have been widely applied in RUL prediction. However, many existing deep learning approaches overlook the inherent ordered relationship between samples in the direct mapping from sliced data to RUL pattern. To capture the faithful and ordered health representation of a given system, a Contrastive Bidirectional LSTM-enabled Health Representation Learning (CBHRL) framework is proposed. Firstly, the supervised contrastive regression loss (SupCR) is implemented to extract continuous health representation. The SupCR is designed to rank the similarity among health representations from different samples, prompting them highly correlated with linear RUL label. Among the process of contrastive learning, the series odd-even decomposition (SOED) method is devised to construct multi-view degradation data, which improves generalization ability. Finally, since the health representation is constructed on basis of similarity, a new similarity prediction method is proposed as the complement of regression prediction method. Experimental results show the health representations extracted by CBHRL achieve improved ratio ranging from a minimum of 17.19% to a maximum of 291.30% in monotonicity, smoothness and trendability.

Comparison of Two Methods, Gradient Boosting and Extreme Gradient Boosting to Pre- dict Survival in Covid-19 Data

Introduction: The present study discusses the importance of having a predictive method to determine the prognosis of patients with diseases like Covid-19. This method can assist physicians in making treatment decisions that improve survival rates and avoid unnecessary treatments. This research also highlights the importance of calibration, which is often overlooked in model evaluation. Without proper calibration, incorrect decisions can be made in disease treatment and preventive care. Therefore, the current study compares two highly accurate machine learning algorithms, Gradient boosting and Extreme gradient boosting, not only in terms of prediction accuracy but also in terms of model calibration and speed. Methods: This study involved analyzing data from Covid-19 patients who were admitted to two hospitals in Mashhad city, Razavi Khorasan province, over a span of 18 months. The k-fold cross-validation method was employed on the training dataset (K=5) to conduct the study. The accuracy and calibration of two methods (Gradient boosting and Extreme gradient boosting) in predicting survival were compared using the Concordance Index and calibration. Results: The Concordance Index values obtained for gradient boosting and Extreme gradient boosting models were 0.734 and 0.736, in the imbalanced and In the balanced data, the Concordance Index values were 0.893 for gradient boosting and 0.894 for Extreme gradient boosting. The surv.calib_beta index, the gradient boosting model had an estimated value of 0.59 in the imbalanced data and 0.66 in the balanced data. The Extreme gradient boosting model had an estimated value of 0.86 in the balanced data and 0.853 in the imbalanced data. The Extreme gradient boosting model was faster in the learning process compared to the gradient boosting model. Conclusion: The Gradient boosting and Extreme gradient boosting models exhibited similar prediction accuracy and discrimination power, but the Extreme gradient boosting model demonstrated relatively good calibration compare to Gradient boosting model.

Association of Life’s Essential 8 and Simple 7 Scores With Mortality: Comparison With Pooled Cohort Equation

BackgroundIn 2022, the Life’s Simple 7 (LS7) score was replaced with the Life’s Essential 8 (LE8) score as a tool to measure cardiovascular health. The risk prediction values of LE8 and LS7 scores for mortality have not been compared. Additionally, the risk prediction value of these scores has not been compared with the pooled cohort equations (PCE) in individuals aged 40 to 79 years. ObjectivesThis study compared the risk prediction value of the: 1) LE8 and LS7 scores in the overall population; and 2) LE8 score, LS7 score, and PCE in the 40- to 79-year-old age group for all-cause and cardiovascular mortality in a nationally representative US population. MethodsThe LS7 and LE8 scores and the PCE were calculated in the National Health and Nutrition Examination Survey cycles 2007 to 2018. All-cause and cardiovascular mortality were identified by linking the participants to the National Death Index. The C-statistics of the respective weighted Cox models were used to compare the risk prediction value of the standardized scores. ResultsAmong 21,721 individuals included, the C-statistics for all-cause mortality were 0.823 (95% CI: 0.803-0.843) and 0.819 (95% CI: 0.799-0.838) in the LE8 and LS7 score-based models, respectively. The C-statistics for cardiovascular mortality were 0.887 (95% CI: 0.857-0.905) for the LE8 score-based model and 0.883 (95% CI: 0.861-0.905) for the LS7 score-based model. Among 12,943 individuals aged 40 to 79 years, the C-statistics for the outcome of all-cause mortality were 0.756 (95% CI: 0.732-0.779), 0.674 (95% CI: 0.646-0.701), and 0.681 (95% CI: 0.656-0.706) for the PCE, LS7 score, and LE8 score-based models, respectively. ConclusionsThe LS7 and LE8 scores had similar risk prediction values for all-cause and cardiovascular mortality. Among 40- to 79-year-old individuals, the PCE had better risk discrimination in the LE8 and LS7 scores in predicting all-cause mortality.

Femtosecond laser-assisted in situ keratomileusis versus small-incision lenticule extraction: current approach based on evidence.

Laser keratorefractive surgery achieves excellent visual outcomes for refractive error correction. With femtosecond laser, small incision lenticule extraction (SMILE) is an increasingly viable alternative to laser-assisted in situ keratomileusis (LASIK). Comparative studies demonstrate similar efficacy and predictability between SMILE and LASIK, making it difficult for clinicians to choose which to use. This review thus compares femtosecond-LASIK (FS-LASK) and SMILE in various scenarios, to assist clinicians in deciding which refractive surgery procedure to recommend. SMILE may be superior for highly myopic eyes due to a smaller decrease in functional optical zone. SMILE further induces less spherical aberration and less overall higher order aberrations in mesopic conditions. SMIILE also has less postoperative dry eye, making it suitable those with preexisting dry eye. For low to moderate myopic astigmatism correction, FS-LASIK has less undercorrection compared to SMILE. Lastly, SMILE has not yet received Food and Drug Administration or Conformité Européenne approval for hyperopic correction, rendering FS-LASIK the choice of procedure for hyperopic correction. Both FS-LASIK and SMILE demonstrate good efficacy and predictability. Understanding specific clinical scenarios where one may be superior to the other will aid clinicians in choosing the most suitable procedure for personalized care.