Loss Metrics Research Articles

An Analysis of Loss Functions for Heavily Imbalanced Lesion Segmentation.

Heavily imbalanced datasets are common in lesion segmentation. Specifically, the lesions usually comprise less than 5% of the whole image volume when dealing with brain MRI. A common solution when training with a limited dataset is the use of specific loss functions that rebalance the effect of background and foreground voxels. These approaches are usually evaluated running a single cross-validation split without taking into account other possible random aspects that might affect the true improvement of the final metric (i.e., random weight initialisation or random shuffling). Furthermore, the evolution of the effect of the loss on the heavily imbalanced class is usually not analysed during the training phase. In this work, we present an analysis of different common loss metrics during training on public datasets dealing with brain lesion segmentation in heavy imbalanced datasets. In order to limit the effect of hyperparameter tuning and architecture, we chose a 3D Unet architecture due to its ability to provide good performance on different segmentation applications. We evaluated this framework on two public datasets and we observed that weighted losses have a similar performance on average, even though heavily weighting the gradient of the foreground class gives better performance in terms of true positive segmentation.

Emotion Detection from Photos Using MobleNet-based Deep Learning

In the era of twenty-first century, an era characterized by the proliferation of digital technology, big data and so on, the ability to identify human emotions through visual content from images has gained much importance and its popularity is increasing worldwide. This project deals with the task of detecting emotions from images using deep learning techniques with a specific emphasis on Mobile Net-based architectures. We start the project by preparing the dataset of various images showing diverse emotions. The Mobile Net architecture, a powerful convolutional neural network is fine-tuned with a custom dense layer to classify emotions into seven distinct categories. Data argumentation techniques such as zooming, shearing and horizontal flipping are incorporated to enhance robustness and prevent overfitting. The training dataset is preprocessed and normalized while a segregated validation dataset ensures stringent evaluation. During training we implemented early stopping and model checkpoint mechanisms to get optimal performance while avoiding overfitting. After training the analysis of accuracy and loss metrics provides an insight into the model’s trajectory. In practical applicability we use the trained model to predict emotion from single images, showcasing its potential in various domains, including digital marketing, healthcare, and user experience design. In today’s digital landscape the project findings hold relevance for a wide spectrum of applications, promising advancements in human computer interactions and emotion aware systems.

Evaluation of a multi-function in-ear device performance in the presence of impulse noise using acoustic test fixtures

Real-time noise exposure monitoring has the potential to assess risk before damage occurs. We present on the development of an in- ear device that (1) provides hearing protection; (2) measures the noise in the ear canal to estimate exposure dose; and (3) monitors distortion product otoacoustic emissions (DPOAE) over time. To validate the device’s ability to measure impulse noise under hearing protection, we measured the response in a field test. A single AR15 was shot multiple times, and four prototypes were used in multiple acoustic test fixtures (ATFs) at different relative distances from the exit of the gun. The ATFs included a G.R.A.S 45 CB in addition to a head simulator designed to detect both air- and bone-conducted sound. We evaluated the impulse peak insertion loss (IPIL) metric of the device inserted by itself or in combination with other hearing protection (“double protection” condition) to assess attenuation of impulsive noises. We also compared the device's microphone measurements to those obtained through the ATF microphones, and to the recordings of external microphones located at or near the ear of each ATF. Results demonstrate the value of testing in-ear devices in acoustic test fixtures during development phases.

Investigating the sensitivity of losses to time-dependent components of seismic risk modeling

Conventional earthquake risk modeling involves several notable simplifications, which neglect: (1) the effects on seismicity of interactions between adjacent faults and the long-term elastic rebound behavior of faults; (2) short-term hazard increases associated with aftershocks; and (3) the accumulation of damage in assets due to the occurrence of multiple earthquakes in a short time window, without repairs. Several recent earthquake events (e.g. 2010–2011 Canterbury earthquakes, New Zealand; 2019 Ridgecrest earthquakes, USA; and 2023 Turkey–Syria earthquakes) have emphasized the need for risk models to account for the aforementioned short- and long-term time-dependent characteristics of earthquake risk. This study specifically investigates the sensitivity of monetary loss (i.e. a possible earthquake-risk-model output) to these time dependencies, for a case-study portfolio in Central Italy. The investigation is intended to provide important insights for the catastrophe risk insurance and reinsurance industry. In addition to salient catastrophe risk insurance features, the end-to-end approach for time-dependent earthquake risk modeling used in this study incorporates recent updates in long-term time-dependent fault modeling, aftershock forecasting, and vulnerability modeling that accounts for damage accumulation. The sensitivity analysis approach presented may provide valuable guidance on the importance and appropriate treatment of time dependencies in regional (i.e. portfolio) earthquake risk models. We find that the long-term fault and aftershock occurrence models are the most crucial features of a time-dependent seismic risk model to constrain, at least for the monetary loss metrics examined in this study. Accounting for damage accumulation is also found to be important, if there is a high insurance deductible associated with portfolio assets.

Selecting and Interpreting Multiclass Loss and Accuracy Assessment Metrics for Classifications with Class Imbalance: Guidance and Best Practices

Evaluating classification accuracy is a key component of the training and validation stages of thematic map production, and the choice of metric has profound implications for both the success of the training process and the reliability of the final accuracy assessment. We explore key considerations in selecting and interpreting loss and assessment metrics in the context of data imbalance, which arises when the classes have unequal proportions within the dataset or landscape being mapped. The challenges involved in calculating single, integrated measures that summarize classification success, especially for datasets with considerable data imbalance, have led to much confusion in the literature. This confusion arises from a range of issues, including a lack of clarity over the redundancy of some accuracy measures, the importance of calculating final accuracy from population-based statistics, the effects of class imbalance on accuracy statistics, and the differing roles of accuracy measures when used for training and final evaluation. In order to characterize classification success at the class level, users typically generate averages from the class-based measures. These averages are sometimes generated at the macro-level, by taking averages of the individual-class statistics, or at the micro-level, by aggregating values within a confusion matrix, and then, calculating the statistic. We show that the micro-averaged producer’s accuracy (recall), user’s accuracy (precision), and F1-score, as well as weighted macro-averaged statistics where the class prevalences are used as weights, are all equivalent to each other and to the overall accuracy, and thus, are redundant and should be avoided. Our experiment, using a variety of loss metrics for training, suggests that the choice of loss metric is not as complex as it might appear to be, despite the range of choices available, which include cross-entropy (CE), weighted CE, and micro- and macro-Dice. The highest, or close to highest, accuracies in our experiments were obtained by using CE loss for models trained with balanced data, and for models trained with imbalanced data, the highest accuracies were obtained by using weighted CE loss. We recommend that, since weighted CE loss used with balanced training is equivalent to CE, weighted CE loss is a good all-round choice. Although Dice loss is commonly suggested as an alternative to CE loss when classes are imbalanced, micro-averaged Dice is similar to overall accuracy, and thus, is particularly poor for training with imbalanced data. Furthermore, although macro-Dice resulted in models with high accuracy when the training used balanced data, when the training used imbalanced data, the accuracies were lower than for weighted CE. In summary, the significance of this paper lies in its provision of readers with an overview of accuracy and loss metric terminology, insight regarding the redundancy of some measures, and guidance regarding best practices.

Analysis of Investors’ Choices in Technology Companies

This paper examines how to analyze which technology companies are suitable for different types of investors through various financial indicators. First, this paper chooses four technology companies, namely Tesla, Tencent, Microsoft and Apple. Next, Yahoo searches for various profit and loss metrics for several companies and groups these metrics into three categories, including risk, profitability and market ratio. Finally, reasonable suggestions are given by analyzing the investment preferences of nine kinds of investors. The nine types of Investors are price/earnings to growth ratio investors (PEG investors), discounted cash flow Investors (DCF investors), income investors, value investors, index investors, momentum investors, insider investors, ratio analysis and stock buyer. According to the analysis results, PEG investors and DCF investors may choose Tesla, while DCF investors and income investors may choose Tencent. In addition to index investors, other investors are likely to choose Microsoft. Apple is likely to be picked by PEG investors, value investors and momentum investors. The research significance of this paper is to help investors who lack certain financial knowledge to choose their own types of stocks, and to describe the characteristics of different types of investors to help investors judge which investment method is suitable for them.

A Deep Learning Approach for the Automated Classification of Geomagnetically Induced Current Scalograms

During geomagnetic storms, which are a result of solar wind’s interaction with the Earth’s magnetosphere, geomagnetically induced currents (GICs) begin to flow in the long, high-voltage electrical networks on the Earth’s surface. It causes a number of negative phenomena that affect the normal operation of the entire electric power system. To investigate the nature of the phenomenon and its effects on transformers, a GIC monitoring system was created in 2011. The system consists of devices that are installed in the neutrals of autotransformers at five substations of the Kola–Karelian power transit in northwestern Russia. Considering the significant amount of data accumulated over 12 years of operating the GIC monitoring system, manual analysis becomes impractical. To analyze the constantly growing volume of recorded data effectively, a method for the automatic classification of GICs in autotransformer neutrals was proposed. The method is based on a continuous wavelet transform of the neutral current data combined with a convolutional neural network (CNN) to classify the obtained scalogram images. The classifier’s performance is evaluated using accuracy and binary cross-entropy loss metrics. As the result of comparing four CNN architectures, a model that showed high GIC classification performance on the validation set was chosen as the final model. The proposed CNN model, in addition to the main layers, includes pre-processing layers and a dropout layer.

An evaluation of machine learning approaches for early diagnosis of autism spectrum disorder

Autistic Spectrum Disorder (ASD) is a neurological disease characterized by difficulties with social interaction, communication, and repetitive activities. While its primary origin lies in genetics, early detection is crucial, and leveraging machine learning offers a promising avenue for a faster and more cost-effective diagnosis. This study employs diverse machine learning methods to identify crucial ASD traits, aiming to enhance and automate the diagnostic process. We study eight state-of-the-art classification models to determine their effectiveness in ASD detection. We evaluate the models using accuracy, precision, recall, specificity, F1-score, area under the curve (AUC), kappa, and log loss metrics to find the best classifier for these binary datasets. Among all the classification models, for the children dataset, the SVM and LR models achieve the highest accuracy of 100% and for the adult dataset, the LR model produces the highest accuracy of 97.14%. Our proposed ANN model provides the highest accuracy of 94.24% for the new combined dataset when hyperparameters are precisely tuned for each model. As almost all classification models achieve high accuracy which utilize true labels, we become interested in delving into five popular clustering algorithms to understand model behavior in scenarios without true labels. We calculate Normalized Mutual Information (NMI), Adjusted Rand Index (ARI), and Silhouette Coefficient (SC) metrics to select the best clustering models. Our evaluation finds that spectral clustering outperforms all other benchmarking clustering models in terms of NMI and ARI metrics while demonstrating comparability to the optimal SC achieved by k-means. The implemented code is available at GitHub.

Multi-scale and multi-refinement framework for seismic risk assessment of urban areas

As a part of a wider research project, this paper aims to develop an innovative framework for seismic risk assessment of urban areas. The investigated procedure employs a multi-scale approach, from individual elements (e.g., single buildings, water pipelines), to various clusters and interconnected layers and, finally, the entire urban area. The last one is represented by a multi-layered model, including the “building stock” and the utilities layers (e.g. water distribution networks). Moreover, a key feature relies on a multi-refinement level framework, which allows for the evaluation of loss metrics even in the case of limited knowledge. Alternative refinement levels involve: i) typological-based vulnerability assessment; ii) analytical/mechanical procedures, and iii) numerical (software-based) simulations. The uncertainties for each refinement level are assessed and propagated. The paper also discusses preliminary results for the building stock and water utilities. Implementing such a framework in a City Digital Twin environment can enhance the awareness of stakeholders and the general population towards seismic risk prompting decision-makers in taking action.

Research on incentive mechanisms for anti-heterogeneous federated learning based on reputation and contribution

<abstract> <p>An optimization algorithm for federated learning, equipped with an incentive mechanism, is introduced to tackle the challenges of excessive iterations, prolonged training durations, and suboptimal efficiency encountered during model training within the federated learning framework. Initially, the algorithm establishes reputation values that are tied to both time and model loss metrics. This foundation enables the creation of incentive mechanisms aimed at rewarding honest nodes while penalizing malicious ones. Subsequently, a bidirectional selection mechanism anchored in blockchain technology is developed, allowing smart contracts to enroll nodes with high reputations in training sessions, thus filtering out malicious clients and enhancing local training efficiency. Furthermore, the integration of the Earth Mover's Distance (EMD) mechanism serves to lessen the impact of non-IID (non-Independent and Identically Distributed) data on the global model, leading to a reduction in the frequency of model training cycles and an improvement in model accuracy. Experimental results confirm that this approach maintains high model accuracy in non-IID data settings, outperforming traditional federated learning algorithms.</p> </abstract>

Application of Machine Learning in Estimating Milk Yield According to the Phenotypic and Pedigree Data of Holstein-Friesian Cattle in Serbia

Summary This paper presents a deep neural network (DNN) approach designed to estimate the milk yield of Holstein-Friesian cattle. The DNN comprised stacked dense (fully connected) layers, each hidden layer followed by a dropout layer. Various configurations of the DNN were tested, incorporating 2 and 3 hidden layers containing 8 to 54 neurons. The experiment involved testing the DNN with different activation functions such as the sigmoid, tanh, and rectified linear unit (ReLU). The dropout rates ranging from 0 to 0.3 were employed, with the output layer using a linear activation function. The DNN models were trained using the Adam, SGD, and RMSprop optimizers, with the root mean square error serving as the loss metric. The training dataset comprised information from a unique database containing records of dairy cows in the Republic of Serbia, totaling 3,406 cows. The input parameters (a total of 27) for the DNN included breeding and milk yield data from the cow’s mother, as well as the father’s ID, whereas the output parameters (a total of 8) consisted of milk yield parameters (a total of 3) and breeding parameters of the cow (a total of 5). Training iterations were conducted using a batch size of 8 over 500, 1000, and 5000 epochs.

A Study on The Improvement of Information Loss Metrics in Real-Time Stream Data Anonymization

A Study on The Improvement of Information Loss Metrics in Real-Time Stream Data Anonymization

Fully Automated Analysis of Muscle Architecture from B-Mode Ultrasound Images with DL_Track_US

B-mode ultrasound can be used to image musculoskeletal tissues, but one major bottleneck is analyses of muscle architectural parameters (i.e., muscle thickness, pennation angle and fascicle length), which are most often performed manually. In this study we trained two different neural networks (classic U-Net and U-Net with VGG16 pre-trained encoder) to detect muscle fascicles and aponeuroses using a set of labeled musculoskeletal ultrasound images. We determined the best-performing model based on intersection over union and loss metrics. We then compared neural network predictions on an unseen test set with those obtained via manual analysis and two existing semi/automated analysis approaches (simple muscle architecture analysis [SMA] and UltraTrack). DL_Track_US detects the locations of the superficial and deep aponeuroses, as well as multiple fascicle fragments per image. For single images, DL_Track_US yielded results similar to those produced by a non-trainable automated method (SMA; mean difference in fascicle length: 5.1 mm) and human manual analysis (mean difference: -2.4 mm). Between-method differences in pennation angle were within 1.5°, and mean differences in muscle thickness were less than 1 mm. Similarly, for videos, there was overlap between the results produced with UltraTrack and DL_Track_US, with intraclass correlations ranging between 0.19 and 0.88. DL_Track_US is fully automated and open source and can estimate fascicle length, pennation angle and muscle thickness from single images or videos, as well as from multiple superficial muscles. We also provide a user interface and all necessary code and training data for custom model development.

Surpassing early stopping: A novel correlation-based stopping criterion for neural networks

During the training of neural networks, selecting the right stopping criterion is crucial to prevent overfitting and conserve computing power. While the early stopping and the maximum number of epochs stopping methods are simple to implement, they have limitations in identifying the point during training where the training and the validation loss start to diverge. To overcome these limitations, we propose a general correlation-based stopping criterion called the Correlation-Driven Stopping Criterion (CDSC). The CDSC stops the training process when the rolling Pearson correlation of the loss metrics between the training and validation datasets decreases below a pre-defined threshold. To show the effectiveness of the newly proposed Correlation-Driven Stopping Criterion, its effectiveness was compared with the effectiveness of the early stopping and the maximum number of epochs stopping methods across multiple common machine learning problems and neural network models. Our study shows that the proposed Correlation-Driven Stopping Criterion can enhance the out-of-sample performance of all tested neural network models while conserving computing power.

Postoperative Hemodynamics of Total Knee Arthroplasty Unaffected by Cementless Approach under Contemporary Patient Blood Management Protocol: A Propensity Score-Matched Study.

(1) Background: A cementless total knee arthroplasty (TKA) is a recent and an increasingly popular innovation that enhances porous fixation surfaces. However, the lack of cemented sealing of an exposed resected bone has raised concerns about the potential for greater blood loss. The goals of this study were to determine if a cementless approach impacts post-TKA hemodynamics and to identify risk factors for blood loss in instances of cementless (vs. cemented) TKAs under a contemporary patient blood management (PBM) protocol. (2) Methods: We recruited 153 consecutive patients undergoing unilateral TKAs between 2019 and 2023. All enrollees received cementless or cemented prostheses of the same design (cementless, 87; cemented, 66). After propensity score matching for demographics, there were 46 patients in each group. We then compared blood loss metrics (total [TBL] and estimated [EBL]), drainage volumes, hemoglobin (Hb) levels, and transfusion rates by group. (3) Results: Post-TKA hemodynamics (i.e., TBL, EBL, drainage, Hb level, and transfusion rate) of cementless (n = 46) and cemented (n = 46) TKA groups did not differ significantly. In addition, the proportions of patients with Hb drops > 3.0 g/dL were similar for the two groups. A logistic regression analysis revealed that only preoperative Hb and EBL during the early postoperative period were predictive of a substantial fall in Hb levels. The fixation method was not associated with Hb decline > 3.0 g/dL by postoperative Day 3. (4) Conclusion: The cementless TKA has no impact on customary post-TKA hemodynamics and is not associated with greater TKA-related blood loss when implementing a contemporary PBM protocol.

Transforming Text Generation in NLP: Deep Learning with GPT Models and 2023 Twitter Corpus Using Transformer Architecture

This paper presents the design, implementation, and evaluation of a Transformer-based Generative Pre-trained Transformer (GPT) model tailored for character-level text generation. Leveraging the robust architecture of the Transformer, the model has been trained on a corpus sourced from social media text data, with the aim of exploring the intricacies of language patterns within a condensed and informal text setting. Key aspects of the model include a multi-head self-attention mechanism with a custom head configuration, positional embeddings, and layer normalization to promote stability in learning. It operates with a defined set of hyperparameters: a batch size of 32, a block size of 128, 200 iterations, a learning rate of 3e-4, and employs 4 attention heads across 4 layers with an embedding dimension of 384. The model has been optimized using the AdamW optimizer and includes regularization through dropout to prevent overfitting.Through a series of training iterations, the model demonstrates a converging behavior in loss metrics, indicating effective learning, and showcases the capacity to generate coherent text sequences post-training. Training and validation losses have been reported, revealing the nuances in model performance and generalization capabilities. The generated text samples postulate the model's potential in capturing the contextual flow of the dataset. This study further plots the loss curves, visually representing the training dynamics and convergence patterns. The final model, encapsulated within a PyTorch framework, presents a step forward in the realm of neural text generation, contributing to the ongoing advancements in language modeling and its applications in understanding and generating human-like text.

Enhancing Image Quality: A Comparative Study of Spatial, Frequency Domain, and Deep Learning Methods

Image restoration and noise reduction methods have been created to restore deteriorated images and improve their quality. These methods have garnered substantial significance in recent times, mainly due to the growing utilization of digital imaging across diverse domains, including but not limited to medical imaging, surveillance, satellite imaging, and numerous others.
 In this paper, we conduct a comparative analysis of three distinct approaches to image restoration: the spatial method, the frequency domain method, and the deep learning method. The study was conducted on a dataset of 10,000 images, and the performance of each method was evaluated using the accuracy and loss metrics. The results show that the deep learning method outperformed the other two methods, achieving a validation accuracy of 72.68% after 10 epochs. The spatial method had the lowest accuracy of the three, achieving a validation accuracy of 69.98% after 10 epochs. The FFT frequency domain method had a validation accuracy of 52.87% after 10 epochs, significantly lower than the other two methods. The study demonstrates that deep learning is a promising approach for image classification tasks and outperforms traditional methods such as spatial and frequency domain techniques.

Enhancing Multi-Output Time Series Forecasting with Encoder-Decoder Networks

Background: Multi-output Time series forecasting is a complex problem that requires handling interdependencies and interactions between variables. Traditional statistical approaches and machine learning techniques often struggle to predict such scenarios accurately. Advanced techniques and model reconstruction are necessary to improve forecasting accuracy in complex scenarios. Objective: This study proposed an Encoder-Decoder network to address multi-output time series forecasting challenges by simultaneously predicting each output. This objective is to investigate the capabilities of the Encoder-Decoder architecture in handling multi-output time series forecasting tasks. Methods: This proposed model utilizes a 1-Dimensional Convolution Neural Network with Bidirectional Long Short-Term Memory, specifically in the encoder part. The encoder extracts time series features, incorporating a residual connection to produce a context representation used by the decoder. The decoder employs multiple unidirectional LSTM modules and Linear transformation layers to generate the outputs each time step. Each module is responsible for specific output and shares information and context along the outputs and steps. Results: The result demonstrates that the proposed model achieves lower error rates, as measured by MSE, RMSE, and MAE loss metrics, for all outputs and forecasting horizons. Notably, the 6-hour horizon achieves the highest accuracy across all outputs. Furthermore, the proposed model exhibits robustness in single-output forecast and transfer learning, showing adaptability to different tasks and datasets. Conclusion: The experiment findings highlight the successful multi-output forecasting capabilities of the proposed model in time series data, with consistently low error rates (MSE, RMSE, MAE). Surprisingly, the model also performs well in single-output forecasts, demonstrating its versatility. Therefore, the proposed model effectively various time series forecasting tasks, showing promise for practical applications. Keywords: Bidirectional Long Short-Term Memory, Convolutional Neural Network, Encoder-Decoder Networks, Multi-output forecasting, Multi-step forecasting, Time-series forecasting

Vision Transformers and Transfer Learning Approaches for Arabic Sign Language Recognition

Sign languages are complex, but there are ongoing research efforts in engineering and data science to recognize, understand, and utilize them in real-time applications. Arabic sign language recognition (ArSL) has been examined and applied using various traditional and intelligent methods. However, there have been limited attempts to enhance this process by utilizing pretrained models and large-sized vision transformers designed for image classification tasks. This study aimed to create robust transfer learning models trained on a dataset of 54,049 images depicting 32 alphabets from an ArSL dataset. The goal was to accurately classify these images into their corresponding Arabic alphabets. This study included two methodological parts. The first one was the transfer learning approach, wherein we utilized various pretrained models namely MobileNet, Xception, Inception, InceptionResNet, DenseNet, and BiT, and two vision transformers namely ViT, and Swin. We evaluated different variants from base-sized to large-sized pretrained models and vision transformers with weights initialized from the ImageNet dataset or otherwise randomly. The second part was the deep learning approach using convolutional neural networks (CNNs), wherein several CNN architectures were trained from scratch to be compared with the transfer learning approach. The proposed methods were evaluated using the accuracy, AUC, precision, recall, F1 and loss metrics. The transfer learning approach consistently performed well on the ArSL dataset and outperformed other CNN models. ResNet and InceptionResNet obtained a comparably high performance of 98%. By combining the concepts of transformer-based architecture and pretraining, ViT and Swin leveraged the strengths of both architectures and reduced the number of parameters required for training, making them more efficient and stable than other models and existing studies for ArSL classification. This demonstrates the effectiveness and robustness of using transfer learning with vision transformers for sign language recognition for other low-resourced languages.

The Impact of Early-Stage Chronic Kidney Disease on Weight Loss Outcomes After Gastric Bypass

PurposeWeight loss achieved through bariatric metabolic surgery was demonstrated to be effective at reversing chronic kidney dysfunction associated with obesity-related glomerulopathy. However, robust data on how pre-operative kidney status impacts on bariatric metabolic surgery weight loss outcomes is still lacking. The aim of this study was to evaluate the impact of kidney dysfunction on weight loss outcomes after bariatric metabolic surgery.MethodsPatients with obesity to be submitted to gastric bypass surgery underwent a pre-operative evaluation of creatinine clearance, estimated glomerular filtration rate (eGFR), proteinuria, and albuminuria in 24-hour urine. Body mass index (BMI), % total weight loss (%TWL), and % excess BMI loss (%EBMIL) were assessed at 6 and 12 months after surgery.ResultsBefore surgery, patients (N=127) had a mean BMI of 39.6 ± 3.0 kg/m2, and 56.7% (n=72) had a creatinine clearance > 130 mL/min, 23.6% (n= 30) presented proteinuria > 150 mg/24h, and 15.0% (n= 19) presented albuminuria > 30 mg/24h. After surgery, the mean BMI was 27.7 kg/m2 and 25.0 kg/m2 at 6 and 12 months, respectively (p<0.0001). The %TWL was lower in patients with pre-operative eGFR < percentile 25 (34.4 ± 5.8% vs 39.4 ± 4.9%, p=0.0007, at 12 months). There were no significant correlations between weight loss metrics and pre-operative creatinine clearance rate, proteinuria, or albuminuria.ConclusionEarly-stage chronic kidney disease (G2) has a negative impact on short-term weight loss outcomes after bariatric metabolic surgery, albeit in a magnitude inferior to the clinically relevant threshold.Graphical abstract