Improving remote sensing observations of the water cycle with analytical methods, simple statistical models, and more complex machine-learning models

The integration of machine learning (ML) models with cloud computing has transformed the landscape of predictive analytics, offering scalable, efficient, and flexible solutions for organizations. Cloud platforms such as AWS, Google Cloud, and Microsoft Azure enable businesses to deploy and manage complex ML models without the need for extensive on-premise infrastructure. However, optimizing these ML models for performance and cost-efficiency in cloud environments presents unique challenges, including resource management, latency, scalability, and data security. This paper focuses on strategies to optimize machine learning models specifically for predictive analytics in cloud environments. It explores key techniques such as auto-scaling, model compression, and hyperparameter tuning, which are critical for improving the accuracy and speed of predictions while minimizing computational costs. The research also examines advanced tools such as containerization, serverless computing, and cloud-native services that further streamline the deployment and management of ML models. In the Indian context, where cloud adoption is growing rapidly, optimizing ML models is crucial for businesses across various sectors, including finance, healthcare, and e-commerce. By leveraging cloud-based ML solutions, Indian companies can enhance their predictive analytics capabilities, driving smarter decision-making and operational efficiency. This abstract presents an overview of how optimized machine learning models can unlock the full potential of predictive analytics in cloud environments, leading to better business outcomes. Through case studies and practical applications, this paper provides actionable insights into the best practices for optimizing ML models in a cloud-based setting.

Although complex machine learning models (e.g., Random Forest, Neural Networks) are commonly outperforming the traditional simple interpretable models (e.g., Linear Regression, Decision Tree), in the healthcare domain, clinicians find it hard to understand and trust these complex models due to the lack of intuition and explanation of their predictions. With the new General Data Protection Regulation (GDPR), the importance for plausibility and verifiability of the predictions made by machine learning models has become essential. To tackle this challenge, recently, several machine learning interpretability techniques have been developed and introduced. In general, the main aim of these interpretability techniques is to shed light and provide insights into the predictions process of the machine learning models and explain how the model predictions have resulted. However, in practice, assessing the quality of the explanations provided by the various interpretability techniques is still questionable. In this paper, we present a comprehensive experimental evaluation of three recent and popular local model agnostic interpretability techniques, namely, LIME, SHAP and Anchors on different types of real-world healthcare data. Our experimental evaluation covers different aspects for its comparison including identity, stability, separability, similarity, execution time and bias detection. The results of our experiments show that LIME achieves the lowest performance for the identity metric and the highest performance for the separability metric across all datasets included in this study. On average, SHAP has the smallest average time to output explanation across all datasets included in this study. For detecting the bias, SHAP enables the participants to better detect the bias.

Simple SummaryStudies comparing performances of machine learning (ML) methods in building predictive models of toxicity in RT are rare. Thanks to the availability of a large cohort (n = 1314) of breast cancer patients homogeneously treated with tangential fields, different ML approaches could be compared. This work shows how more complex models typically achieve higher performances. At the same time, for this test case, the importance is given mainly by a few variables, and toxicity can be predicted by simpler models with similar performances. The availability of more individually characterizing features (here partially missing) is expected to have a likely much higher impact than the choice of the best-performing ML/DL approach.Purpose. Different ML models were compared to predict toxicity in RT on a large cohort (n = 1314). Methods. The endpoint was RTOG G2/G3 acute toxicity, resulting in 204/1314 patients with the event. The dataset, including 25 clinical, anatomical, and dosimetric features, was split into 984 for training and 330 for internal tests. The dataset was standardized; features with a high p-value at univariate LR and with Spearman were excluded; synthesized data of the minority were generated to compensate for class imbalance. Twelve ML methods were considered. Model optimization and sequential backward selection were run to choose the best models with a parsimonious feature number. Finally, feature importance was derived for every model. Results. The model’s performance was compared on a training–test dataset over different metrics: the best performance model was LightGBM. Logistic regression with three variables (LR3) selected via bootstrapping showed performances similar to the best-performing models. The AUC of test data is slightly above 0.65 for the best models (highest value: 0.662 with LightGBM). Conclusions. No model performed the best for all metrics: more complex ML models had better performances; however, models with just three features showed performances comparable to the best models using many (n = 13–19) features.

This study aims to develop and validate a machine learning-based mortality risk prediction model for V-A ECMO patients to improve the precision of clinical decision-making. This multicenter retrospective cohort study included 280 patients receiving V-A ECMO from the Second Affiliated Hospital of Guangxi Medical University, Yulin First People’s Hospital, and the MIMIC-IV database. The data from the Second Affiliated Hospital of Guangxi Medical University and the MIMIC-IV database were merged and randomly divided in a 7:3 ratio into a training set and an internal validation set, respectively. The dataset from Yulin First People’s Hospital was reserved as an external validation cohort. The primary study outcome was defined as in-hospital mortality.Feature selection was conducted using Lasso regression, followed by the development of six machine learning models: Logistic Regression, Random Forest (RF), Deep Neural Network (DNN), Support Vector Machine (SVM), LightGBM, and CatBoost. Model performance was assessed using the Area Under the Curve (AUC), accuracy, sensitivity, specificity, and F1 score. Model validation was performed through calibration and decision curve analysis. Feature importance was evaluated using SHAP, and subgroup analysis was conducted to assess the model’s applicability across different clinical scenarios. In internal validation, the Logistic Regression model performed the best, with an AUC of 0.86 (95% CI: 0.77–0.93), accuracy of 0.76, sensitivity of 0.73, specificity of 0.79, and an F1 score of 0.73. It outperformed other models (RF: AUC = 0.79, DNN: AUC = 0.78, SVM: AUC = 0.76, LightGBM: AUC = 0.71, CatBoost: AUC = 0.77). External validation yielded consistent results, with the Logistic Regression model’s AUC at 0.75 (95% CI: 0.56–0.92), accuracy of 0.69, sensitivity of 0.64, specificity of 0.73, and an F1 score of 0.66. Calibration curve analysis revealed that the Logistic Regression model had the lowest Brier score (0.1496), indicating the most reliable predicted probabilities. Decision curve analysis demonstrated that the model provided the highest net benefit across most decision thresholds. SHAP analysis identified lactate, age, and albumin as key predictors of mortality, with lactate and age positively correlated, and albumin negatively correlated. Subgroup analysis revealed better performance in the cardiac arrest group (AUC = 0.81), non-sepsis group (AUC = 0.75), and non-diabetes group (AUC = 0.78). The Logistic Regression-based mortality risk prediction model for V-A ECMO patients demonstrated comparable or even favorable performance to more complex machine learning models, with the advantage of higher interpretability.By explicitly incorporating lactate, age, and albumin as the principal predictors, this model facilitates precise risk stratification and provides practical support for clinical decision-making in ECMO management.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-25423-4.

Machine-Learning-as-a-Service (MLaaS) has become a widespread paradigm, making even the most complex Machine Learning models available for clients via, e.g., a pay-per-query principle. This allows users to avoid time-consuming processes of data collection, hyperparameter tuning, and model training. However, by giving their customers access to the (predictions of their) models, MLaaS providers endanger their intellectual property such as sensitive training data, optimised hyperparameters, or learned model parameters. In some cases, adversaries can create a copy of the model with (almost) identical behaviour using the the prediction labels only. While many variants of this attack have been described, only scattered defence strategies that address isolated threats have been proposed. To arrive at a comprehensive understanding why these attacks are successful and how they could be holistically defended against, a thorough systematisation of the field of model stealing is necessary. We address this by categorising and comparing model stealing attacks, assessing their performance, and exploring corresponding defence techniques in different settings. We propose a taxonomy for attack and defence approaches and provide guidelines on how to select the right attack or defence strategy based on the goal and available resources. Finally, we analyse which defences are rendered less effective by current attack strategies.

Accurately predicting the performance of Earth Pressure Balance Tunnel Boring Machines (EPB-TBMs) in soft ground conditions is crucial yet challenging due to the complex interaction of geological and operational factors. This study investigates Mashhad Metro Line 3, where a TBM was employed to excavate a 1831-m section through variable soil compositions, including significant cobble and boulder content, presenting unique challenges to performance prediction. To address these complexities, several machine learning models—Multiple Linear Regression (MLR), Decision Trees (DT), and Multi-Layer Perceptron (MLP) neural networks—were applied to predict TBM penetration rates and assess model efficacy. Beginning with a dataset of 438,960 rows, rigorous feature selection and data processing yielded a final dataset of 1833 rows. Among the models, MLR achieved an R2 score of 0.991, closely matching the more complex MLP model, which reached an R2 score of 0.988. In contrast, the Decision Tree model demonstrated a lower R2 score of 0.923, suggesting a tendency to overfit. While MLR provided an effective, straightforward approach, MLP proved valuable for capturing non-linear patterns that could improve predictive accuracy in more variable tunneling conditions. These findings underscore the practical applications of both simple and complex machine learning models in enhancing TBM performance prediction.

Modern medicine is supported by increasingly sophisticated algorithms. In diagnostics or screening, statistical models are commonly used to assess the risk of disease development, the severity of its course, and expected treatment outcome. The growing availability of very detailed data and increased interest in personalized medicine are leading to the development of effective but complex machine learning models. For these models to be trusted, their predictions must be understandable to both the physician and the patient, hence the growing interest in the area of Explainable Artificial Intelligence (XAI). In this paper, we present selected methods from the XAI field in the example of models applied to assess lung cancer risk in lung cancer screening through low-dose computed tomography. The use of these techniques provides a better understanding of the similarities and differences between three commonly used models in lung cancer screening, i.e., BACH, PLCOm2012, and LCART. For the presentation of the results, we used data from the Domestic Lung Cancer Database. The XAI techniques help to better understand (1) which variables are most important in which model, (2) how they are transformed into model predictions, and facilitate (3) the explanation of model predictions for a particular screenee.

Accurate hospital bed occupancy forecasting is essential for effective resource planning and patient flow management. While complex machine learning models have gained popularity in healthcare forecasting, their operational utility often falls short due to high maintenance costs and limited interpretability. This study evaluates the performance and practicality of Prophet, a parsimonious time-series model, for mid-term hospital bed occupancy forecasting. We applied the Prophet model to daily bed occupancy data from the Medical Center - University of Freiburg (2010-2023), incorporating public holidays and a COVID-19 pandemic indicator as exogenous regressors. Prophet decomposes time series into trend, seasonality, and holiday effects, offering interpretable components. Forecast accuracy was assessed via rolling cross-validation over 2022-2023 for horizons of 30, 60, 90, and 180 days. A production-ready forecasting pipeline and dashboard were also implemented using cloud-native tools. Prophet achieved low MAPE values across all horizons (3.21%-3.53%) with coverage above 80%, demonstrating reliable accuracy comparable to or better than more complex models that often require higher computational resources and operational costs, such as deep neural networks. Component analysis revealed patterns aligned with hospital operations; weekly and yearly cycles, and holiday effects, highlighting the model's interpretability. This study shows that mid-term hospital bed occupancy can be accurately forecasted using a simple, interpretable model like Prophet. In contrast to more complex architectures, Prophet offers robust performance with minimal tuning, faster deployment, and clearer insights that are critical in operational settings. These findings reinforce the argument that, for structured forecasting tasks like bed occupancy, simple models can rival complex ones, not only in accuracy, but also in reproducibility, scalability, and operational value.

Student drop-out is one of the most burning issues in STEM higher education, which induces considerable social and economic costs. Using machine learning tools for the early identification of students at risk of dropping out has gained a lot of interest recently. However, there has been little discussion on dropout prediction using interpretable machine learning (IML) and explainable artificial intelligence (XAI) tools.In this work, using the data of a large public Hungarian university, we demonstrate how IML and XAI tools can support educational stakeholders in dropout prediction. We show that complex machine learning models – such as the CatBoost classifier – can efficiently identify at-risk students relying solely on pre-enrollment achievement measures, however, they lack interpretability. Applying IML tools, such as permutation importance (PI), partial dependence plot (PDP), LIME, and SHAP values, we demonstrate how the predictions can be explained both globally and locally. Explaining individual predictions opens up great opportunities for personalized intervention, for example by offering the right remedial courses or tutoring sessions. Finally, we present the results of a user study that evaluates whether higher education stakeholders find these tools interpretable and useful.

Many modern day applications can be solved with the usage of machine learning, which involves training a computer to learn on large amounts of data without direct programmer guidance. Conventional computers typically use normal general purpose central processing units, though more specialized tasks may take advantage of more parallel hardware such as graphics processing units. In the pursuit of increased performance to facilitate increasingly more complex machine learning models, researchers in both academia and industry look towards field-programmable gate arrays and application specific integrated circuits for their needs. Various implementations, both theoretical and practical, exist across a wide variety of designs. A custom design, using systolic arrays and built on the existing RISC-V Instruction Set Architecture, will be used to accelerate matrix calculations, with example performance on the MNIST dataset measured.

Machine learning (ML) often provides applicable high-performance models to facilitate decision-makers in various fields. However, this high performance is achieved at the expense of the interpretability of these models, which has been criticized by practitioners and has become a significant hindrance in their application. Therefore, in highly sensitive decisions, black boxes of ML models are not recommended. We proposed a novel methodology that uses complex supervised ML models and transforms them into simple, interpretable, transparent statistical models. This methodology is like stacking ensemble ML in which the best ML models are used as a base learner to compute relative feature weights. The index of these weights is further used as a single covariate in the simple logistic regression model to estimate the likelihood of an event. We tested this methodology on the primary dataset related to cardiovascular diseases (CVDs), the leading cause of mortalities in recent times. Therefore, early risk assessment is an important dimension that can potentially reduce the burden of CVDs and their related mortality through accurate but interpretable risk prediction models. We developed an artificial neural network and support vector machines based on ML models and transformed them into a simple statistical model and heart risk scores. These simplified models were found transparent, reliable, valid, interpretable, and approximate in predictions. The findings of this study suggest that complex supervised ML models can be efficiently transformed into simple statistical models that can also be validated.

The adoption of complex machine learning (ML) models in recent years has brought along a new challenge related to how to interpret, understand, and explain the reasoning behind these complex models' predictions. Treating complex ML systems as trustworthy black boxes without domain knowledge checking has led to some disastrous outcomes. In this context, interpretability and explainability are often used unintelligibly, and fairness, on the other hand, has become lately popular due to some discrimination problems in ML. While closely related, interpretability and explainability denote different features of prediction. In this sight, the aim of this paper is to give an overview of the interpretability, explainability and the fairness concepts in the literature and to evaluate the performance of the Patient Rule Induction Method (PRIM) concerning these aspects.

Can Machine Learning Models Capture Correlations in Corporate Distresses?

Printed electronics (PE) feature low non-recurring engineering costs and low per unit-area fabrication costs, enabling thus extremely low-cost and on-demand hardware. Such low-cost fabrication allows for high customization that would be infeasible in silicon, and bespoke architectures prevail to improve the efficiency of emerging PE machine learning (ML) applications. However, even with bespoke architectures, the large feature sizes in PE constraint the complexity of the ML models that can be implemented. In this work, we bring together, for the first time, approximate computing and PE design targeting to enable complex ML models, such as Multi-Layer Perceptrons (MLPs) and Support Vector Machines (SVMs), in PE. To this end, we propose and implement a cross-layer approximation, tailored for bespoke ML architectures. At the algorithmic level we apply a hardware-driven coefficient approximation of the ML model and at the circuit level we apply a netlist pruning through a full search exploration. In our extensive experimental evaluation we consider 14 MLPs and SVMs and evaluate more than 4300 approximate and exact designs. Our results demonstrate that our cross approximation delivers Pareto optimal designs that, compared to the state-of-the-art exact designs, feature 47% and 44% average area and power reduction, respectively, and less than 1% accuracy loss.

<div>Abstract<p>Patients with mantle cell lymphoma (MCL), an incurable B-cell malignancy, benefit from accurate pretreatment disease stratification. We curated an extensive database of 862 patients diagnosed between 2014 and 2022. A machine learning (ML) gradient-boosted model incorporated baseline features from clinicopathological, cytogenetic, and genomic data with high predictive power discriminating between patients with indolent or responsive MCL and those with aggressive disease (AUC ROC = 0.83). Additionally, we utilized the gradient-boosted framework as a robust feature selection method for multivariate logistic and survival modeling. The best ML models incorporated features from clinical and genomic data types highlighting the need for correlative molecular studies in precision oncology. As proof of concept, we launched our most accurate and practical models using an application interface, which has potential for clinical implementation. We designated the 20-feature ML model-based index the “integrative MIPI” or iMIPI and a similar 10-feature ML index the “integrative simplified MIPI” or iMIPI-s. The top 10 baseline prognostic features represented in the iMIPI-s are: Lactase Dehydrogenase (LDH), Ki-67%, platelet count, bone marrow involvement percentage, hemoglobin levels, the total number of observed somatic mutations, TP53 mutational status, ECOG performance level, beta-2 microglobulin, and morphology. Our findings emphasize that prognostic applications and indices should include molecular features, especially TP53 mutational status. This work demonstrates the clinical utility of complex ML models and provides further evidence for existing prognostic markers in MCL.</p></div>