Search Results (1,703)

Background and Objectives: Liver cancer ranks among the leading causes of cancer-related mortality, necessitating the development of novel diagnostic methods. Deregulated lipid metabolism, a hallmark of hepatocarcinogenesis, offers compelling prospects for biomarker identification. This study aims to employ explainable artificial intelligence (XAI) to identify lipidomic biomarkers for liver cancer and to develop a robust predictive model for early diagnosis. Materials and Methods: This study included 219 patients diagnosed with liver cancer and 219 healthy controls. Serum samples underwent untargeted lipidomic analysis with LC-QTOF-MS. Lipidomic data underwent univariate and multivariate analyses, including fold change (FC), t-tests, PLS-DA, and Elastic Network feature selection, to identify significant biomarker candidate lipids. Machine learning models (AdaBoost, Random Forest, Gradient Boosting) were developed and evaluated utilizing these biomarkers to differentiate liver cancer. The AUC metric was employed to identify the optimal predictive model, whereas SHAP was utilized to achieve interpretability of the model’s predictive decisions. Results: Notable alterations in lipid profiles were observed: decreased sphingomyelins (SM d39:2, SM d41:2) and increased fatty acids (FA 14:1, FA 22:2) and phosphatidylcholines (PC 34:1, PC 32:1). AdaBoost exhibited a superior classification performance, achieving an AUC of 0.875. SHAP identified PC 40:4 as the most efficacious lipid for model predictions. The SM d41:2 and SM d36:3 lipids were specifically associated with an increased risk of low-onset cancer and elevated levels of the PC 40:4 lipid. Conclusions: This study demonstrates that untargeted lipidomics, in conjunction with explainable artificial intelligence (XAI) and machine learning, may effectively identify biomarkers for the early detection of liver cancer. The results suggest that alterations in lipid metabolism are crucial to the progression of liver cancer and provide valuable insights for incorporating lipidomics into precision oncology. Full article

In the field of sustainable mobility, this study highlights the importance of using machine learning for predictive modeling based on real traffic data collected from instrumented bicycles. The advent of advanced technologies like sustainable mobility apps, sensors, and advanced data analysis methods led to the ability to collect data from various sources, which enabled researchers to estimate battery state of charge (SOC) accurately. Most current research uses them in the lab experiments for data collection. In this work, we use real-time sensors data to construct data-driven models for lithium-ion battery SOC estimation. This research integrates both electric bicycle battery, environmental and route variables to achieve the following goals: (1) Collect a multimodal data set including operational, topography, vehicle, and external variables, (2) Preprocess data obtained from sensors installed on the electric bicycle battery, (3) Create models of lithium-ion battery SOC based on electric bicycle battery and environmental variables, and (4) Assess data-driven models and compare their performance for lithium-ion battery SOC with high accuracy. To achieve that, we conducted a real study to predict the Remaining Useful Life (RUL), as a measure of state of charge, of electric bicycle battery. The study was carried out on a 15 km cycle route in Medellín, Colombia, for 28 days. To estimate the RUL, we used four different machine learning algorithms: Long Short-Term Memory (LSTM), Support Vector Regression (SVR), AdaBoost, and Gradient Boost. Notably, data preprocessing techniques played a pivotal role, with a particular focus on smoothing sensor data using Convolutional Neural Networks (CNN). The results showed a significant improvement in prediction accuracy when using data preprocessing, confirming its importance in improving model performance. Furthermore, the comparison of network performance facilitated the selection of the most effective model for the test data. This study underscores the value of using real-world data to develop and validate predictive models in the pursuit of sustainable mobility solutions, and highlights the critical role of data-driven methodologies in addressing today’s urban transportation challenges. Full article

This study investigates the potential of Chirich (Asphodelus aestivus) tuber, one of Turkey’s natural resources, for sustainable bio-hybrid film production. Bio-hybrid films developed from Chirich tuber starch in composite form with polyvinyl alcohol (PVOH) were thoroughly examined for their physical, mechanical, and barrier properties. During the production process, twin-screw extrusion and hydraulic hot pressing methods were employed; the films’ optical, chemical, and barrier performances were analyzed through FT-IR spectroscopy, water vapor permeability, solubility, and mechanical tests. To evaluate the films’ durability against environmental factors and model their properties, advanced computational model algorithms such as Gradient Boosting Regression (GBR), Random Forest Regression (RFR), and AdaBoost Regression (ABR) were utilized. The results showed that the GBR algorithm achieved the highest accuracy with 99.92% R² and presented the most robust model in terms of sensitivity to environmental factors. The results indicate that Chirich tuber-based bio-hybrid films exhibit significantly enhanced mechanical strength and barrier performance compared to conventional corn starch-based biodegradable polymers. These superior properties make them particularly suitable for industrial applications such as food packaging and medical materials, where durability, moisture resistance, and gas barrier characteristics are critical. Moreover, their biodegradability and potential for integration into circular economy frameworks underscore their environmental sustainability, offering a viable alternative to petroleum-derived plastics. The incorporation of ML-driven optimization not only facilitates precise property prediction but also enhances the scalability of bio-hybrid film production. By introducing an innovative, data-driven approach to sustainable material design, this study contributes to the advancement of bio-based polymers in industrial applications, supporting global efforts to mitigate plastic waste and promote environmentally responsible manufacturing practices. Full article

This article investigates the effectiveness of various classification techniques in predicting financial distress for Albanian firms. The dataset includes 16 financial ratios from the financial statements of 187 of the largest non-financial businesses operating in Albania, covering the period from 2011 up to 2014, and ranked by 2014 revenues. The methods used in predicting financial distress are logistic regression, Ada Boost, Naïve Bayes, decision trees, support vector machine (SVM), neural network, and random forest. To compare the effectiveness of the models applied we used Classification Accuracy (CA), confusion matrix, and area under the curve (AUC) as evaluation criteria. The results demonstrate the superior predictive ability of ensemble methods, with random forest achieving more accurate forecasts than other methods, followed by Ada Boost. The research contributes to the literature by showing the added value of machine learning models in emerging markets with unique practice and economic conditions and proposing an alternative classification approach for the classification of financial distress when lacking bankruptcy data. Finally, the empirical findings evidence that the strengths of ensemble learning methods are reinforced in unbalanced not-big datasets of a unique emerging economy. These insights are relevant for lending institutions and researchers aiming to refine credit risk models in unique markets where access to relevant data is a challenge. Full article

A machine learning (ML)-based methodology for predicting frosts was applied to the southern and southeastern regions of Brazil, as well as to other countries including Uruguay, Paraguay, northern Argentina, and southeastern Bolivia. The machine learning model (using TensorFlow (TF)) was compared to the frost index (IG from the Portuguese: Índice de Geada) developed by the National Institute for Space Research (INPE, Brazil). The IG is estimated using meteorological variables from a regional weather numerical model (RWNM). After calculating the two indices using the ML model and the RWNM, a voting committee (VC) was trained to select between the computed outputs. The AdaBoostClassifier algorithm was employed to implement the voting committee. The study area was subdivided into three distinct subregions: R1 (outside Brazil), R2 (the south of Brazil), and R3 (southeastern Brazil). Two forecasting time scales were evaluated: 24 h and 72 h. The 24 h forecasts from both approaches (TF and RWNM) exhibited a similar performance in terms of the number of accurate predictions. However, in the region covering Uruguay and northern Argentina, the TensorFlow model demonstrated superior frost prediction accuracy. Additionally, the TensorFlow model outperformed the RWNM for the 72 h forecast horizon. Full article

Semiconductor radiation detectors usually use a specific signal conditioning circuit, ensuring the required detection system parameters. This paper details the noise properties of specific input stages in photoreceivers that detect various types of radiation. For this purpose, the popular silicon PIN photodiode (BPW34) and two different types of low-noise operational amplifiers (AD797A and ADA4625-1) were used. In the presented experiments, noise measurements were provided for voltage and transimpedance amplifiers operating in input stages, comparing their noise and bandwidths. This made it possible to obtain results for bipolar junction transistor (BJT)- and field-effect transistor (FET)-based input stages of circuity, cooperating directly with a photodiode. Analyzing the obtained characteristics and considering the photodiode operation mode, it is evident that the transimpedance amplifier and photoconductive mode should be considered a typical first-choice solution. In some cases, the performances, such as bandwidth and noise, may be similar to those of voltage. Nevertheless, the bias method used in TIA and feedback compensation can also affect the resulting output noise spectral characteristics due to the photodiode and other capacitances existing in the circuit. In the case of a high transimpedance, the FET-based op-amps ensure lower output noise than the BJT-based ones due to the significantly lower current noise. The simple radiation detector with two-channel differential TIA was also proposed and tested based on the results obtained. Full article

The increasing use of high-resolution cross-sectional imaging has significantly enhanced the detection of pancreatic cystic lesions (PCLs), including pseudocysts and neoplastic entities such as IPMN, MCN, and SCN. However, accurate categorization of PCLs remains a challenge. This study aims to improve PCL evaluation by developing and validating a radiomics-based software tool leveraging machine learning (ML) for lesion classification. The model categorizes PCLs into mucinous and non-mucinous types using a custom dataset of 261 CT examinations, with 156 images for training and 105 for external validation. Three experienced radiologists manually delineated the images, extracting 38 radiological and 214 radiomic features using the Pyradiomics module in Python 3.13.2. Feature selection was performed using Least Absolute Shrinkage and Selection Operator (LASSO) regression, followed by classification with an Adaptive Boosting (AdaBoost) model trained on the optimized feature set. The proposed model achieved an accuracy of 89.3% in the internal validation cohort and demonstrated robust performance in the external validation cohort, with 90.2% sensitivity, 80% specificity, and 88.2% overall accuracy. Comparative analysis with existing radiomics-based studies showed that the proposed model either outperforms or performs on par with the current state-of-the-art methods, particularly in external validation scenarios. These findings highlight the potential of radiomics-driven machine learning approaches in enhancing PCL diagnosis across diverse patient populations. Full article

Mango (Mangifera Indica L.), a key horticultural crop, particularly in Pakistan, has been primarily studied locally using low- to medium-resolution satellite imagery, usually focusing on a particular phenological stage. The large canopy size, complex tree structure, and unique phenology of mango trees further accentuate intrinsic challenges posed by low-spatiotemporal-resolution data. The absence of mango-specific vegetation indices compounds the problem of accurate health classification and yield estimation at the tree level. To overcome these issues, this study utilizes high-resolution multi-spectral UAV imagery collected from two mango orchards in Multan, Pakistan, throughout the annual phenological cycle. It introduces MangiSpectra, an integrated two-staged framework based on Long Short-Term Memory (LSTM) networks. In the first stage, nine conventional and three mango-specific vegetation indices derived from UAV imagery were processed through fine-tuned LSTM networks to classify the health of individual mango trees. In the second stage, associated data such as the trees’ age, variety, canopy volume, height, and weather data were combined with predicted health classes for yield estimation through a decision tree algorithm. Three mango-specific indices, namely the Mango Tree Yellowness Index (MTYI), Weighted Yellowness Index (WYI), and Normalized Automatic Flowering Detection Index (NAFDI), were developed to measure the degree of canopy covered by flowers to enhance the robustness of the framework. In addition, a Cumulative Health Index (CHI) derived from imagery analysis after every flight is also proposed for proactive orchard management. MangiSpectra outperformed the comparative benchmarks of AdaBoost and Random Forest in health classification by achieving 93% accuracy and AUC scores of 0.85, 0.96, and 0.92 for the healthy, moderate and weak classes, respectively. Yield estimation accuracy was reasonable with $R^2=0.21$, and $\mathrm{RMSE}=50.18$. Results underscore MangiSpectra’s potential as a scalable precision agriculture tool for sustainable mango orchard management, which can be improved further by fine-tuning algorithms using ground-based spectrometry, IoT-based orchard monitoring systems, computer vision-based counting of fruit on control trees, and smartphone-based data collection and insight dissemination applications. Full article

Background: Colorectal cancer liver metastasis (CRLM) is a significant contributor to cancer-related illness and death. Neoadjuvant chemotherapy (NAC) is an essential treatment approach; however, optimal patient selection remains a challenge. This study aimed to develop a machine learning-based predictive model using hematological biomarkers to assess the efficacy of NAC in patients with CRLM. Methods: We retrospectively analyzed the clinical data of 214 CRLM patients treated with the XELOX regimen. Blood characteristics before and after NAC, as well as the ratios of these biomarkers, were integrated into the machine learning models. Logistic regression, decision trees (DTs), random forest (RF), support vector machine (SVM), and AdaBoost were used for predictive modeling. The performance of the models was evaluated using the AUROC, F1-score, and external validation. Results: The DT (AUROC: 0.915, F1-score: 0.621) and RF (AUROC: 0.999, F1-score: 0.857) models demonstrated the best predictive performance in the training cohort. The model incorporating the ratio of post-treatment to pre-treatment gamma-glutamyl transferase (rGGT) and carcinoembryonic antigen (rCEA) formed the GCR index, which achieved an AUROC of 0.853 in the external validation. The GCR index showed strong clinical relevance, predicting better chemotherapy responses in patients with lower rCEA and higher rGGT levels. Conclusions: The GCR index serves as a predictive biomarker for the efficacy of NAC in CRLM, providing a valuable clinical reference for the prognostic assessment of these patients. Full article

Background/Objectives: Therapeutic management of inflammatory bowel diseases (IBD) is rapidly evolving in the era of novel biological therapies. However, real-world data relating to the usage trends and treatment persistence remain inconsistent. This study aimed to investigate trends in biological use, dose intensification, and treatment persistence in IBD patients, who received treatment in a large tertiary center in Greece. Methods: Patients with IBD who underwent at least one biological treatment between 2018 and 2022 were included in this retrospective study. Data on patients’ demographics, type of disease, use of biologicals, dose intensification, and treatment persistence were analyzed for time trends. Results: Data from 409 patients with IBD (mean age 39 (range 17–87), female 51%, 56.9% CD, mean duration of disease: 9.3 years) were included in the study. The number of patients on biologics was raised from 133 in 2018 to 368 in 2022 (a 28.1% yearly increase), while the percentage of patients who were treated with anti-TNF biosimilars increased to >60% of the total anti-TNF population in 2022. We observed a gradual increase in non-anti-TNF therapies in bio-naïve patients, in particular vedolizumab (46% of all biologicals in UC; 16% in CD) and ustekinumab (16.3% of all biologicals in UC, 31% in CD). The 3-year persistence rate of IFX was 64% in CD and 56% in UC, whereas it was 61% for ADA in CD. Dose intensification of anti-TNF was efficient in >50% of CD patients and >30% of UC patients; however, the majority of patients who required dose escalation within the first year eventually became unresponsive. The 3-year persistence of vedolizumab as a first-line treatment was 82% for CD and 69% for UC, respectively. The 3-year persistence of ustekinumab as first-line treatment for CD was 65%. No significant differences regarding the efficacy of anti-TNF, ustekinumab, or vedolizumab were detected when they were used as first-line treatments for Crohn’s disease; similarly, no significant differences were detected between infliximab and vedolizumab as first-line treatments for UC. Conclusions: There was a gradual increase in the use of biologicals, including biosimilars, between the years 2018–2022, reflecting adherence to current guidance with adoption of an early escalation strategy. Newer, post-anti-TNF biologics such as vedolizumab and ustekinumab have been rapidly incorporated into therapeutic approaches for both CD and UC. Full article

The trend in all automotive manufacturers is to commercialize vehicles with an increasing number of sophisticated Advanced Driver-Assistance Systems (ADASs). These systems often require that several sensors, such as Light Detection and Ranging (LIDAR), radio detection and ranging (radar), cameras, etc., work in cooperation, which makes the systems very complex. To perform the electromagnetic compatibility (EMC) validation of these complex ADASs, the stimulation of multiple sensors composing the system is necessary. Furthermore, the synchronization of these stimulations is essential to create realistic outdoor scenarios in the usual EMC facilities (on a roller bench in a semi-anechoic chamber). This synchronization is mandatory as the integrated safety systems will disable any ADAS or autonomous system in case of incoherencies in the data delivered by the sensors, rendering the validation challenging. Moreover, the current methodologies proposed are meant to be performed to validate simple ADASs based on simple sensors. In addition, with the current test facilities, one cannot stimulate, in a realistic and synchronous way, multiple sophisticated sensors (e.g., LIDARs and inertial measurement units). For all these reasons, the radiated immunity tests of future automotive systems will be endlessly difficult following current trends. In addition, the complexity of the systems and their increasing number increase the duration and cost of these immunity tests and make their validations more challenging. In this article, we present a new methodology to validate the radiated immunity of complex automotive autonomous systems to address these challenges. The results we present show that this new methodology can be performed to validate ADASs and autonomous automotive systems independently of their complexity. Full article

Timely prediction of collisions enables advanced driver assistance systems to issue warnings and initiate emergency maneuvers as needed to avoid collisions. With recent developments in computer vision and deep learning, collision warning systems that use vision as the only sensory input have emerged. They are less expensive than those that use multiple sensors, but their effectiveness must be thoroughly assessed. We systematically searched academic literature for studies proposing ego-centric, vision-based collision warning systems that use deep learning techniques. Thirty-one studies among the search results satisfied our inclusion criteria. Risk of bias was assessed with PROBAST. We reviewed the selected studies and answer three primary questions: What are the (1) deep learning techniques used and how are they used? (2) datasets and experiments used to evaluate? (3) results achieved? We identified two main categories of methods: Those that use deep learning models to directly predict the probability of a future collision from input video, and those that use deep learning models at one or more stages of a pipeline to compute a threat metric before predicting collisions. More importantly, we show that the experimental evaluation of most systems is inadequate due to either not performing quantitative experiments or various biases present in the datasets used. Lack of suitable datasets is a major challenge to the evaluation of these systems and we suggest future work to address this issue. Full article

This paper examines the energy efficiency of smart electric vehicles equipped with regenerative braking systems under challenging weather conditions. While Advanced Driver Assistance Systems (ADAS) are primarily designed to enhance driving safety, they often overlook energy efficiency. This study proposes a Weather-Adaptive Regenerative Braking Strategy (WARBS) system, which leverages onboard sensors and data processing capabilities to enhance the energy efficiency of regenerative braking across diverse weather conditions while minimizing unnecessary alerts. To achieve this, we develop driving style recognition models that integrate road conditions, such as weather and road friction, with different driving styles. Next, we propose an adaptive deceleration plan that aims to maximize the conversion of kinetic energy into electrical energy for the vehicle’s battery under varying weather conditions, considering vehicle dynamics and speed constraints. Given that the potential for energy recovery through regenerative braking is diminished on icy and snowy roads compared to dry ones, our approach introduces a driving context recognition system to facilitate effective speed planning. Both simulation and experimental validation indicate that this approach can significantly enhance overall energy efficiency. Full article

The increasing complexity of polymer systems in both experimental and computational studies has led to an expanding interest in machine learning (ML) methods to aid in data analysis, material design, and predictive modeling. Among the various ML approaches, boosting methods, including AdaBoost, Gradient Boosting, XGBoost, CatBoost and LightGBM, have emerged as powerful tools for tackling high-dimensional and complex problems in polymer science. This paper provides an overview of the applications of boosting methods in polymer science, highlighting their contributions to areas such as structure–property relationships, polymer synthesis, performance prediction, and material characterization. By examining recent case studies on the applications of boosting techniques in polymer science, this review aims to highlight their potential for advancing the design, characterization, and optimization of polymer materials. Full article

A novel fluoride ion fluorescent probe is designed by introducing the strong electron-withdrawing triazine groups into the triarylboron/acridine conjugation system. The A-D-A′ molecular configuration endows this molecule with multiple charge-transfer channels; upon reaction with F⁻, the triazine groups act as primary acceptors within the molecule, facilitating charge transfer between the acridine units and the triazine groups. During fluoride ion detection, changes in the triarylboron moiety lead to a significant bathochromic-shift in fluorescence emission from green to yellow. Theoretical calculations attribute this phenomenon to a reduction in the molecular S₁ state energy level upon fluorination, resulting in a pronounced visible color change and chromogenic response during detection. Based on fluorescence intensity changes with varying degrees of F⁻ coordination, a detection limit as low as 10⁻⁷ M was determined for TB-1DMAc-2TRZ, demonstrating the high sensitivity of this probe. Full article