Machine Learning-based Approach Research Articles

Activation of epidermal growth factor receptors in triple-negative breast cancer cells by morphine; analysis through Raman spectroscopy and machine learning

Triple negative breast cancer (TNBC) is a very aggressive form of breast cancer, and the analgesic drug morphine has been shown to promote the proliferation of TNBC cells. This article investigates whether morphine causes activation of epidermal growth factor receptors (EGFR), the roles of μ-opioid and EGFR receptors on TNBC cell proliferation and migration. While examining the changes with molecular techniques, we also aimed to investigate the analysis ability of Raman spectroscopy and machine learning-based approach. Effects of morphine on the proliferation and migration of MDA.MB.231 cells were evaluated by MTT and scratch wound-healing tests, respectively. Morphine-induced phosphorylation of the EGFR was analyzed by western blotting in the presence and absence of μ-receptor antagonist naltrexone and the EGFR-tyrosine kinase inhibitor gefitinib. Morphine-induced EGFR phosphorylation and cell migration were significantly inhibited by pretreatments with both naltrexone and gefitinib; however, morphine-increased cell proliferation was inhibited only by naltrexone. While morphine-induced changes were observed in the Raman scatterings of the cells, the inhibitory effect of naltrexone was analyzed with similarity to the control group. Principal component analysis (PCA) of the Raman confirmed the epidermal growth factor (EGF)-like effect of morphine and was inhibited by naltrexone and partly by gefitinib pretreatments. Our in vitro results suggest that combining morphine with an EGFR inhibitor or a peripherally acting opioidergic receptor antagonist may be a good strategy for pain relief without triggering cancer proliferation and migration in TNBC patients. In addition, our results demonstrated the feasibility of the Raman spectroscopy and machine learning-based approach as an effective method to investigate the effects of agents in cancer cells without the need for complex and time-consuming sample preparation. The support vector machine (SVM) with linear kernel automatically classified the effects of drugs on cancer cells with ∼95% accuracy.

Satellite-based ensemble intelligent approach for predicting forest fire: a case of the Hyrcanian forest in Iran.

A machine learning-based approach is applied to simulate and forecast forest fires in the Golestan province in Iran. A dataset for no-fire, medium confidence (MC) fire events, and high confidence (HC) fire events is constructed from MODIS-MOD14A2. Nine climate variables from NASA's FLDAS are used as input variables, and 12 dates and 915 study points are considered. Three machine learning ensemble multi-label classifiers, gradient boosting (GBC), random forest (RFC), and extremely randomized tree (ETC), are used for forest fire simulation for the period 2000 to 2021, and ETC is found to be the most accurate classifier. Future fire projection for the near-future period of 2030 to 2050 is carried out with the ETC model, using CMIP6 EC-Earth3-SSP245 General Circulation Model (GCM) data. It is projected that MC forest fire occurrences will decrease, while HC forest fire occurrences will increase, and that the summer months, especially September, will be the most affected by fire.

Predicting bovine daily milk yield by leveraging genomic breeding values

The main goal of this work, conducted on a herd of 502 Holstein cows situated in Italy, is to propose a machine learning-based approach to forecast the individual bovine daily milk production by explicitly leveraging genotypic information. As part of our study, we also evaluated the importance in the prediction of genotypic and phenotypic variables usually available within herd. The methodology we propose is based on two consecutive models: a genomic prediction one to calculate the animal’s genomic breeding value from marker data, followed by a feed forward neural network combining such additive genetic effect and the environmental features (parity, days in milk, age at calving in months, month of calving) for milk yield forecasting. In particular, we both assess the inclusion of genomic breeding values calculated within herd or provided by the breeders’ association, discovering that the latter ones allow for better final predictions. The results of our model outperform the ones by a linear mixed model with the same inputs on average, day by day and at the individual level. Moreover, we propose a problem formulation that also leverages additional factors partially controllable by breeders: in this case, features such as the number of milkings and the concentrate consumption inside the automatic milking system prove to highly impact on the final prediction, and hence on milk production. To the best of our knowledge, the proposed problem formulation based on genomic breeding values is a novelty in the individual bovine milk yield machine learning forecasting literature. Given the low genotyping costs and the availability of a larger number of environmental features in farms equipped with a wide range of sensors, as automatic milking systems, our solution can support breeders’ herd management and animal monitoring, thanks to the possibility to forecast the full lactation curve in advance even for primiparous bovines and newborn calves. With this work, we successfully achieve our objectives of including genomic information in bovine milk yield machine learning-based forecasting, thus improving the performance on this task, and of evaluating the impact on prediction of common genotypic and phenotypic information available to breeders.

Assessing occupational hazards in welding operations: A machine learning-based approach for worker safety in Indian foundries.

Quantitatively analyzing the impact of UV radiation and noise during welding operations is essential to assess the exposure, identify potential hazards, and develop targeted safety protocols to ensure worker safety and adherence to safety regulations, especially in developing countries with inadequate adherence to safety standards and resources. This study employs machine learning for predicting ultraviolet radiation and noise levels during welding, emphasizing worker safety. The focus is on the Indian foundry sector to gauge actual exposure vis-á-vis safety standards. Ultraviolet radiation and noise emitted during the welding of a ferrous alloy were collected from three foundries in Agra, India. Five machine learning (ML) algorithms were applied for data analysis and prediction of UV radiation and noise levels, and a relative performance comparison was carried out on the compiled data against safety standards. Out of all the ML algorithms applied, the Support Vector Machine regression algorithm (RMSE = 356.93) obtained the best performance on UV radiation data, and the Random Forest algorithm (RMSE = 11.4) was found to give the best results for the noise level prediction task. This work represents the first known application of machine learning techniques for predicting UV radiation and noise levels in arc welding processes. The results show the efficacy of algorithms such as SVM regression and Random Forest for the problem. Further, the datasets and ML algorithms implemented in the work will be made openly available to support further research endeavors in this and related areas.

Wind power forecasting based on a machine learning model: considering a coastal wind farm in Zhejiang as an example

ABSTRACT The unpredictability and instability of wind have hindered the development and utilization of wind power. To harness wind energy and ensure a secure and stable power grid after wind power integration, precise predictions of wind power generation are imperative. Here, we apply one-year data from a coastal wind farm in Zhejiang to train a Random Forest (RF) model for predicting wind power generation. The results indicate that the RF model (mean bias (MB) 1.33) outperforms traditional linear models (MB 87.70). An evaluation of prediction-measurement bias shows a strong agreement between the RF model’s predictions and actual measurements (MB 1.33), especially when wind speeds exceed 5.7 m/s. While the Weather Research and Forecasting (WRF) model yields less accurate results (R2 0.65) compared to using measured wind velocity data (R2 0.97) for wind power prediction, it remains acceptable as it can capture and forecast peak generating power, meeting daily management requirements. Therefore, our machine learning-based approach offers practical guidance for reducing wind power generation uncertainties. Enhancing wind velocity forecasting through the WRF model could further improve the RF model’s accuracy in predicting wind power generation. Our study also verified the future application of RF model in coastal areas of China.

On the use of machine learning and data-transformation methods to predict hydration kinetics and strength of alkali-activated mine tailings-based binders

The escalating production of mine tailings (MT), a byproduct of the mining industry, constitutes significant environmental and health hazards, thereby requiring a cost-effective and sustainable solution for its disposal or reuse. This study proposes the use of MT as the primary ingredient (≥70%mass) in binders for construction applications, thereby ensuring their efficient upcycling as well as drastic reduction of environmental impacts associated with the use of ordinary Portland cement (OPC). The early-age hydration kinetics and compressive strength of MT-based binders are evaluated with an emphasis on elucidating the influence of alkali activation parameters and the amount of slag or cement that are used as minor constituents. This study reveals correlations between cumulative heat release and compressive strengths at different ages; these correlations can be leveraged to estimate the compressive strength based on hydration kinetics. Furthermore, this study presents a random forest (RF) model—in conjunction with fast Fourier and direct cosine transformation techniques to overcome the limitations associated with limited volume and diversity of the database—to enable high-fidelity predictions of time-dependent hydration kinetics and compressive strength of MT-based binders in relation to mixture design. Overall, this study demonstrates a sustainable approach to upcycle mine tailings as the primary component in low-carbon construction binders; and presents both analytical and machine learning-based approaches for accurate a priori predictions of hydration kinetics and compressive strength of these binders.

TV shows popularity prediction of genre-independent TV series through machine learning-based approaches

The use of social media has grown exponentially in recent years up to become a reflection of human social attitudes and to represent today the main channel for conducting discussions and sharing opinions. For this reason, the vast amount of information generated is often used for predicting outcomes of real-world events in different fields, including business, politics, and health, as well as in the entertainment industry. In this paper, we focus on how data from Twitter can be used to predict ratings of a large set of TV shows regardless of their specific genre. Given a show, the idea is to exploit features concerning the pre-release hype on Twitter for rating predictions. We propose a novel machine learning-based approach to the genre-independent TV show popularity prediction problem. We compared the performance of several well-known predictive methods, and as a result, we discovered that LSTM and Random Forest can predict the ratings in the USA entertainment market, with a low mean squared error of 0.058. Furthermore, we tested our model by using data of “never seen” shows, by deriving interesting results in terms of error rates. Finally, we compared performance against relevant solutions available in the literature, with discussions about challenges arousing from the analysis of shows in different languages.

Comparative study of respiratory sounds classification methods based on cepstral analysis and artificial neural networks

In this paper, we investigated and evaluated various machine learning-based approaches for automatically detecting wheezing sounds. We conducted a comprehensive comparison of these proposed systems, assessing their classification performance through metrics such as Sensitivity, Specificity, and Accuracy. The main approach to developing a machine learning-based system for classifying respiratory sounds involved the combination of a technique for extracting features from an unknown input sound with a classification method to determine its belonging class. The characterization techniques used in this study are based on the cepstral analysis, which was extensively employed in the automatic speech recognition field. While MFCC (Mel-Frequency Cepstral Coefficients) feature extraction methods are commonly used in respiratory sounds classification, our study introduces a novelty by employing GFCC (Gammatone-Frequency Cepstral Coefficients) and BFCC (Bark-Frequency Cepstral Coefficients) for this purpose. For the classification task, we employed two types of neural networks: the MLP (Multilayer Perceptron), a feedforward neural network, and a variant of the LSTM (Long Short-Term Memory) recurrent neural network called BiLSTM (Bidirectional LSTM). The proposed classification systems are evaluated using a database consisting of 497 wheezing segments and 915 normal respiratory segments, which are recorded from individuals diagnosticated with asthma and individuals without any respiratory issues, respectively. The highest classification performance was achieved by the BFCC-BiLSTM model, which demonstrated an exceptional accuracy rate of 99.8%.

Machine Learning-Based Approach to Identifying Fall Risk in Seafarers Using Wearable Sensors

Falls on a ship cause severe injuries, and an accident falling off board, referred to as “man overboard” (MOB), can lead to death. Thus, it is crucial to accurately and timely detect the risk of falling. Wearable sensors, unlike camera and radar sensors, are affordable and easily accessible regardless of the weather conditions. This study aimed to identify the fall risk level (i.e., high and low risk) among individuals on board using wearable sensors. We collected walking data from accelerometers during the experiment by simulating the ship’s rolling motions using a computer-assisted rehabilitation environment (CAREN). With the best features selected by LASSO, eight machine learning (ML) models were implemented with a synthetic minority oversampling technique (SMOTE) and the best-tuned hyperparameters. In all ML models, the performance in classifying fall risk showed overall a good accuracy (0.7778 to 0.8519), sensitivity (0.7556 to 0.8667), specificity (0.7778 to 0.8889), and AUC (0.7673 to 0.9204). Logistic regression showed the best performance in terms of the AUC for both training (0.9483) and testing (0.9204). We anticipate that this study will effectively help identify the risk of falls on ships and aid in developing a monitoring system capable of averting falls and detecting MOB situations.

An automatic system to detect colorectal polyp using hybrid fused method from colonoscopy images

The third most common disease in the world that leads to mortality is colorectal cancer (CRC). Colonoscopy, which locates and removes colonic polyps, is the most common procedure worldwide for the early identification and preventative treatment of colorectal cancer. The likelihood of a CRC patient dying can be considerably decreased by early diagnosis and treatment of precancerous polyps. Unidentified polyps may eventually turn into cancer. Although several classification algorithms have been put out to identify polyps, their effectiveness has not yet been compared to that of skilled endoscopes. An ensemble machine learning-based classification approach for detecting colorectal cancer from colonoscopy images is presented in this research. Firstly researchers collect colorectal cancer or polyp images from three standard datasets. In the preprocessing phase, researcher convert the RGB image to grayscale image and identified the region of interest (ROI) by removing the unwanted regions. A hybrid anisotropic diffusion filtering (HADF) approach was employed to eliminate noise from each image. Then the system extracts features from individual feature extractor methods and fused extracted features in a vector. The fused features help to detect colorectal polyp or cancer for increasing cancer or polyp identification rates. Finally, an ensemble classifier classifies the colorectal cancer or normal images and achieves better accuracy. The suggested technique exceeds the prior conventional methods, according to tests on widely used public datasets, improving accuracy by 99.45 %, sensitivity by 99.30 %, specificityby 99.58 %, and precision 99.53 %.

Machine learning-based characterisation of urban morphology with the street pattern

Streets are a crucial part of the built environment, and their layouts, the street patterns, are widely researched and contribute to a quantitative understanding of urban morphology. However, traditional street pattern analysis only considers a few broadly defined characteristics. It uses administrative boundaries and grids as units of analysis that fail to encompass the diversity and complexity of street networks. To address these challenges, this research proposes a machine learning-based approach to automatically recognise street patterns that employs an adaptive analysis unit based on street-based local areas (SLAs). SLAs use a network partitioning technique that can adapt to distinct street networks, making it particularly suitable for different urban contexts. By calculating several streets’ network metrics and performing a hierarchical clustering method, streets with similar characters are grouped under the same street pattern. A case study is carried out in six cities worldwide. The results show that street pattern types are rather diverse and hierarchical, and categorising them into clearly demarcated taxonomy is challenging. The study derives a set of new morphometrics-based street patterns with four major types that resemble conventional street patterns and eleven sub-types to significantly increase their diversity for broader coverage of urban morphology. The new patterns capture urban structural differences across cities, such as the urban-suburban division and the number of urban centres present. In conclusion, the proposed machine learning-based morphometric street pattern to characterise urban morphology has an enhanced ability to encompass more information from the built environment while maintaining the intuitiveness of using patterns.

Beyond mirkwood: Enhancing SED Modeling with Conformal Predictions

Traditional spectral energy distribution (SED) fitting techniques face uncertainties due to assumptions in star formation histories and dust attenuation curves. We propose an advanced machine learning-based approach that enhances flexibility and uncertainty quantification in SED fitting. Unlike the fixed NGBoost model used in mirkwood, our approach allows for any scikit-learn-compatible model, including deterministic models. We incorporate conformalized quantile regression to convert point predictions into error bars, enhancing interpretability and reliability. Using CatBoost as the base predictor, we compare results with and without conformal prediction, demonstrating improved performance using metrics such as coverage and interval width. Our method offers a more versatile and accurate tool for deriving galaxy physical properties from observational data.

Monitoring the Workflow of Earthquake by Using Machine Learning

This study explores the innovative application of machine learning techniques in monitoring and analyzing the workflow associated with earthquakes. The primary objective is to enhance the accuracy and efficiency of earthquake detection, prediction, and response mechanisms the development of machine learning technology enables more robust real-time earthquake monitoring through automated implementations. However, the application of machine learning to earthquake location problems faces challenges in regions with limited available training data. This paper proposes a machine learning-based approach to monitor seismic workflows efficiently. We apply various Machine Learning algorithms, including deep learning neural networks and decision trees, to process and interpret vast amounts of seismic data. These algorithms were trained on historical earthquake records to identify patterns and anomalies that precede seismic events. Our approach not only focused on the detection of seismic activities but also on the prediction of potential aftershock locations and magnitudes. Earthquake monitoring and response systems play a crucial role in mitigating potential damages and ensuring rapid response to seismic events. Traditional methods rely on sensor networks and manual analysis, often leading to delays in detection and response. The system leverages real-time data from seismic sensors, incorporating advanced machine learning algorithms for data analysis and pattern recognition. By employing deep learning models, such as convolution neural networks (CNNs) and recurrent neural networks (RNNs), the system can detect and classify seismic activities accurately. Additionally, anomaly detection techniques enable the identification of unusual seismic patterns that might indicate impending earthquakes or aftershocks. The system integrates geographical information systems (GIS) to map the seismic activities spatially, aiding in the visualization and understanding of seismic patterns across regions. The machine learning models are trained on historical seismic data to enhance their predictive capabilities and adaptability to varying seismic behaviors. This proposed system offers a real-time monitoring solution that can assist in early earthquake detection, accurate event classification, and timely response coordination. Its ability to continuously learn from incoming data ensures adaptability to changing seismic behaviors, thereby improving overall efficiency and reliability in earthquake monitoring workflows.

A deep learning framework for non-functional requirement classification

AbstractAnalyzing, identifying, and classifying nonfunctional requirements from requirement documents is time-consuming and challenging. Machine learning-based approaches have been proposed to minimize analysts’ efforts, labor, and stress. However, the traditional approach of supervised machine learning necessitates manual feature extraction, which is time-consuming. This study presents a novel deep-learning framework for NFR classification to overcome these limitations. The framework leverages a more profound architecture that naturally captures feature structures, possesses enhanced representational power, and efficiently captures a broader context than shallower structures. To evaluate the effectiveness of the proposed method, an experiment was conducted on two widely-used datasets, encompassing 914 NFR instances. Performance analysis was performed on the applied models, and the results were evaluated using various metrics. Notably, the DReqANN model outperforms the other models in classifying NFR, achieving precision between 81 and 99.8%, recall between 74 and 89%, and F1-score between 83 and 89%. These significant results highlight the exceptional efficacy of the proposed deep learning framework in addressing NFR classification tasks, showcasing its potential for advancing the field of NFR analysis and classification.

In silico identification of human microRNAs pointing centrin genes in Leishmania donovani: Considering the RNAi-mediated gene control.

Leishmaniasis, a parasitic disease caused by different species of the protozoa parasite Leishmania, is a neglected tropical human disease that is endemic in about a hundred countries worldwide. According to the World Health Organization (WHO), the annual incidence of cutaneous leishmaniasis (CL) is estimated to be 0.7-1.2 million cases globally, whereas the annual incidence of visceral leishmaniasis is estimated to be 0.2-0.4 million cases. In many eukaryotic organisms, including human beings and protozoan parasites, centrin genes encode proteins that play essential roles within the centrosome or basal body. Human microRNAs (miRNAs) have been linked to several infectious and non-infectious diseases associated with pathogen-host interactions, and they play the emphatic roles as gene expression regulators. In this study, we used the MirTarget bioinformatics tool, which is a machine learning-based approach implemented in miRDB, to predict the target of human miRNAs in Leishmania donovani centrin genes. For cross-validation, we utilized additional prediction algorithms, namely, RNA22 and RNAhybrid, targeting all five centrin isotypes. The centrin-3 (LDBPK_342160) and putative centrin-5 (NC_018236.1) genes in L. donovani were targeted by eight and twelve human miRNAs, respectively, among 2,635 known miRNAs (miRBase). hsa-miR-5193 consistently targeted both genes. Using TargetScan, TarBase, miRecords, and miRTarBase, we identified miRNA targets and off-targets in human homologs of centrin, inflammation, and immune-responsive genes. Significant targets were screened based on GO terminologies and KEGG pathway-enrichment analysis (Log10p-value >0.0001). In silico tools that predict the biological roles of human miRNAs as primary gene regulators in pathogen-host interactions help unravel the regulatory patterns of these miRNAs, particularly in the early stages of inflammatory responses. It is also noted that these miRNAs played an important role in the late phase of adaptive immune response, inclusively their impacts on the immune system's response to L. donovani.

Machine learning analysis of RB-TnSeq fitness data predicts functional gene modules in Pseudomonas putida KT2440.

This study demonstrates a rapid, automated approach for elucidating functional modules within complex genetic networks. While Pseudomonas putida randomly barcoded transposon insertion sequencing data were used as a proof of concept, this approach is applicable to any organism with existing functional genomics data sets and may serve as a useful tool for many valuable applications, such as guiding metabolic engineering efforts in other microbes or understanding functional relationships between virulence-associated genes in pathogenic microbes. Furthermore, this work demonstrates that comparison of data obtained from independent component analysis of transcriptomics and gene fitness datasets can elucidate regulatory-functional relationships between genes, which may have utility in a variety of applications, such as metabolic modeling, strain engineering, or identification of antimicrobial drug targets.

Machine Learning-Based Approach for Predicting the Altcoins Price Direction Change from a High-Frequency Data of Seven Years Based on Socio-Economic Factors, Bitcoin Prices, Twitter and News Sentiments

Machine Learning-Based Approach for Predicting the Altcoins Price Direction Change from a High-Frequency Data of Seven Years Based on Socio-Economic Factors, Bitcoin Prices, Twitter and News Sentiments

“To share or not to share?” – A hybrid SEM-ANN-NCA study of the enablers and enhancers for mobile sharing economy

The proliferation of mobile technologies have facilitated the popularity of mobile sharing economy (MSE). However, extant literature that used sufficient logic has fallen short in explaining this phenomenon as the findings are insufficient to explain the necessary conditions that activate users' intention and actual usage. The purpose of this study is to identify the enablers (must-have factors) and enhancers (should-have factors) of consumers' intention and actual use of mobile sharing economy. A must-have factor is needed to initialize the effect of independent variables while a should-have factor enhances the effect. Drawing from the extended valence framework, a combined SEM-ANN-NCA analysis is used to examine users' actual adoption of MSE. The application of the “sufficiency logic” and “necessity logic” and the AI and machine learning-based ANN approach enables us to ascertain the linear-nonlinear relationships and the enablers and enhancers in mobile sharing economy. The findings show that mobile ease of use is a sufficient and necessary condition for mobile usefulness while trust is a sufficient and necessary condition for epistemic benefit. Mobile usefulness and trust are the sufficient and necessary conditions for behavioral intention while behavioral intention is the sufficient and necessary condition for actual use. Trust is a sufficient but unnecessary condition for convenience and psychological risk. Epistemic benefit and psychological risk are sufficient but not necessary conditions for behavioral intention. Education is a sufficient but not necessary condition for actual use. The study has provided important new practical and theoretical insights to MSE service providers, operators, users and scholars.

Prediction of Hydration Heat for Diverse Cementitious Composites through a Machine Learning-Based Approach.

Hydration plays a crucial role in cement composites, but the traditional methods for measuring hydration heat face several limitations. In this study, we propose a machine learning-based approach to predict hydration heat at specific time points for three types of cement composites: ordinary Portland cement pastes, fly ash cement pastes, and fly ash-metakaolin cement composites. By adjusting the model architecture and analyzing the datasets, we demonstrate that the optimized artificial neural network model not only performs well during the learning process but also accurately predicts hydration heat for various cement composites from an extra dataset. This approach offers a more efficient way to measure hydration heat for cement composites, reducing the need for labor- and time-intensive sample preparation and testing. Furthermore, it opens up possibilities for applying similar machine learning approaches to predict other properties of cement composites, contributing to efficient cement research and production.

Psychological Assessment of Autism Spectrum Disorders using Machine Learning

This research aims to advance the field of autism spectrum disorder (ASD) assessment by proposing a machine learning-based approach tailored for school and community settings. Leveraging a diverse dataset encompassing behavioral, physiological, and neuroimaging data, we apply advanced machine learning algorithms to develop predictive models for early ASD detection. Our study integrates expert interviews to validate the clinical utility of these models and explores ethical considerations surrounding data privacy and bias. Preliminary results show promising accuracy in ASD identification. This research contributes to a more accessible and objective ASD assessment, with implications for early intervention and inclusive support, particularly in educational and community contexts.