Random Forest Machine Learning Model Research Articles

HRpredict: Introducing a Web-Based Application for Heart Rate Prediction and Lifestyle Recommendations

Background: Monitoring heart health requires early detection of deviations in HR, which makes it easier to detect and address heart irregularities at an early stage. Health remote systems when combined with artificial intelligence (AI) can assist in better health outcomes through early detection of heart problems. Aims: Our main goal is to create a website application (Web-App) for web browser access, aiming to utilize a Random Forest (RF) machine learning (ML) model trained to predict the average heart rate (HR) over 10 days for different periods, and to enable lifestyle and activity recommendations. Methods: The Web-App is created using Laravel, an open-source Personal Home Page (PHP) web framework that follows the model-view-controller (MVC) architectural pattern. Results: This research resulted in a web-based ML model that can be used to predict future heart rates over a 10-day period which are utilized to establish average HR values, considering baseline and three distinct periods: morning, noon, and evening across the 10-day duration. Through this Web-App lifestyle, habit, activity, and 10-day reassessment recommendations are also provided. Conclusion: The Web-App was designed to be accessed and used through a web browser, to provide lifestyle recommendations based on predicted HR readings. To determine the impact of users adhering to recommendations, further research is required.

Multi-Objective Stochastic Paint Optimizer for Solving Dynamic Economic Emission Dispatch with Transmission Loss Prediction Using Random Forest Machine Learning Model

Dynamic economic emission dispatch problems are complex optimization tasks in power systems that aim to simultaneously minimize both fuel costs and pollutant emissions while satisfying various system constraints. Traditional methods often involve solving intricate nonlinear load flow equations or employing approximate loss formulas to account for transmission losses. These methods can be computationally expensive and may not accurately represent the actual transmission losses, affecting the overall optimization results. To address these limitations, this study proposes a novel approach that integrates transmission loss prediction into the dynamic economic emission dispatch (DEED) problem. A Random Forest machine learning model was offline-trained to predict transmission losses accurately, eliminating the need for repeated calculations during each iteration of the optimization process. This significantly reduced the computational burden of the algorithm and improved its efficiency. The proposed method utilizes a powerful multi-objective stochastic paint optimizer to solve the highly constrained and complex dynamic economic emission dispatch problem integrated with random forest-based loss prediction. A fuzzy membership-based approach was employed to determine the best compromise Pareto-optimal solution. The proposed algorithm integrated with loss prediction was validated on widely used five and ten-unit power systems with B-loss coefficients. The results obtained using the proposed algorithm were compared with seventeen algorithms available in the literature, demonstrating that the multi-objective stochastic paint optimizer (MOSPO) outperforms most existing algorithms. Notably, for the Institute of Electrical and Electronics Engineers (IEEE) thirty bus system, the proposed algorithm achieves yearly fuel cost savings of USD 37,339.5 and USD 3423.7 compared to the existing group search optimizer algorithm with multiple producers (GSOMP) and multi-objective multi-verse optimization (MOMVO) algorithms.

A supervised machine learning model for identifying predictive factors for recommending head and neck cancer surgery.

New patient referrals are often processed by practice coordinators with little-to-no medical background. Treatment delays due to incorrect referral processing, however, have detrimental consequences. Identifying variables that are associated with a higher likelihood of surgical oncological resection may improve patient referral processing and expedite the time to treatment. The study objective is to develop a supervised machine learning (ML) platform that identifies relevant variables associated with head and neck surgical resection. A retrospective cohort study was conducted on 64 222 patient datapoints from the SEER database. The random forest ML model correctly classified patients who were offered head and neck surgery with an 81% accuracy rate. The sensitivity and specificity rates were 86% and 71%. The positive and negative predictive values were 85% and 73%. ML modeling accurately predicts head and neck cancer surgery recommendations based on patient and cancer information from a large population-based dataset. ML adjuncts for referral processing may decrease the time to treatment for patients with cancer.

Machine learning-based optimal distributed generation and electric vehicle load management

The load profile of radial distribution networks (RDNs) is significantly impacted when plug-in electric vehicles (PEVs) are connected to them in large numbers. The disturbances in the load profile may lead to increased power losses in distribution lines and deterioration of the network voltage profile. The provision of distributed generation at strategic locations in the distribution network can help to compensate for the impact on the electrical network due to PEV loads. This paper proposes the use of machine learning (ML)-based models for determining the optimal location of distributed generators (DGs) in an RDN. The proposed method considered time-varying load in addition to PEV load. The suggested method determines optimal placement of DGs based on the power loss reduction index and voltage deviation index reduction index. Four distinct types of ML models were used in the proposed approach using synthesised data on the Institute of Electrical and Electronics Engineers' 33-bus RDN. The performance of the ML models was evaluated with respect to mean squared error and mean absolute percentage error and, for the ML models considered, the random forest ML model gave the best performance.

Digital assessment of speech in Huntington disease.

Speech changes are an early symptom of Huntington disease (HD) and may occur prior to other motor and cognitive symptoms. Assessment of HD commonly uses clinician-rated outcome measures, which can be limited by observer variability and episodic administration. Speech symptoms are well suited for evaluation by digital measures which can enable sensitive, frequent, passive, and remote administration. We collected audio recordings using an external microphone of 36 (18 HD, 7 prodromal HD, and 11 control) participants completing passage reading, counting forward, and counting backwards speech tasks. Motor and cognitive assessments were also administered. Features including pausing, pitch, and accuracy were automatically extracted from recordings using the BioDigit Speech software and compared between the three groups. Speech features were also analyzed by the Unified Huntington Disease Rating Scale (UHDRS) dysarthria score. Random forest machine learning models were implemented to predict clinical status and clinical scores from speech features. Significant differences in pausing, intelligibility, and accuracy features were observed between HD, prodromal HD, and control groups for the passage reading task (e.g., p < 0.001 with Cohen'd = -2 between HD and control groups for pause ratio). A few parameters were significantly different between the HD and control groups for the counting forward and backwards speech tasks. A random forest classifier predicted clinical status from speech tasks with a balanced accuracy of 73% and an AUC of 0.92. Random forest regressors predicted clinical outcomes from speech features with mean absolute error ranging from 2.43-9.64 for UHDRS total functional capacity, motor and dysarthria scores, and explained variance ranging from 14 to 65%. Montreal Cognitive Assessment scores were predicted with mean absolute error of 2.3 and explained variance of 30%. Speech data have the potential to be a valuable digital measure of HD progression, and can also enable remote, frequent disease assessment in prodromal HD and HD. Clinical status and disease severity were predicted from extracted speech features using random forest machine learning models. Speech measurements could be leveraged as sensitive marker of clinical onset and disease progression in future clinical trials.

Decoding Diagnostic Precision: Evaluating Error Implications with Random Forest and Support Vector Regression Machine Learning Models

Decoding Diagnostic Precision: Evaluating Error Implications with Random Forest and Support Vector Regression Machine Learning Models

Diversity, composition, metabolic characteristics, and assembly process of the microbial community in sewer system at the early stage.

Sewer systems play vital roles in wastewater treatment facilities, and the microbial communities contribute significantly to the transformation of domestic wastewater. Therefore, this study conducted a 180-day experiment on a sewer system and utilized the high-throughput sequencing technology to characterize the microbial communities. Additionally, community assembly analysis was performed to understand the early-stage dynamics within the sewer system. The results demonstrated that the overall diversity of microbial communities exhibited fluctuations as the system progressed. The dominant phyla observed were Chloroflexi, Bacteroidetes, Firmicutes, and Proteobacteria, accounting for over 85.4% of the total relative abundances. At the genus level, bacteria associated with fermentation displayed a high relative abundance, particularly during days 75 to 180. A random-forest machine-learning model identified a group of microbes that confirmed the substantial contribution of fermentation. During the process of fermentation, microorganisms predominantly utilized propionate formation as the main pathway for acidogenesis, followed by acetate and butyrate formation. In terms of nitrogen and sulfur cycles, dissimilatory nitrate reduction and assimilatory sulfate reduction played significant roles. Furthermore, stochastic ecological processes had a dominant effect during the experiment. Dispersal limitation primarily governed the assembly process almost the entire experimental period, indicating the strong adaptability and metabolic plasticity of microorganisms in response to environmental variations. This experiment provides valuable insights into the metabolic mechanisms and microbial assembly associated with sewer systems.

Can adverse childhood experiences predict chronic health conditions? Development of trauma-informed, explainable machine learning models.

Decades of research have established the association between adverse childhood experiences (ACEs) and adult onset of chronic diseases, influenced by health behaviors and social determinants of health (SDoH). Machine Learning (ML) is a powerful tool for computing these complex associations and accurately predicting chronic health conditions. Using the 2021 Behavioral Risk Factor Surveillance Survey, we developed several ML models-random forest, logistic regression, support vector machine, Naïve Bayes, and K-Nearest Neighbor-over data from a sample of 52,268 respondents. We predicted 13 chronic health conditions based on ACE history, health behaviors, SDoH, and demographics. We further assessed each variable's importance in outcome prediction for model interpretability. We evaluated model performance via the Area Under the Curve (AUC) score. With the inclusion of data on ACEs, our models outperformed or demonstrated similar accuracies to existing models in the literature that used SDoH to predict health outcomes. The most accurate models predicted diabetes, pulmonary diseases, and heart attacks. The random forest model was the most effective for diabetes (AUC = 0.784) and heart attacks (AUC = 0.732), and the logistic regression model most accurately predicted pulmonary diseases (AUC = 0.753). The strongest predictors across models were age, ever monitored blood sugar or blood pressure, count of the monitoring behaviors for blood sugar or blood pressure, BMI, time of last cholesterol check, employment status, income, count of vaccines received, health insurance status, and total ACEs. A cumulative measure of ACEs was a stronger predictor than individual ACEs. Our models can provide an interpretable, trauma-informed framework to identify and intervene with at-risk individuals early to prevent chronic health conditions and address their inequalities in the U.S.

Predicting Mood Based on the Social Context Measured Through the Experience Sampling Method, Digital Phenotyping, and Social Networks

Social interactions are essential for well-being. Therefore, researchers increasingly attempt to capture an individual's social context to predict well-being, including mood. Different tools are used to measure various aspects of the social context. Digital phenotyping is a commonly used technology to assess a person's social behavior objectively. The experience sampling method (ESM) can capture the subjective perception of specific interactions. Lastly, egocentric networks are often used to measure specific relationship characteristics. These different methods capture different aspects of the social context over different time scales that are related to well-being, and combining them may be necessary to improve the prediction of well-being. Yet, they have rarely been combined in previous research. To address this gap, our study investigates the predictive accuracy of mood based on the social context. We collected intensive within-person data from multiple passive and self-report sources over a 28–day period in a student sample (Participants: N = 11, ESM measures: N = 1313). We trained individualized random forest machine learning models, using different predictors included in each model summarized over different time scales. Our findings revealed that even when combining social interactions data using different methods, predictive accuracy of mood remained low. The average coefficient of determination over all participants was 0.06 for positive and negative affect and ranged from − 0.08 to 0.3, indicating a large amount of variance across people. Furthermore, the optimal set of predictors varied across participants; however, predicting mood using all predictors generally yielded the best predictions. While combining different predictors improved predictive accuracy of mood for most participants, our study highlights the need for further work using larger and more diverse samples to enhance the clinical utility of these predictive modeling approaches.

Using machine learning to identify indicators of rare earth element enrichment in sedimentary strata with applications for metal prospectivity

Rare earth elements (REE), classified as critical minerals which are crucial for clean energy technologies, face soaring demand. While economic deposits are found in limited geologic environments including carbonatites and ion-adsorption clays, unconventional, secondary sources such as those from sedimentary basins could hold potential to meet this increased demand. Coal and its associated combustion by-products, phosphorites, oil sands tailings, and formation waters have all garnered interest for REE recovery, yet they remain significantly underexplored. Accordingly, new tools for data analysis and optimization such as machine learning can assist in mineral prospectivity, with these tools being subject to rapid proliferation in the Earth sciences.This work leverages compositional data analysis principles and machine learning to probe geochemical relationships and predict REE abundances in sedimentary lithologies using unsupervised (correlation, principal component, and cluster analysis) and supervised (regression, support vector machine, random forest, and boosting) machine learning models. These three unsupervised models display similar results, with REE typically being associated with incompatible elements (e.g., Th, Nb, and Hf). Gradient boosting, Adaboost, and Random Forest had the highest performance for predicting REE concentrations, with Th and P commonly being the most important predictor variables. Identifying geochemical indicators of REE enrichment that may be used to assist in discovering potentially exploitable REE resources based on existing data, as well as increasing the understanding of metal behaviour in sedimentary systems, is a step forward in understanding novel secondary and unconventional REE sources. Although REE concentrations from these sources are generally lower than primary ore deposits, the amount of available feedstock, potentially simpler, cheaper, and less environmentally taxing extraction processes, and the added benefit of remediating waste streams and contributing to the circular economy make these sources alluring.

The rising human footprint in the Tibetan Plateau threatens the effectiveness of ecological restoration on vegetation growth

Ecological restoration projects in the Tibetan Plateau aimed to reverse ecosystem degradation and safeguard the fragile alpine ecological environment. However, it is still being determined if the vegetation restoration is successful on a large scale or reaches the expected magnitude, restricting our ability to adapt practices to maximise the benefit. With multiple vegetation indices (VIs: NDVI, LAI, and GPP) from satellite observations and random forest machine-learning models, we performed an attribution study on vegetation growth trends caused by climate change and human activities. Then, we further explored the relationship between vegetation growth and ecological projects and human footprint without the influence of climate. The results showed that climatic change was a relatively strong driver of vegetation growth. The positive contributions of ecological restoration occurred only in half of the plateau due to the increased human footprint. Vegetation enhancement resulting from ecological restoration occurred in 13.1%–23.1% of the plateau. Among these values, ecological restoration counteracted the negative climate effects in 4.7%–8.3% of the plateau (about half of the negative climate effect area). In forest and grassland protection areas, the ecological restoration was more successful. The study provides a better understanding of the role of ecological projects in vegetation restoration and potential threats to its effectiveness. This is essential to improve future restoration projects.

Identification of immune microenvironment subtypes and clinical risk biomarkers for osteoarthritis based on a machine learning model.

Osteoarthritis (OA) is a degenerative disease with a high incidence worldwide. Most affected patients do not exhibit obvious discomfort symptoms or imaging findings until OA progresses, leading to irreversible destruction of articular cartilage and bone. Therefore, developing new diagnostic biomarkers that can reflect articular cartilage injury is crucial for the early diagnosis of OA. This study aims to explore biomarkers related to the immune microenvironment of OA, providing a new research direction for the early diagnosis and identification of risk factors for OA. We screened and downloaded relevant data from the Gene Expression Omnibus (GEO) database, and the immune microenvironment-related genes (Imr-DEGs) were identified using the ImmPort data set by combining weighted coexpression analysis (WGCNA). Functional enrichment of GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) were conducted to explore the correlation of Imr-DEGs. A random forest machine learning model was constructed to analyze the characteristic genes of OA, and the diagnostic significance was determined by the Receiver Operating Characteristic Curve (ROC) curve, with external datasets used to verify the diagnostic ability. Different immune subtypes of OA were identified by unsupervised clustering, and the function of these subtypes was analyzed by gene set enrichment analysis (GSVA). The Drug-Gene Interaction Database was used to explore the relationship between characteristic genes and drugs. Single sample gene set enrichment analysis (ssGSEA) revealed that 16 of 28 immune cell subsets in the dataset significantly differed between OA and normal groups. There were 26 Imr-DEGs identified by WGCNA, showing that functional enrichment was related to immune response. Using the random forest machine learning model algorithm, nine characteristic genes were obtained: BLNK (AUC = 0.809), CCL18 (AUC = 0.692), CD74 (AUC = 0.794), CSF1R (AUC = 0.835), RAC2 (AUC = 0.792), INSR (AUC = 0.765), IL11 (AUC = 0.662), IL18 (AUC = 0.699), and TLR7 (AUC = 0.807). A nomogram was constructed to predict the occurrence and development of OA, and the calibration curve confirmed the accuracy of these 9 genes in OA diagnosis. This study identified characteristic genes related to the immune microenvironment in OA, providing new insight into the risk factors of OA.

Using machine learning for the optimisation of operations and management in electric systems and networks

This research employs the Random Forest Machine Learning model to predict electricity consumption and detect anomalies in electrical networks. Addressing the energy sector’s challenges, such as supply reliability and renewable energy integration, this model processes historical electricity consumption data, weather conditions, and network events to efficiently forecast demand and identify anomalies. Data cleansing and normalisation preceded the training phase, where the model was fine-tuned using historical data to balance forecast accuracy and overfitting avoidance. The dataset was divided into training (80%) and testing (20%) sets for performance evaluation. Through cross-validation, optimal model hyperparameters were determined. The findings highlight the model’s efficacy in accurately predicting daily electricity consumption in a small, homogenous town. The model achieved a Mean Absolute Error (MAE) of 198.73 MWh and a coefficient of determination (R²) of 0.9387. Temperature, humidity, and wind speed were identified as key influencing factors on consumption levels. Conclusively, the Random Forest model presents a valuable tool for energy management, offering precise consumption forecasting and anomaly detection capabilities. Future work will address computational demands and enhance model integration with other Machine Learning methods for improved performance. This contribution is significant for efficient energy system planning and operation.

Metabolic Heterogeneity of Tumor Cells and its Impact on Colon Cancer Metastasis: Insights from Single-Cell and Bulk Transcriptome Analyses.

Background: Metabolic reprogramming plays a crucial role in the development of colorectal cancer (CRC), influencing tumor heterogeneity, the tumor microenvironment, and metastasis. While the interaction between metabolism and CRC is critical for developing personalized treatments, gaps remain in understanding how tumor cell metabolism affects prognosis. Our study introduces novel insights by integrating single-cell and bulk transcriptome analyses to explore the metabolic landscape within CRC cells and its mechanisms influencing disease progression. This approach allows us to uncover metabolic heterogeneity and identify specific metabolic genes impacting metastasis, which have not been thoroughly examined in previous studies. Methods: We sourced microarray and single-cell RNA sequencing datasets from the Gene Expression Omnibus (GEO) and bulk sequencing data for CRC from The Cancer Genome Atlas (TCGA). We employed Gene Set Variation Analysis (GSVA) to assess metabolic pathway activity, consensus clustering to identify CRC-specific transcriptome subtypes in bulkseq, and rigorous quality controls, including the exclusion of cells with high mitochondrial gene expression in scRNA seq. Advanced analyses such as AUCcell, infercnvCNV, Non-negative Matrix Factorization (NMF), and CytoTRACE were utilized to dissect the cellular landscape and evaluate pathway activities and tumor cell stemness. The hdWGCNA algorithm helped identify prognosis-related hub genes, integrating these findings using a random forest machine learning model. Results: Kaplan-Meier survival curves identified 21 significant metabolic pathways linked to prognosis, with consensus clustering defining three CRC subtypes (C3, C2, C1) based on metabolic activity, which correlated with distinct clinical outcomes. The metabolic activity of the 13 cell subpopulations, particularly the epithelial cell subpopulation with active metabolic levels, was evaluated using AUCcell in scRNA seq. To further analyze tumor cells using infercnv, NMF disaggregated these cells into 10 cellular subpopulations. Among these, the C2 subpopulation exhibited higher stemness and tended to have a poorer prognosis compared to C6 and C0. Conversely, the C8, C3, and C1 subpopulations demonstrated a higher level of the five metabolic pathways, and the C3 and C8 subpopulations tended to have a more favorable prognosis. hdWGCNA identified 20 modules, from which we selected modules primarily expressed in high metabolic tumor subgroups and highly correlated with clinical information, including blue and cyan. By applying variable downscaling of RF to a total of 50 hub genes, seven gene signatures were obtained. Furthermore, molecules that were validated to be protective in GEO were screened alongside related molecules, resulting in the identification of prognostically relevant molecules such as UQCRFS1 and GRSF1. Additionally, the expression of GRSF1 was examined in colon cancer cell lines using qPCR and phenotypically verified by in vitro experiments. Conclusion: Our findings emphasize that high activity in specific metabolic pathways, including pyruvate metabolism and the tricarboxylic acid cycle, correlates with improved colon cancer outcomes, presenting new avenues for metabolic-based therapies. The identification of hub genes like GRSF1 and UQCRFS1 and their link to favorable metabolic profiles offers novel insights into tumor neovascularization and metastasis, with significant clinical implications for targeting metabolic pathways in CRC therapy.

Artificial intelligence performance in testing microfluidics for point-of-care.

Artificial intelligence (AI) is revolutionizing medicine by automating tasks like image segmentation and pattern recognition. These AI approaches support seamless integration with existing platforms, enhancing diagnostics, treatment, and patient care. While recent advancements have demonstrated AI superiority in advancing microfluidics for point of care (POC) diagnostics, a gap remains in comparative evaluations of AI algorithms in testing microfluidics. We conducted a comparative evaluation of AI models specifically for the two-class classification problem of identifying the presence or absence of bubbles in microfluidic channels under various imaging conditions. Using a model microfluidic system with a single channel loaded with 3D transparent objects (bubbles), we challenged each of the tested machine learning (ML) (n = 6) and deep learning (DL) (n = 9) models across different background settings. Evaluation revealed that the random forest ML model achieved 95.52% sensitivity, 82.57% specificity, and 97% AUC, outperforming other ML algorithms. Among DL models suitable for mobile integration, DenseNet169 demonstrated superior performance, achieving 92.63% sensitivity, 92.22% specificity, and 92% AUC. Remarkably, DenseNet169 integration into a mobile POC system demonstrated exceptional accuracy (>0.84) in testing microfluidics at under challenging imaging settings. Our study confirms the transformative potential of AI in healthcare, emphasizing its capacity to revolutionize precision medicine through accurate and accessible diagnostics. The integration of AI into healthcare systems holds promise for enhancing patient outcomes and streamlining healthcare delivery.

Secure-Tech Triad Enhancing Electronic Voting System Security through Integrated Blockchain, AI, and IoT Technologies

As electronic voting systems become increasingly prevalent, the urgent need for robust security measures to combat evolving cyber threats has never been more critical. This paper introduces a ground-breaking architectural framework Secure-Tech Triad that synergistically combines Blockchain technology with Machine Learning (ML) algorithms and Internet of Things (IoT) capabilities to enhance the security and efficiency of electronic voting systems. This architectural framework utilizes a modified Proof-of-Stake (PoS) Blockchain algorithm, a Random Forest ML model for real-time anomaly detection, and an MQTT protocol for IoT-based data collection to create a more secure, efficient, and responsive voting environment. Rigorous testing and evaluation show that the integrated framework significantly outperforms existing Blockchain-only solutions in key performance indicators, such as security breach detection rate, system latency, and cost efficiency. This integrated approach is the best-performing model, achieving a 97% security breach detection rate, a 30% reduction in system latency (down to 2.3 seconds), and a 25% decrease in operational costs. These results underscore the combined effectiveness of Blockchain, AI, and IoT in enhancing security, speed, and cost-effectiveness. Specifically, the Random Forest algorithm has been instrumental in achieving an exceptional security breach detection rate, while IoT data collection has played a pivotal role in enabling real-time anomaly detection and proactive threat mitigation.

Prediction of tuberculosis clusters in the riverine municipalities of the Brazilian Amazon with machine learning.

Tuberculosis (TB) is the second most deadly infectious disease globally, posing a significant burden in Brazil and its Amazonian region. This study focused on the "riverine municipalities" and hypothesizes the presence of TB clusters in the area. Wealso aimed to train a machine learning model to differentiate municipalities classified as hot spots vs. non-hot spots using disease surveillance variables as predictors. Data regarding the incidence of TB from 2019 to 2022 in the riverine town was collected from the Brazilian Health Ministry Informatics Department. Moran's I was used to assess global spatial autocorrelation, while the Getis-Ord GI* method was employed to detect high and low-incidence clusters. A Random Forest machine-learning model was trained using surveillance variables related to TB cases to predict hot spots among non-hot spot municipalities. Our analysis revealed distinct geographical clusters with high and low TB incidence following a west-to-east distribution pattern. The Random Forest Classification model utilizes six surveillance variables to predict hot vs. non-hot spots. The machine learning model achieved an Area Under the Receiver Operator Curve (AUC-ROC) of 0.81. Municipalities with higher percentages of recurrent cases, deaths due to TB, antibiotic regimen changes, percentage of new cases, and cases with smoking history were the best predictors of hot spots. This prediction method can be leveraged to identify the municipalities at the highest risk of being hot spots for the disease, aiding policymakers with an evidenced-based tool to direct resource allocation for disease control in the riverine municipalities.

Signature extension as a machine learning strategy for mapping plantation forest genera with Sentinel-2 imagery

Plantation forest inventory data such as tree genus is important for supporting forestry decisions and policymaking, but such data are traditionally collected in-field, which is time-consuming and costly. Although machine learning and remote sensing technologies have shown great potential to reduce the time and effort for mapping forest plantation genera at regional scales, they rely on training (labelled) data, which are also costly to collect over large areas. One approach to reducing the effort of training data collection is to make use of signature extension, whereby training data collected in one area is used to train and apply a machine learning model in a different area. This study aimed to evaluate the viability of training data signature extension for constructing random forest (RF) machine learning models to differentiate between acacia, eucalyptus and pine trees using Sentinel-2 imagery as input. The study was carried out over a large area (about 4920 km2) in South Africa. The study area was divided into 19 tiles of 100 × 100 km (each tile coincides with the footprint of a Sentinel-2 tile) from which three were chosen for sourcing (collecting) training data. Four experiments were conducted. In the first experiment, a fixed number (3000 per genera) of training samples were collected in the first source tile and used to build an RF model. The resulting model was then applied and assessed in all 19 tiles. This protocol was repeated in the second and third experiments using the training data collected in the other two source tiles respectively. In the final experiment, training data from all three source tiles were combined and applied to all 19 tiles. The mean overall classification accuracy of each classified tile was compared to the extension distance (i.e. the distance between the target and source tile), differences in rainfall seasonality, and variation in the mean annual temperature among tiles to gain an understanding of how signature extension efficiency is influenced by distance and environmental conditions. The results show that signature extension is viable (∼70% overall accuracies) over distances of up to 500 km, but only if the source and target tiles represent areas with similar rainfall regimes.

Predicting Asthma Exacerbation Risk in the Adult South Korean Population Using Integrated Health Data and Machine Learning Models.

Asthma is a chronic inflammatory airway disease with significant burden; exacerbations can severely affect quality of life and healthcare costs. Advances in big data analysis and artificial intelligence have made it easier to predict future exacerbations more accurately. This study used an integrated dataset of Korean National Health Insurance, meteorological, air pollution, and viral data from national public databases to develop a model to predict asthma exacerbations on a daily basis in South Korea. We merged these sources and applied random forest, AdaBoost, XGBoost, and LightGBM machine learning models to compare their performances at predicting future exacerbations. Of the models, XGBoost (AUROC of 0.68 and accuracy of 0.96) and LightGBM (AUROC of 0.67 and accuracy of 0.96) were the most promising. Common important variables were the number of visits and exacerbations per year, and medical resource utilization, including the prescription of asthma medications. Comorbid diabetes, hypertension, gastroesophageal reflux, arthritis, metabolic syndrome, osteoporosis, and ischemic heart disease were also associated with elevated exacerbation risk. The models examined in this study highlight the importance of previous exacerbations, use of medical resources, and comorbidities in the prediction of future exacerbations in patients with asthma.

Machine learning based forest species susceptibility mapping using Sentinel 2A data and GIS: A case study of Ait Bouzid forest (Central High Atlas, Morocco)

In arid and semi-arid environments, producing accurate maps of forest tree cover using optical remote sensing data is essential to understand their spatial distributions and dynamics. In this respect, the current study aimed to explore the effectiveness of support vector machine (SVM), K nearest neighbors (KNN), and random forest (RF)machine learning (ML) models to map the forest tree species of Ait Bouzid region (Central High Atlas, Morocco) by using Sentinel-2A data. The results from all models showed that about 19-28%, 21-27%, 16-24%, 15-18%, and 0,3-0,32% of the area was covered by euphorbia, red juniper, cedar, holm oak, bare ground, and water body, respectively. According to the overall accuracy (OA) and kappa coefficient, the SVM classifier showed the highest OA (73%) and kappa (0.66) values, followed by KNN (OA=70%, kappa=0.62) and RF (OA=67%, kappa=0.59). Regarding LC classes, water, bare soil, and holm oak could be identified with the producer's accuracy attaining 100%, while red juniper and cedar were the most challenging classes to determine for all ML classifiers, with the producer's accuracy of 40-50% and 40-67%. This study revealed the potential of ML approaches coupled with multispectral Sentinel-2A data for forest species cartography in arid areas with high accuracy. Furthermore, it provides crucial information about forest tree species distribution for developing forest management plans.