Baseline Machine Learning Models Research Articles

Collaborative artificial intelligence system for investigation of healthcare claims compliance

Healthcare fraud, waste and abuse are costly problems that have huge impact on society. Traditional approaches to identify non-compliant claims rely on auditing strategies requiring trained professionals, or on machine learning methods requiring labelled data and possibly lacking interpretability. We present Clais, a collaborative artificial intelligence system for claims analysis. Clais automatically extracts human-interpretable rules from healthcare policy documents (0.72 F1-score), and it enables professionals to edit and validate the extracted rules through an intuitive user interface. Clais executes the rules on claim records to identify non-compliance: on this task Clais significantly outperforms two baseline machine learning models, and its median F1-score is 1.0 (IQR = 0.83 to 1.0) when executing the extracted rules, and 1.0 (IQR = 1.0 to 1.0) when executing the same rules after human curation. Professionals confirm through a user study the usefulness of Clais in making their workflow simpler and more effective.

HBCVTr: an end-to-end transformer with a deep neural network hybrid model for anti-HBV and HCV activity predictor from SMILES

Hepatitis B and C viruses (HBV and HCV) are significant causes of chronic liver diseases, with approximately 350 million infections globally. To accelerate the finding of effective treatment options, we introduce HBCVTr, a novel ligand-based drug design (LBDD) method for predicting the inhibitory activity of small molecules against HBV and HCV. HBCVTr employs a hybrid model consisting of double encoders of transformers and a deep neural network to learn the relationship between small molecules’ simplified molecular-input line-entry system (SMILES) and their antiviral activity against HBV or HCV. The prediction accuracy of HBCVTr has surpassed baseline machine learning models and existing methods, with R-squared values of 0.641 and 0.721 for the HBV and HCV test sets, respectively. The trained models were successfully applied to virtual screening against 10 million compounds within 240 h, leading to the discovery of the top novel inhibitor candidates, including IJN04 for HBV and IJN12 and IJN19 for HCV. Molecular docking and dynamics simulations identified IJN04, IJN12, and IJN19 target proteins as the HBV core antigen, HCV NS5B RNA-dependent RNA polymerase, and HCV NS3/4A serine protease, respectively. Overall, HBCVTr offers a new and rapid drug discovery and development screening method targeting HBV and HCV.

Graph Convolutional Capsule Regression (GCCR): A Model for Accelerated Filtering of Novel Potential Candidates for SARS-CoV-2 Based on Binding Affinity.

There has been a growing interest in discovering a viable drug for the new coronavirus (SARS-CoV-2) since the beginning of the pandemic. Protein-ligand interaction studies are a crucial step in the drug discovery process, as it helps us narrow the search space for potential ligands with high drug-likeness. Derivatives of popular drugs like Remdesivir generated through tools employing evolutionary algorithms are usually considered potential candidates. However, screening promising molecules from such a large search space is difficult. In a conventional screening process, for each ligand-target pair, there are time-consuming interaction studies that use docking simulations before downstream tasks like thermodynamic, kinetic, and electrostatic-potential evaluation. In this work, 'Graph Convolutional Capsule Regression' (GCCR), a model which uses Capsule Neural Networks (CapsNet) and Graph Convolutional Networks (GCN) to predict the binding energy of a protein-ligand complex is being proposed. The model's predictions were further validated with kinetic and free energy studies like Molecular Dynamics (MD) for kinetic stability and MM/GBSA analysis for free energy calculations. The GCCR showed an RMSE value of 0.0978 for 81.3% of the concordance index. The RMSE of GCCR converged around the iteration of just 50 epochs scoring a lower RMSE than GCN and GAT. When training with Davis Dataset, GCCR gave an RMSE score of 0.3806 with a CI score of 87.5%. The proposed GCCR model shows great potential in improving the screening process based on binding affinity and outperforms baseline machine learning models like DeepDTA, KronRLS, SimBoost, and other Graph Neural Networks (GNN) based models like Graph Convolutional Networks (GCN) and Graph Attention Networks (GAT).

A spatial–temporal deep learning-based warning system against flooding hazards with an empirical study in Taiwan

With climate change intensifying, increasingly extreme rainfall events which lead to flooding are occurring at a higher frequency. As such, it is crucial to develop and implement useful flood forecasting models which can provide early knowledge of the severity of flood risks and allow decision makers to take preventive and response actions which reduce human casualties. This study, carried out in collaboration with the National Science and Technology Center for Disaster Reduction (NCDR) of Taiwan, utilizes deep learning techniques combined with attention mechanisms to establish a data-driven flood forecasting framework which can serve as an early warning system for flooding hazards. The overall framework, which utilizes historical flood depth data from an array of flood sensors while engineering additional features from KRID (estimation Rainfall with Integrated radar-gauge Data) observed rainfall data consists of two phases. The first phase is a flood event detection model built using a classification-based random forest algorithm. The second phase is a flood depth prediction model which utilizes a spatial–temporal attention mechanism combined with a transformation-gated long short-term memory (LSTM).This novel model, known as the Spatial–Temporal Attention Transformation-Gated LSTM (STA-TGLSTM), is capable of predicting flood levels over the next three hours. In an empirical study based in Changhua County, Taiwan, STA-TGLSTM achieves the lowest RMSE losses for predicting the next three hours of flood levels compared to baseline machine learning models. Notably, the accuracy of predicting the flood sensor depth for the first hour (T+1) reaches 76% when STA-TGLSTM is applied. In comparison to traditional hydrological models, this data-driven model offers greater computational efficiency, providing future flood depth information within seconds. Moreover, as demonstrated in the empirical study, by visualizing the weight values of the spatial–temporal attention mechanism, valuable insights into the relationship between each flood sensor and the spatiotemporal variables can be readily gained.

Breaking barriers: a statistical and machine learning-based hybrid system for predicting dementia.

Introduction: Dementia is a condition (a collection of related signs and symptoms) that causes a continuing deterioration in cognitive function, and millions of people are impacted by dementia every year as the world population continues to rise. Conventional approaches for determining dementia rely primarily on clinical examinations, analyzing medical records, and administering cognitive and neuropsychological testing. However, these methods are time-consuming and costly in terms of treatment. Therefore, this study aims to present a noninvasive method for the early prediction of dementia so that preventive steps should be taken to avoid dementia. Methods: We developed a hybrid diagnostic system based on statistical and machine learning (ML) methods that used patient electronic health records to predict dementia. The dataset used for this study was obtained from the Swedish National Study on Aging and Care (SNAC), with a sample size of 43040 and 75 features. The newly constructed diagnostic extracts a subset of useful features from the dataset through a statistical method (F-score). For the classification, we developed an ensemble voting classifier based on five different ML models: decision tree (DT), naive Bayes (NB), logistic regression (LR), support vector machines (SVM), and random forest (RF). To address the problem of ML model overfitting, we used a cross-validation approach to evaluate the performance of the proposed diagnostic system. Various assessment measures, such as accuracy, sensitivity, specificity, receiver operating characteristic (ROC) curve, and Matthew's correlation coefficient (MCC), were used to thoroughly validate the devised diagnostic system's efficiency. Results: According to the experimental results, the proposed diagnostic method achieved the best accuracy of 98.25%, as well as sensitivity of 97.44%, specificity of 95.744%, and MCC of 0.7535. Discussion: The effectiveness of the proposed diagnostic approach is compared to various cutting-edge feature selection techniques and baseline ML models. From experimental results, it is evident that the proposed diagnostic system outperformed the prior feature selection strategies and baseline ML models regarding accuracy.

Optimization of RNN-LSTM Model Using NSGA-II Algorithm for IOT-based Fire Detection Framework

The occurrence of fire leads to unparalleled loss of resources as well as human life and hence, fire detection systems must be trustworthy and less erroneous. Real-time assessment of fire conditions through predictive learning models could lead to easier decision-making and timely rescue operations. Reported works are often restricted to the use of singular rule-based algorithms which can hardly offer a comprehensive solution by adapting to changing dynamics of fire conditions due to their static features, mostly leading to inaccurate classification. In this research, an efficient fire prediction framework has been proposed by efficiently combining the outputs of recurrent neural networks which bear the advantages of short-term predictions with long short-term memory that takes care of long-term predictions. Weights to combine multi-objective optimization functions to minimize mean absolute error and root mean square error have been designed using non-dominated sorting genetic algorithm II (NSGA-II) which offers a lower time complexity. The benchmark dataset from the NIST website has been chosen for analysis and performance validation of the proposed classifier on experimental data pertaining to different fire scenarios. A Pareto optimal front has been obtained from the proposed algorithm which represents the optimum solutions. The performance of the proposed model has been exhaustively evaluated through different factors such as accuracy, RMSE, MAE, F-Measure, binary classification rate, negative predictive value, recall and precision which justifies its contribution. The proposed model reduced RMSE by 14.95–19.88% compared to baseline machine learning models along with an enhanced accuracy of 95.05% and reduced false positive rate which is better compared to reported works along with improvement in F-Measure. Results show that the proposed NSGA-II-based RNN LSTM model accurately predicts the occurrence of fire events with reduced false alarms while maintaining a low computational overhead.

Scientific reference style using rule-based machine learning

Regular Expressions (RegEx) can be employed as a technique for supervised learning to define and search for specific patterns inside text. This work devised a method that utilizes regular expressions to convert the reference style of academic papers into several styles, dependent on the specific needs of the target publication or conference. Our research aimed to detect distinctive patterns of reference styles using RegEx and compare them with a dataset including various reference styles. We gathered a diverse range of reference format categories, encompassing seven distinct classes, from various sources such as academic papers, journals, conference proceedings, and books. Our approach involves employing RegEx to convert one referencing format to another based on the user's specific preferences. The proposed model demonstrated an accuracy of 57.26% for book references and 57.56% for journal references. We used the similarity ratio and Levenshtein distance to evaluate the dataset's performance. The model achieved a 97.8% similarity ratio with a Levenshtein distance of 2. Notably, the APA style for journal references yielded the best results. However, the effectiveness of the extraction function varies depending on the reference style. For APA style, the model showed a 99.97% similarity ratio with a Levenshtein distance of 1. Overall, our proposed model outperforms baseline machine learning models in this task. This study introduces an automated program that utilizes regular expressions to modify academic reference formats. This will enhance the efficiency, precision, and adaptability of academic publishing.

Designing NLP applications to support ICD coding: an impact analysis and guidelines to enhance baseline performance when processing patient discharge notes

Financial costs are a major concern in the healthcare system, with medical billing and coding playing a key role in facilitating transactions and financing procedures. Billing involves filing claims with insurance companies and requires scrutiny of clinical summaries and electronic health records to correctly match diagnoses, prescriptions, and procedures to standardized codes. Accuracy in assigning International Classification of Diseases (ICD) codes is critical to proper reimbursement of care. Incorrect codes waste time and resources, and cause administrative and financial problems for hospitals, insurance companies and patients. Manual medical coding is a labor-intensive and error-prone process that creates additional administrative burden and inconvenience for hospitals, insurance companies, and patients. To simplify the process, clinical records are often processed to automatically identify and extract clinical concepts and corresponding ICD codes. Deep learning and natural language processing techniques have shown promise in a variety of tasks but applying them to medical coding has been challenging. Accurate coding requires a deep understanding of medical terminology, context, and guidelines that may be difficult to capture with traditional deep learning methods. Although deep learning shows promise in healthcare, its specific impact on ICD coding is not fully understood, and translating scalable deep learning methods into practical improvements in ICD coding remains a challenge. Evaluating deep learning models under the scenarios of real-world coding and comparing them to established practice is critical to determining their true effectiveness. In this work, we address the automation of ICD coding by highlighting pitfalls and contrasting different perspectives. We investigated automatic ICD coding using baseline machine learning models, with a focus on identifying ICD-9 codes in discharge notes from Medical Information Mart for Intensive Care (MIMIC) database. A thorough evaluation of different models and approaches is crucial to avoid over-reliance on any method. Our findings show that simpler methods can achieve comparable results to deep learning models while still requiring fewer computational resources.

LAUREATE

The latest developments in wearable sensors have resulted in a wide range of devices available to consumers, allowing users to monitor and improve their physical activity, sleep patterns, cognitive load, and stress levels. However, the lack of out-of-the-lab labelled data hinders the development of advanced machine learning models for predicting affective states. Furthermore, to the best of our knowledge, there are no publicly available datasets in the area of Human Memory Augmentation. This paper presents a dataset we collected during a 13-week study in a university setting. The dataset, named LAUREATE, contains the physiological data of 42 students during 26 classes (including exams), daily self-reports asking the students about their lifestyle habits (e.g. studying hours, physical activity, and sleep quality) and their performance across multiple examinations. In addition to the raw data, we provide expert features from the physiological data, and baseline machine learning models for estimating self-reported affect, models for recognising classes vs breaks, and models for user identification. Besides the use cases presented in this paper, among which Human Memory Augmentation, the dataset represents a rich resource for the UbiComp community in various domains, including affect recognition, behaviour modelling, user privacy, and activity and context recognition.

Semantic-aware quality assessment of building elements using graph neural networks

Automated construction quality assessments often face challenges due to noise and occlusions in on-site data when independently comparing elements in as-designed building information models (ad-BIMs) with registered 3D point clouds. This paper proposes BIM-Graph Neural Network (BIM-GNN), a learning-based approach that leverages semantics in ad-BIM to enhance element-wise quality assessments. BIM-GNN transforms ad-BIM into a graph, where nodes represent BIM objects and edges capture their topological and spatial relationships. The node features incorporate ad-BIM semantics, scan alignments, and graph structure information, while edge attributes describe relationship types. BIM-GNN employs transductive learning to classify nodes as verified, deviated, missing, or no-data, capturing project-level context. Experimental results show that BIM-GNN outperforms baseline machine learning models by over 27% on the median weighted F1-Score and provides the ability to infer the status of partially observed and unobserved elements. This work inspires future research on semantic-aware automated construction control.

Automated Machine Learning to Evaluate the Information Content of Tropospheric Trace Gas Columns for Fine Particle Estimates Over India: A Modeling Testbed

AbstractIndia is largely devoid of high‐quality and reliable on‐the‐ground measurements of fine particulate matter (PM2.5). Ground‐level PM2.5 concentrations are estimated from publicly available satellite Aerosol Optical Depth (AOD) products combined with other information. Prior research has largely overlooked the possibility of gaining additional accuracy and insights into the sources of PM using satellite retrievals of tropospheric trace gas columns. We evaluate the information content of tropospheric trace gas columns for PM2.5 estimates over India within a modeling testbed using an Automated Machine Learning (AutoML) approach, which selects from a menu of different machine learning tools based on the data set. We then quantify the relative information content of tropospheric trace gas columns, AOD, meteorological fields, and emissions for estimating PM2.5 over four Indian sub‐regions on daily and monthly time scales. Our findings suggest that, regardless of the specific machine learning model assumptions, incorporating trace gas modeled columns improves PM2.5 estimates. We use the ranking scores produced from the AutoML algorithm and Spearman’s rank correlation to infer or link the possible relative importance of primary versus secondary sources of PM2.5 as a first step toward estimating particle composition. Our comparison of AutoML‐derived models to selected baseline machine learning models demonstrates that AutoML is at least as good as user‐chosen models. The idealized pseudo‐observations (chemical‐transport model simulations) used in this work lay the groundwork for applying satellite retrievals of tropospheric trace gases to estimate fine particle concentrations in India and serve to illustrate the promise of AutoML applications in atmospheric and environmental research.

Using graph neural networks for site-of-metabolism prediction and its applications to ranking promiscuous enzymatic products.

While traditionally utilized for identifying site-specific metabolic activity within a compound to alter its interaction with a metabolizing enzyme, predicting the site-of-metabolism (SOM) is essential in analyzing the promiscuity of enzymes on substrates. The successful prediction of SOMs and the relevant promiscuous products has a wide range of applications that include creating extended metabolic models (EMMs) that account for enzyme promiscuity and the construction of novel heterologous synthesis pathways. There is therefore a need to develop generalized methods that can predict molecular SOMs for a wide range of metabolizing enzymes. This article develops a Graph Neural Network (GNN) model for the classification of an atom (or a bond) being an SOM. Our model, GNN-SOM, is trained on enzymatic interactions, available in the KEGG database, that span all enzyme commission numbers. We demonstrate that GNN-SOM consistently outperforms baseline machine learning models, when trained on all enzymes, on Cytochrome P450 (CYP) enzymes, or on non-CYP enzymes. We showcase the utility of GNN-SOM in prioritizing predicted enzymatic products due to enzyme promiscuity for two biological applications: the construction of EMMs and the construction of synthesis pathways. A python implementation of the trained SOM predictor model can be found at https://github.com/HassounLab/GNN-SOM. Supplementary data are available at Bioinformatics online.

Prediction of ultimate condition of FRP-confined recycled aggregate concrete using a hybrid boosting model enriched with tabular generative adversarial networks

Despite its multiple benefits, recycled aggregate concrete (RAC) usually exhibits inferior properties compared with natural aggregate concrete, which has been deemed as a hurdle to its widespread use. One way to overcome this roadblock is to apply FRP jacketing. However, it appears not easy to quantify the axial performance of FRP-confined recycled aggregate concrete columns (FRACC), due largely to the complex load-resisting mechanisms involved. Also, the weakness of RAC itself further compound the problem. This paper aspires to deliver an alternative means to address this difficulty. A powerful boosting approach, XGBoost, was developed to fulfill the goal, where its hyperparameters was fine-tuned by a beetle antennae search metaheuristic algorithm. Meanwhile, a synthetic data generator, tabular generative adversarial networks, was introduced to supplement the limited training data. The model developed outperformed existing empirical equations and several baseline machine learning models. Teng et al.’s FRP-confined concrete model was also improved for better tracing the axial stress–strain behavior. Besides, interpreting the model allows better understanding of the underlying mechanisms such as the minimum reinforcement ratio of FRP required to mitigate the negative effects of RAC. Finally, two data-driven, explicit design equations are given for practical design of FRACC.

Multi-dimensional patient acuity estimation with longitudinal EHR tokenization and flexible transformer networks.

Transformer model architectures have revolutionized the natural language processing (NLP) domain and continue to produce state-of-the-art results in text-based applications. Prior to the emergence of transformers, traditional NLP models such as recurrent and convolutional neural networks demonstrated promising utility for patient-level predictions and health forecasting from longitudinal datasets. However, to our knowledge only few studies have explored transformers for predicting clinical outcomes from electronic health record (EHR) data, and in our estimation, none have adequately derived a health-specific tokenization scheme to fully capture the heterogeneity of EHR systems. In this study, we propose a dynamic method for tokenizing both discrete and continuous patient data, and present a transformer-based classifier utilizing a joint embedding space for integrating disparate temporal patient measurements. We demonstrate the feasibility of our clinical AI framework through multi-task ICU patient acuity estimation, where we simultaneously predict six mortality and readmission outcomes. Our longitudinal EHR tokenization and transformer modeling approaches resulted in more accurate predictions compared with baseline machine learning models, which suggest opportunities for future multimodal data integrations and algorithmic support tools using clinical transformer networks.

ClimateBench v1.0: A Benchmark for Data‐Driven Climate Projections

AbstractMany different emission pathways exist that are compatible with the Paris climate agreement, and many more are possible that miss that target. While some of the most complex Earth System Models have simulated a small selection of Shared Socioeconomic Pathways, it is impractical to use these expensive models to fully explore the space of possibilities. Such explorations therefore mostly rely on one‐dimensional impulse response models, or simple pattern scaling approaches to approximate the physical climate response to a given scenario. Here we present ClimateBench—the first benchmarking framework based on a suite of Coupled Model Intercomparison Project, AerChemMIP and Detection‐Attribution Model Intercomparison Project simulations performed by a full complexity Earth System Model, and a set of baseline machine learning models that emulate its response to a variety of forcers. These emulators can predict annual mean global distributions of temperature, diurnal temperature range and precipitation (including extreme precipitation) given a wide range of emissions and concentrations of carbon dioxide, methane and aerosols, allowing them to efficiently probe previously unexplored scenarios. We discuss the accuracy and interpretability of these emulators and consider their robustness to physical constraints such as total energy conservation. Future opportunities incorporating such physical constraints directly in the machine learning models and using the emulators for detection and attribution studies are also discussed. This opens a wide range of opportunities to improve prediction, robustness and mathematical tractability. We hope that by laying out the principles of climate model emulation with clear examples and metrics we encourage engagement from statisticians and machine learning specialists keen to tackle this important and demanding challenge.

GNPI: Graph normalization to integrate phylogenetic information for metagenomic host phenotype prediction

Microorganisms play important roles in our lives especially on metabolism and diseases. Determining the probability of human suffering from specific diseases and the severity of the disease based on microbial genes is the crucial research for understanding the relationship between microbes and diseases. Previous could extract the topological information of phylogenetic trees and integrate them to metagenomic datasets, thus enable classifiers to learn more information in limited datasets and thus improve the performance of the models. In this paper, we proposed a GNPI model to better learn the structure of phylogenetic trees. GNPI maintained the original vector format of metagenomic datasets, while previous research had to change the input form to matrices. The vector-like form of the input data can be easily adopted in the baseline machine learning models and is available for deep learning models. The datasets processed with GNPI help enhance the accuracy of machine learning and deep learning models in three different datasets. GNPI is an interpretable data processing method for host phenotype prediction and other bioinformatics tasks.

MedML: Fusing medical knowledge and machine learning models for early pediatric COVID-19 hospitalization and severity prediction

SummaryThe COVID-19 pandemic has caused devastating economic and social disruption. This has led to a nationwide call for models to predict hospitalization and severe illness in patients with COVID-19 to inform the distribution of limited healthcare resources. To address this challenge, we propose a machine learning model, MedML, to conduct the hospitalization and severity prediction for the pediatric population using electronic health records. MedML extracts the most predictive features based on medical knowledge and propensity scores from over 6 million medical concepts and incorporates the inter-feature relationships in medical knowledge graphs via graph neural networks. We evaluate MedML on the National Cohort Collaborative (N3C) dataset. MedML achieves up to a 7% higher AUROC and 14% higher AUPRC compared to the best baseline machine learning models. MedML is a new machine learnig framework to incorporate clinical domain knowledge and is more predictive and explainable than current data-driven methods.

An improved machine learning technique for identify informative COVID-19 tweets

Twitter users are increasingly using the platform to share information, particularly in the case of disease outbreaks such as COVID-19. It's difficult to find informative tweets about coronavirus on Twitter. Recognizing tweets associated with disease evaluation in social media is a critical endeavour because it is a subset of associated data. Existing works rely solely on subject identification, vocabulary construction, idea extraction, polarity detection, descriptive Terms, and disease-related statistical characteristics, resulting in a lack of precision in detecting tweet content. To solve this problem, this study used parts of speech tags and high-resolution graphics. To address this issue, we proposed an IPSH (Informative POS statistical High Frequency) model for predicting COVID-19 tweet content that incorporates parts of speech tags and high-frequency words as features into the existing machine learning model. The model was found to be more efficient when compared to baseline machine learning models using the Twitter COVID-19 disease dataset.

Cryptocurrency malware detection in real-world environment: Based on multi-results stacking learning

Cryptocurrency mining malware (CryptocMal) has been proliferating due to its high profitability and anonymity. There are many studies using machine learning methods to build CryptocMal detectors and other malware detectors. However, these detection methods tend to test their performance on small datasets only, and such testing conditions make researchers often doubt the real-world performance of these machine learning methods. Some deep learning methods that do not require expert knowledge of malware further reinforce this suspicion. In this paper, different from previous studies, the heuristic rule features set for the machine learning model are designed based on CryptocMal characteristics. Furthermore, the heuristic rule features are integrated as a domain knowledge component in an ensemble learning framework, called CMalHunt. CMalHunt utilized the stacking method to combine results of domain knowledge features, behavior features and binary bytes features. Through integrating classification models with different feature types, the experimental results show that CMalHunt significantly outperforms the baseline machine learning models. These results also validate our conjecture about feature types integration, indicating that each feature can play a role in the CryptocMal detection task. This paper is informative for real-world applications of machine learning in malware identification and malware family classification.

How to get best predictions for road monitoring using machine learning techniques.

Road condition monitoring is essential for improving traffic safety and reducing accidents. Machine learning methods have recently gained prominence in the practically important task of controlling road surface quality. Several systems have been proposed using sensors, especially accelerometers present in smartphones due to their availability and low cost. However, these methods require practitioners to specify an exact set of features from all the sensors to provide more accurate results, including the time, frequency, and wavelet-domain signal features. It is important to know the effect of these features change on machine learning model performance in handling road anomalies classification tasks. Thus, we address such a problem by conducting a sensitivity analysis of three machine learning models which are Support Vector Machine, Decision Tree, and Multi-Layer Perceptron to test the effectiveness of the model by selecting features. We built a feature vector from all three axes of the sensors that boosts classification performance. Our proposed approach achieved an overall accuracy of 94% on four types of road anomalies. To allow an objective analysis of different features, we used available accelerometer datasets. Our objective is to achieve a good classification performance of road anomalies by distinguishing between significant and relatively insignificant features. Our chosen baseline machine learning models are based on their comparative simplicity and powerful empirical performance. The extensive analysis results of our study provide practical advice for practitioners wishing to select features effectively in real-world settings for road anomalies detection.