MethodsX最新文献

{"title":"Enhancing rotary dryer efficiency: Adjusting blower speed for optimal temperature stability in coffee bean drying","authors":"Ika Noer Syamsiana ,&nbsp;Nihayatun Nafisah ,&nbsp;Arwin Datumaya Wahyudi Suma ,&nbsp;Nilawati Fiernaningsih ,&nbsp;Rahma Nur Amalia ,&nbsp;Wahyu Aulia Nurwicaksana","doi":"10.1016/j.mex.2025.103477","DOIUrl":"10.1016/j.mex.2025.103477","url":null,"abstract":"<div><div>This study evaluates the quality of coffee beans dried using a rotary dryer with Fuzzy Logic Control (FLC) for temperature with airspeed regulation. Laboratory tests assessed gravimetric water content, acidity levels, and antioxidant content to determine the effects of drying parameters. The FLC system effectively stabilized the drying temperature at a setpoint of 60 °C, with errors below 0.17 % under and 2 % above the setpoint. This precise control enhanced drying efficiency while maintaining the quality standards of coffee beans, achieving a moisture content below 12.5 % (SNI 01–2907–2008). Results showed that coffee beans dried at 270 min (1 kg mass) achieved a water content of 11.28 %, while 330 min (2 kg mass) reduced the water content to 9.92 %. For 450 min (3 kg mass), the water content stabilized at 10.92 %. Acidity and antioxidant levels remained optimal across all tests. This study demonstrates the effectiveness of fuzzy logic-based rotary dryers for efficient and high-quality coffee bean processing, regardless of drying mass.</div><div>Key Methodology:<ul><li><span>•</span><span><div>Developed a fuzzy logic control system for real-time blower speed to regulated temperature adjustments.</div></span></li><li><span>•</span><span><div>Stabilized hot air distribution to reduce temperature deviations and relative humidity.</div></span></li><li><span>•</span><span><div>Conducted gravimetric, acidity, and antioxidant analyses on coffee beans with varying drying times and masses.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103477"},"PeriodicalIF":1.6,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144535815","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrated deep learning for cardiovascular risk assessment and diagnosis: An evolutionary mating algorithm-enhanced CNN-LSTM","authors":"Ahmed Mohammed Ahmed Alsarori,&nbsp;Mohd Herwan Sulaiman","doi":"10.1016/j.mex.2025.103466","DOIUrl":"10.1016/j.mex.2025.103466","url":null,"abstract":"<div><div>Cardiovascular diseases (CVD) remain the leading cause of mortality worldwide, emphasizing the urgent need for accurate and efficient predictive models. This study proposes a dual-output deep learning model based on a hybrid Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM) model, optimized using the Evolutionary Mating Algorithm (EMA). The model predicts both a continuous risk score and a binary diagnostic outcome, supporting both quantitative assessment and early clinical decision-making. EMA was applied for hyperparameter optimization, demonstrating improved convergence and generalization over conventional methods. Performance was benchmarked against CNN-LSTM models optimized using Particle Swarm Optimization (PSO) and Barnacle Mating Optimization (BMO). The EMA-based model achieved superior results, with a Mean Absolute Error (MAE) of 0.018, Mean Squared Error (MSE) of 0.0006, Root Mean Squared Error (RMSE) of 0.024, and a coefficient of determination (R²) of 0.98 for risk prediction. For the diagnostic task, the model attained 70 % accuracy and 80 % precision. These findings validate EMA’s effectiveness in tuning dual-output deep learning models and highlight its potential in enhancing cardiovascular risk stratification and early diagnosis in clinical settings.<ul><li><span>•</span><span><div>Dual-output CNN-LSTM model optimized using EMA.</div></span></li><li><span>•</span><span><div>Continuous risk scores and binary diagnostic classification predictions.</div></span></li><li><span>•</span><span><div>EMA outperformed PSO and BMO in predictive accuracy and model robustness.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103466"},"PeriodicalIF":1.6,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144518793","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"eHealth interventions for chronic pain: Protocol for a systematic review and meta-analysis","authors":"Rosario Caruso ,&nbsp;Pier Mario Perrone ,&nbsp;Karen Barros Parron Fernandes ,&nbsp;Cristina Arrigoni ,&nbsp;Arianna Magon ,&nbsp;Gianluca Conte ,&nbsp;Silvia Belloni ,&nbsp;Silvana Castaldi","doi":"10.1016/j.mex.2025.103475","DOIUrl":"10.1016/j.mex.2025.103475","url":null,"abstract":"<div><div>Chronic pain is a prevalent and costly condition that significantly impairs functional and emotional status as well as quality of life, representing a major cause of incapacity worldwide that requires long-term management strategies.</div><div>eHealth interventions, including telemedicine, mobile health applications, and internet-based programs, have emerged as promising approaches to improve pain management by enhancing access to education, psychological support, and self-monitoring tools. However, the efficacy of these interventions remains unclear due to variability in study designs or intervention components, differences in pain conditions (e.g., somatic or neuropathic pain), and outcome measures. This systematic review and meta-analysis aims to synthesize the evidence on the efficacy of eHealth interventions for chronic non-cancer pain in adults, assessing their impact on pain severity, functional outcomes, quality of life, mental health, medication adherence, patient satisfaction, and cost-effectiveness. A network meta-analysis (NMA) will be conducted to compare different eHealth modalities and identify which features contribute most to positive outcomes. Statistical analyses will follow Cochrane guidelines, with the risk of bias and evidence certainty evaluated using the Cochrane RoB 2.0 tool and GRADE framework, respectively. The findings will provide valuable insights for clinicians, policymakers, and researchers, informing best practices for integrating digital health solutions into chronic pain management.<ul><li><span>•</span><span><div>This systematic review and meta-analysis will assess the efficacy of eHealth interventions for chronic pain management.</div></span></li><li><span>•</span><span><div>A network meta-analysis will compare different digital health modalities to identify the most effective intervention components.</div></span></li><li><span>•</span><span><div>The study will evaluate clinical, functional, psychological, adherence, and cost-related outcomes, informing evidence-based practice and policy.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103475"},"PeriodicalIF":1.6,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144535813","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"QCAE-QOC-SVM: A hybrid quantum machine learning model for DoS and Fuzzy attack detection on autonomous vehicle CAN bus","authors":"Meghana R,&nbsp;Sowmyashree Sakrepatna Ramesha,&nbsp;Adwitiya Mukhopadhyay","doi":"10.1016/j.mex.2025.103471","DOIUrl":"10.1016/j.mex.2025.103471","url":null,"abstract":"<div><div>In this study, we introduce a hybrid quantum machine learning method to identify Normal signal, DoS, and Fuzzy attacks on the CAN bus utilized in autonomous vehicles. Our approach is a combination of a Quantum Convolutional Autoencoder (QCAE) and a Quantum Orthogonal Classifier based on Support Vector Machines (QOC-SVM). The method effectively extracts patterns from CAN bus traffic with the help of quantum-powered classification for accurate anomaly detection. The model was assessed using a public and custom dataset of 300,000 instances generated through the CARLA simulator and was run on a high-performance computing facility. Results from the experiments show that the QCAE-QOC-SVM model performs better than conventional machine learning (ML), deep learning (DL), and other quantum machine learning (QML) models with an F1 score of 99.43 % when the batches-to-batch size ratio is 7741:31. These findings indicate the possibility of quantum machine learning to significantly improve strong defense mechanisms against cyber-attacks for intelligent transportation systems. The high accuracy and resistance of the model proposed indicate good prospects for real-time autonomous vehicle security, with enhanced detection of sophisticated attack patterns. Our contribution is substantial in the creation of future-proof cybersecurity solutions for the fast-changing autonomous vehicle technology and intelligent transportation system domain.<ul><li><span>•</span><span><div>Introduction of a hybrid quantum machine learning model for attack detection on autonomous vehicle CAN buses.</div></span></li><li><span>•</span><span><div>Demonstrated superior performance with an F1 score of 99.43 % compared to traditional ML, DL, and QML models.</div></span></li><li><span>•</span><span><div>Showed the potential of quantum machine learning in strengthening defense systems for intelligent transportation networks.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103471"},"PeriodicalIF":1.6,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144535814","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Proximal Policy Optimization-based Task Offloading Framework for Smart Disaster Monitoring using UAV-assisted WSNs","authors":"C.N. Vanitha ,&nbsp;P. Anusuya ,&nbsp;Rajesh Kumar Dhanaraj ,&nbsp;Dragan Pamucar ,&nbsp;Mahmoud Ahmad Al-Khasawneh","doi":"10.1016/j.mex.2025.103472","DOIUrl":"10.1016/j.mex.2025.103472","url":null,"abstract":"<div><div>Unmanned Aerial Vehicles (UAVs) are increasingly employed in Wireless Sensor Networks (WSNs) to enhance communication, coverage, and energy efficiency, particularly in disaster monitoring and remote surveillance scenarios. However, challenges such as limited energy resources, dynamic task allocation, and UAV trajectory optimization remain critical. This paper presents Energy-efficient Task Offloading using Reinforcement Learning for UAV-assisted WSNs (ETORL-UAV), a novel framework that integrates Proximal Policy Optimization (PPO) based reinforcement learning to intelligently manage UAV-assisted operations in edge-enabled WSNs. The proposed approach utilizes a multi-objective reward model to adaptively balance energy consumption, task success rate, and network lifetime. Extensive simulation results demonstrate that ETORL-UAV outperforms five state-of-the-art methods Meta-RL, g-MAPPO, Backscatter Optimization, Hierarchical Optimization, and Game Theory based Pricing achieving up to 9.3 % higher task offloading success, 18.75 % improvement in network lifetime, and 27 % reduction in energy consumption. These results validate the framework's scalability, reliability, and practical applicability for real-world disaster-response WSN deployments.<ul><li><span>•</span><span><div>Proposes ETORL-UAV: Energy-efficient Task Offloading using Reinforcement Learning for UAV-assisted WSNs</div></span></li><li><span>•</span><span><div>Leverages PPO-based reinforcement learning and a multi-objective reward model</div></span></li><li><span>•</span><span><div>Demonstrates superior performance over five benchmark approaches in disaster-response simulations</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103472"},"PeriodicalIF":1.6,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144511150","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"IoT security approach based random distribution of communication frequency","authors":"Manga Tonga V Joseph Salomon ,&nbsp;Mpoh Lobe Henri Pascal-Will ,&nbsp;Malong Yannick ,&nbsp;Maka Maka Ebenezer ,&nbsp;Ihonock Eyembe Luc ,&nbsp;Essiben Dikoundou Jean-François ,&nbsp;Paul-Marie Moulema Douala ,&nbsp;Yong Suk Joe","doi":"10.1016/j.mex.2025.103465","DOIUrl":"10.1016/j.mex.2025.103465","url":null,"abstract":"<div><div>IoT is a technology that can be found everywhere in our daily lives. The bulk of Internet of Things (IoT) technologies relies on wireless transmissions; however, these are often subjected to security threats such as communication intercepts. The aims of this paper is to address these threats by suggesting a method based on a random distribution of communication frequencies. The proposed approach was tested using LoRa technology within an experimental testbed. The validation results show that randomly changing communication frequencies introduce a processing delay of less than one second during transmission and reception. Furthermore, the experiments showed that the proposed strategy enhances coexistence with other LoRa devices. The performance achieved is promising for secure and efficient data transmission applications.<ul><li><span>•</span><span><div>The concentrate on the vulnerabilities that arise in the network of IoT systems at the physical layer</div></span></li><li><span>•</span><span><div>We propose a method to close this security gap by using a random allocation of communication frequency between LoRa cards.</div></span></li><li><span>•</span><span><div>The results show that the coexistence between IoT is possible and the random allocation of communication frequencies introduced less than a second in transmission and reception of packets.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103465"},"PeriodicalIF":1.6,"publicationDate":"2025-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144522804","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A study protocol of developing contemporary family pocketbook for preparing pregnancy and childbirth for Indonesian women","authors":"Dhiana Ayu Novitasari ,&nbsp;Novita Nur Rohma ,&nbsp;Harti Rahmi Aunurul Lisa ,&nbsp;Fitriana Murriya Ekawati","doi":"10.1016/j.mex.2025.103467","DOIUrl":"10.1016/j.mex.2025.103467","url":null,"abstract":"<div><div>Pregnancy and childbirth are two significant events for women and families, that readiness for these two events is linked to its successful outcomes. Worldwide, most of pregnancies are unplanned, which in some cases, limit the optimum pregnancy care affecting the prevention for maternal morbidity and mortality. This protocol aims to describe a method to develop a pocketbook for women and family for preparing pregnancy and childbirth. The study design follows the model of Research and Development (R&amp;D) using 4D development model: namely define, design, develop and disseminate. Data collection applied up-to focus group (FG) discussion with women, families and healthcare professionals in Bantul District Yogyakarta. The first and second FG will discuss about the importance of the pocketbook for women, and its expected content. The third-and fourth FG will finalize the book and its potential implementation in practice. Analysis of the data applied thematic models. The study period will take place from July 2024-July 2025. This study is expected to develop a comprehensive and easy-to-understand family pocketbook, containing important information related to pregnancy and childbirth preparation. Subsequently the pocketbook is expected to help prepare families’ readiness in supporting women during pregnancy and childbirth in Indonesia.</div><div>Three points:<ul><li><span>•</span><span><div>There is a need to prepare women and families about planned pregnancy and childbirth, aiming to provide information and appropriate arrangements.</div></span></li><li><span>•</span><span><div>This protocol aims to describe a method to develop a pocketbook for women and family for preparing pregnancy and childbirth using 4D development model.</div></span></li><li><span>•</span><span><div>The expected developed pocketbook is able to assist women and families’ readiness during pregnancy and childbirth and subsequently optimize the success of pregnancy care.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103467"},"PeriodicalIF":1.6,"publicationDate":"2025-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144535816","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Achieving cloud resource optimization with trust-based access control: A novel ML strategy for enhanced performance","authors":"Bala Subramanian C,&nbsp;Bharathi ST,&nbsp;Shanmugapriya S","doi":"10.1016/j.mex.2025.103461","DOIUrl":"10.1016/j.mex.2025.103461","url":null,"abstract":"<div><div>Cloud computing continues to rise, increasing the demand for more intelligent, rapid, and secure resource management. This paper presents AdaPCA—a novel method that integrates the adaptive capabilities of AdaBoost with the dimensionality-reduction efficacy of PCA. What is the objective? Enhance trust-based access control and resource allocation decisions while maintaining a minimal computational burden. High-dimensional trust data frequently hampers systems; however, AdaPCA mitigates this issue by identifying essential aspects and enhancing learning efficacy concurrently. To evaluate its performance, we conducted a series of simulations comparing it with established methods such as Decision Trees, Random Forests, and Gradient Boosting. We assessed execution time, resource use, latency, and trust accuracy. Results show that AdaPCA achieved a trust score prediction accuracy of 99.8 %, a resource utilization efficiency of 95 %, and reduced allocation time to 140 ms, outperforming the benchmark models across all evaluated parameters. AdaPCA had superior performance overall—expedited decision-making, optimized resource utilization, reduced latency, and the highest accuracy in trust evaluation among the evaluated models. AdaPCA is not merely another model; it represents a significant advancement towards more intelligent and safe cloud systems designed for the future.<ul><li><span>•</span><span><div>Introduces AdaPCA, a novel hybrid approach that integrates AdaBoost with PCA to optimize cloud resource allocation and improve trust-based access control.</div></span></li><li><span>•</span><span><div>Outperforms conventional techniques such as Decision Tree, Random Forest, and Gradient Boosting by attaining superior trust accuracy, expedited execution, enhanced resource utilization, and reduced latency.</div></span></li><li><span>•</span><span><div>Presents an intelligent, scalable, and adaptable architecture for secure and efficient management of cloud resources, substantiated by extensive simulation experiments.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103461"},"PeriodicalIF":1.6,"publicationDate":"2025-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489602","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Bayesian spatio-temporal conditional autoregressive localized modeling techniques for socioeconomic factors and stunting in Indonesia","authors":"Aswi Aswi ,&nbsp;Septian Rahardiantoro ,&nbsp;Anang Kurnia ,&nbsp;Bagus Sartono ,&nbsp;Dian Handayani ,&nbsp;Nurwan Nurwan","doi":"10.1016/j.mex.2025.103464","DOIUrl":"10.1016/j.mex.2025.103464","url":null,"abstract":"<div><div>Stunting remains a persistent public health issue in Indonesia, exhibiting significant spatial and temporal variation. To address this, we employed a hierarchical Bayesian spatio-temporal localized Conditional Autoregressive (CAR) model that includes a clustering component to identify risk factors and estimate relative risk (RR) across 34 provinces from 2020 to 2022. A total of 480 models were evaluated, encompassing three variants of the Bayesian spatio-temporal localized CAR model, 32 covariate combinations, and five hyperprior settings. Assuming a Poisson likelihood for stunting counts, the optimal model was estimated using Markov Chain Monte Carlo methods and included two covariates, namely the poverty rate and the incidence of low birth weight, with up to five spatial clusters. Higher poverty levels and increased prevalence of low birth weight were significantly associated with elevated stunting risk among children under five. Spatio-temporal clustering patterns and the estimated relative risks of stunting varied across Indonesian provinces from 2020 to 2022. Nusa Tenggara Timur consistently ranked among the top three provinces with the highest risk (RR = 2.421 in 2020; 2.384 in 2021; 2.676 in 2022). The highest risk was observed in Sulawesi Barat in 2022 (RR = 2.768), while DKI Jakarta consistently showed the lowest (RR = 0.004).</div><div>Some key points of the article are:<ul><li><span>•</span><span><div>Bayesian spatio-temporal models facilitate the classification of distinct area groups</div></span></li><li><span>•</span><span><div>The models were employed to analyze stunting patterns in Indonesia.</div></span></li><li><span>•</span><span><div>The inclusion of covariates influenced the number of groups identified.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103464"},"PeriodicalIF":1.6,"publicationDate":"2025-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144490653","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"AI-driven pharmacovigilance: Enhancing adverse drug reaction detection with deep learning and NLP","authors":"Dr. Bharti Khemani ,&nbsp;Dr. Sachin Malave ,&nbsp;Samyukta Shinde ,&nbsp;Mandvi Shukla ,&nbsp;Razzaq Shikalgar ,&nbsp;Harshita Talwar","doi":"10.1016/j.mex.2025.103460","DOIUrl":"10.1016/j.mex.2025.103460","url":null,"abstract":"<div><div>In the healthcare industry, the ever-increasing volume of clinical trial data presents challenges for ensuring drug safety and detecting adverse drug reactions (ADRs). This study aims to address the challenge of accurately detecting Serious Adverse Events (SAEs) in pharmacovigilance, a critical component in ensuring drug safety during and after clinical trials. The key problem lies in the underreporting and delayed detection of Adverse Drug Reactions (ADRs) due to the heterogeneous nature of medical data, class imbalance, and the limited scope of traditional monitoring techniques. This study proposes a hybrid AI-driven framework that integrates structured (e.g., patient demographics, lab results) and unstructured data (e.g., clinical notes) to detect ADRs using advanced deep learning and NLP methods. The objective is to outperform traditional signal detection methods and provide interpretable predictions to aid clinicians in real-time. By leveraging advanced Machine Learning (ML) and Deep Learning (DL) techniques, including Random Forests, Gradient Boosting Machines, and Convolutional Neural Networks (CNNs), our model aims to identify potential ADRs across different patient subgroups. Through meticulous feature engineering and the application of techniques to address data imbalance, our model demonstrates improved accuracy and interpretability in predicting ADRs. The CNN model achieved an accuracy of 85 %, outperforming traditional models, such as Logistic Regression (78 %) and Support Vector Machines (80 %). These findings suggest that specific demographic and clinical factors significantly influence the likelihood of adverse reactions, offering valuable insights for targeted monitoring and risk mitigation strategies[11]. This research underscores the potential of predictive modeling to enhance pharmacovigilance efforts and ensure safer clinical trial outcomes.<ul><li><span>•</span><span><div>The research methodology includes a comparison of supervised learning algorithms, such as Logistic Regression, Random Forest, Gradient Boost, CNN, and genetic algorithms, to identify patterns and anomalies in clinical trial data. BERT and GPT, were also employed to provide the functionality of textual interactions over medical data.</div></span></li><li><span>•</span><span><div>Performance metrics such as accuracy, precision, recall, and F1-score were systematically applied to evaluate each model’s performance. Among the models tested, the CNN model with BERT achieved the highest accuracy, providing valuable insights into the potential of deep learning for enhancing pharmacovigilance practices.</div></span></li><li><span>•</span><span><div>These findings suggest that an inclusion of diverse clinical data when supplied to advanced ML and NLP techniques can significantly improve the detection of ADRs, leading to better alignment with the fundamental principles of Good Clinical Practice (GCP).</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103460"},"PeriodicalIF":1.6,"publicationDate":"2025-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144518792","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}