Article
Biology
Lauren J. Beesley, Jeremy M. G. Taylor
Summary: MICE is a popular approach for handling missing data, and we propose a novel strategy to directly incorporate the analysis model by stacking multiple imputations.
Article
Computer Science, Artificial Intelligence
Li Zhang, Xiaohan Zheng, Qingqing Pang, Weida Zhou
Summary: This paper investigates the issue of computational complexity in GKSVM-RFE and proposes two fast versions for feature ranking. By introducing approximate Gaussian kernels, two ranking scores based on different approximate schemes are designed to calculate and rank features quickly in iterations.
APPLIED INTELLIGENCE
(2021)
Article
Computer Science, Interdisciplinary Applications
Xiaojian Ding, Fan Yang, Fuming Ma
Summary: The paper addresses the issue of model selection in support vector machine-based recursive feature elimination (SVM-RFE), proposing an approximation method to evaluate the generalization error and a new criterion to tune the penalty parameter C. The expensive computational cost of the algorithm is mitigated by several alpha seeding approaches, showing superior performance on bioinformatics datasets and empirical time savings.
JOURNAL OF BIOMEDICAL INFORMATICS
(2022)
Article
Public, Environmental & Occupational Health
Hannah S. Laqueur, Aaron B. Shev, Rose M. C. Kagawa
Summary: This paper proposes a data-adaptive approach to model selection for addressing missing data, using Super Learner and local kernel estimation in MICE to predict the missing values, which results in final parameter estimates with lower bias and better coverage.
AMERICAN JOURNAL OF EPIDEMIOLOGY
(2022)
Article
Automation & Control Systems
Abtin Ijadi Maghsoodi, Ali Ebadi Torkayesh, Lincoln C. Wood, Enrique Herrera-Viedma, Kannan Govindan
Summary: The development of state-of-the-art solutions in decision sciences has been driven by the move towards an era of big data structured problems. Large Scale Decision-Making (LSDM) and Data-Driven Decision-Making (DDDM) approaches have emerged to tackle the challenges posed by incomplete data in LSDM problems. This study proposes a machine learning-driven DDDM method for solving LSDM problems with incomplete data and a large number of decision attributes. The method includes data imputation using the Expectation-Maximization algorithm and the extraction of core criteria using Recursive Feature Elimination with Least Square Support Vector. The proposed method is validated by evaluating the sustainability performance of countries worldwide under Sustainable Development Goals.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Environmental Sciences
Christopher A. A. Ramezan
Summary: Feature selection is important in remote sensing analysis to improve classification accuracy and reduce computational complexity. However, the generalizability and transferability of feature selection results depend on different classification models and datasets. While feature selection results can provide insights for analysis, they may not always provide comparable accuracies when applied to other classification models or similar remotely sensed datasets. Therefore, feature selection should be individually conducted for each training set to determine the optimal feature set for the classification model.
Article
Engineering, Electrical & Electronic
Zhou Shuai, Li Tao, Li Yongzhao
Summary: This paper proposes a modulation recognition algorithm based on feature selection. By using the hyperplane of the support vector machine and the weight vector of features, cumulative features are selected and the modulation type employed at the transmitter is identified. Simulation results show that the proposed algorithm can optimize feature selection for modulation recognition and improve identification efficiency when compared with existing feature selection algorithms.
CHINESE JOURNAL OF ELECTRONICS
(2023)
Article
Agricultural Engineering
Yuzhen Xiao, Guozhao Mo, Xiya Xiong, Jiawen Pan, Bingbing Hu, Caicong Wu, Weixin Zhai
Summary: This study proposes a DR-XGBoost model based on dual feature extraction and recursive feature elimination for field-road segmentation in the processing of agricultural machinery trajectory. The model achieves improved segmentation accuracy through feature expansion and elimination and demonstrates significant improvement in experiments.
INTERNATIONAL JOURNAL OF AGRICULTURAL AND BIOLOGICAL ENGINEERING
(2023)
Article
Geochemistry & Geophysics
Fernanda Gontijo Fernandes Niquini, Andre Miranda Brito Branches, Joao Felipe Coimbra Leite Costa, Gabriel de Castro Moreira, Claudio Luiz Schneider, Florence Cristiane de Araujo, Luciano Nunes Capponi
Summary: This study proposed a solution using feature engineering to select the best set of explanatory variables for predicting the production and recovery of output variables in a phosphate flotation plant. The results showed that using a reduced set of input variables resulted in an increase in mean squared error and root mean squared error of less than 2.6%. The trade-off between simplicity and quality of the model needs to be considered when choosing the final neural network model.
Article
Neurosciences
Jiaxi Su, Xiaoyan Zhang, Ziyuan Zhang, Hongmei Wang, Jia Wu, Guangming Shi, Chenwang Jin, Minghao Dong
Summary: This study investigates how short-term real-world visual experience modulates baseline neuronal activity in the resting state. The results suggest that real-world visual experience alters the resting-state brain representation in multidimensional neurobehavioral components, which are closely interrelated with high-order cognitive and low-order visual factors.
FRONTIERS IN NEUROSCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Yinan Guo, Zirui Zhang, Fengzhen Tang
Summary: Feature selection is important in machine learning to reduce complexity and simplify interpretation. A novel non-linear method proposed in this paper uses kernelized multi-class support vector machines and fast recursive feature elimination to select features that work well for all classes, resulting in lower computational time complexity.
PATTERN RECOGNITION
(2021)
Article
Computer Science, Cybernetics
G. Mariammal, A. Suruliandi, S. P. Raja, E. Poongothai
Summary: Crop cultivation prediction relies on various factors, and different regions have different suitable crops. Machine learning techniques can assist farmers in selecting the most suitable crops for specific regions, improving crop prediction accuracy.
IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS
(2021)
Article
Physics, Multidisciplinary
Tao Wang, Mengyu Jiao, Xiaoxia Wang
Summary: This paper proposes a novel stacking ensemble framework for link prediction and conducts extensive experiments on six networks to demonstrate that the proposed method can achieve better prediction results and applicability robustness.
Article
Oncology
Golestan Karami, Marco Giuseppe Orlando, Andrea Delli Pizzi, Massimo Caulo, Cosimo Del Gratta
Summary: The study aimed to predict survival time of GBM patients using machine learning methods integrating multi-modality MRI and feature selection. The RF-RFE Gboost machine could predict survival time with 75% accuracy, and rCBV in the low perfusion area had a significant impact on survival time. The integration of multi-modality MRI and feature selection method can enhance classifier performance.
Article
Automation & Control Systems
Qingjian Ni, Xuehan Cao
Summary: The problem of missing values in time series data is common in data mining and analysis. This paper proposes a new model based on Generative Adversarial Networks to impute missing values from raw data. The model is tested on 4 real-world datasets and outperforms the other 10 state-of-the-art methods in most cases.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2022)
Article
Green & Sustainable Science & Technology
Karpagam Sundararajan, Kathiravan Srinivasan
Summary: Drought has a direct impact on environmental sustainability. Predicting drought early on can help in implementing drought mitigation plans. In India, the standardized precipitation index (SPI) and standardized precipitation evapotranspiration index (SPEI) are used to predict meteorological drought. This study evaluates the ability of these indices to predict meteorological drought in Tamil Nadu using 62 years of data.
Review
Computer Science, Information Systems
Shriniket Dixit, Khitij Bohre, Yashbir Singh, Yassine Himeur, Wathiq Mansoor, Shadi Atalla, Kathiravan Srinivasan
Summary: Parkinson's disease is a devastating neurological disease that requires the development of a faster, less expensive diagnostic instrument. This article provides a thorough analysis of AI-based ML and DL techniques used to diagnose PD and their influence on developing additional research directions.
Review
Medicine, General & Internal
Saransh Bhachawat, Eashwar Shriram, Kathiravan Srinivasan, Yuh-Chung Hu
Summary: Degenerative nerve diseases like Alzheimer's and Parkinson's have become a global concern, affecting approximately 1/6th of the world's population. Early detection through machine learning algorithms, which can infer based on patient data and history, is crucial for effective treatment. The use of machine learning and deep learning in the diagnosis of these diseases has shown promising results.
Article
Public, Environmental & Occupational Health
Venkatesan Rajinikanth, P. M. Durai Raj Vincent, Kathiravan Srinivasan, G. Ananth Prabhu, Chuan-Yu Chang
Summary: Cancer rates in the kidney are on the rise, and accurate detection and management are crucial. This study focuses on developing a framework to classify renal CT images using deep-learning schemes, with a pre-processing scheme to improve accuracy. The experimental results show that the KNN classifier achieves 100% detection accuracy with the pre-processed CT slices, making it clinically significant.
FRONTIERS IN PUBLIC HEALTH
(2023)
Article
Public, Environmental & Occupational Health
Sanchit Vijay, Thejineaswar Guhan, Kathiravan Srinivasan, P. M. Durai Raj Vincent, Chuan-Yu Chang
Summary: Brain tumor diagnosis has been time-consuming, but automating the segmentation process can speed it up. This paper introduces SPP-U-Net, a model that replaces residual connections with a combination of Spatial Pyramid Pooling (SPP) and Attention blocks, allowing for greater context and scope in the segmentation. The proposed approach achieves comparable results to existing literature without increasing training parameters.
FRONTIERS IN PUBLIC HEALTH
(2023)
Article
Computer Science, Interdisciplinary Applications
David Susic, Shabbir Syed-Abdul, Erik Dovgan, Jitendra Jonnagaddala, Anton Gradisek
Summary: This study evaluated the performance of machine learning algorithms in predicting the survival of colorectal cancer patients 1 to 5 years after diagnosis and identified the most important variables. The results showed that machine learning algorithms can predict the survival probability of colorectal cancer patients and can be used to assist decision-making in clinical care management.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2023)
Review
Green & Sustainable Science & Technology
Amogh Gyaneshwar, Anirudh Mishra, Utkarsh Chadha, P. M. Durai Raj Vincent, Venkatesan Rajinikanth, Ganapathy Pattukandan Ganapathy, Kathiravan Srinivasan
Summary: Deep learning models have proven to be effective in drought forecasting, providing more accurate and timely predictions to mitigate the impacts of drought on crop failure, water shortages, and economic losses.
Article
Medicine, General & Internal
A. Angel Nancy, Dakshanamoorthy Ravindran, Durai Raj Vincent, Kathiravan Srinivasan, Chuan-Yu Chang
Summary: The fast-paced technology trend has led to continuous transformation, with cloud computing being the prime provider of various services on a pay-per-use basis. Cloud computing supports the internet of things (IoT) by providing computation and storage capabilities. The inclusion of decentralized fog computing addresses latency and connectivity issues in the cloud-IoT interaction. In the healthcare domain, a fog-assisted smart healthcare system combining fuzzy inference system (FIS) and recurrent neural network (RNN) variants has shown significant improvements in performance, achieving a classification accuracy of 99.125%.
Article
Medicine, General & Internal
Venkatesan Rajinikanth, P. M. Durai Raj Vincent, C. N. Gnanaprakasam, Kathiravan Srinivasan, Chuan-Yu Chang
Summary: This research aims to develop an efficient deep-learning-based brain-tumor detection scheme using FLAIR- and T2-modality MRI slices. The scheme includes preprocessing, deep-feature extraction, tumor segmentation, feature optimization, and binary classification. Experimental results show that the integrated feature-based scheme achieves a classification accuracy of 99.6667% when using a support-vector-machine classifier.
Article
Medicine, General & Internal
Jayakumar Kaliappan, Apoorva Reddy Bagepalli, Shubh Almal, Rishabh Mishra, Yuh-Chung Hu, Kathiravan Srinivasan
Summary: Intrauterine fetal demise is a significant issue in developing and underdeveloped countries, and machine learning models can help detect it. This study used 22 features from fetal heart rate obtained from CTG for 2126 patients and applied various cross-validation techniques to enhance the performance of ML algorithms. Gradient Boosting and Voting Classifier achieved 99% accuracy after cross-validation.
Review
Medicine, General & Internal
Somit Jain, Dharmik Naicker, Ritu Raj, Vedanshu Patel, Yuh-Chung Hu, Kathiravan Srinivasan, Chun-Ping Jen
Summary: Cancer is a dangerous disease that can have negative consequences for the body, is a leading cause of death, and is difficult to detect. Doctors use different methods, including imaging tests, to diagnose cancer. This article evaluates computational-intelligence approaches in cancer diagnosis using machine learning and deep learning models, and explores their advantages and disadvantages. Despite some clinical issues, artificial intelligence has significant potential to enhance cancer imaging and diagnosis.
Article
Oncology
Bernardo Pereira Cabral, Luiza Amara Maciel Braga, Shabbir Syed-Abdul, Fabio Batista Mota
Summary: Cancer is a major cause of global mortality, resulting in 9.3 million deaths annually. The use of artificial intelligence (AI) applications in various domains of oncology has been proposed to alleviate this burden. However, the potential applications and barriers to widespread adoption of AI remain unclear. A global web-based survey of over 1000 AI and cancer researchers was conducted to address this gap. The results indicated that most respondents believed AI would have positive impacts on cancer grading and classification, follow-up services, and diagnostic accuracy. However, limitations such as difficulties incorporating AI into clinical practice and the lack of standardization in cancer health data were also identified. These limitations pose significant challenges in testing, validation, certification, and auditing of AI algorithms and systems. The findings of this study provide valuable insights for informed decision-making in AI and cancer research and development.
Article
Biochemistry & Molecular Biology
Nivedhitha Mahendran, P. M. Durai Raj Vincent
Summary: Alzheimer's disease is a form of Dementia with uncertain mechanism and no vital genetic factor. Recent advancements in bioinformatics have enabled the discovery of genetic risk factors associated with Alzheimer's disease. A Deep Belief Network-based prediction model using DNA Methylation and Gene Expression Microarray Data has been developed, overcoming the challenge of high dimension low sample size. The proposed feature selection technique and prediction model outperform existing methods, indicating promising results for multi-omics data.
COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL
(2023)
Review
Health Care Sciences & Services
Eshita Dhar, Umashankar Upadhyay, Yaoru Huang, Mohy Uddin, George Manias, Dimosthenis Kyriazis, Usman Wajid, Hamza AlShawaf, Shabbir Syed Abdul
Summary: Due to the challenges posed by the COVID-19 pandemic, technology and digital solutions have played a crucial role in providing necessary healthcare services, particularly in medical education and clinical care. This scoping review examined recent developments in the use of Virtual Reality (VR) for therapeutic care and medical education, with a focus on training medical students and patients. The findings showed significant improvements in medical education and clinical care through the use of VR, with participants endorsing its safety, engagement, and benefits. Collaboration between researchers, the VR industry, and healthcare professionals is needed to further enhance patient care and refine VR content and simulation development.
Review
Health Care Sciences & Services
Ritik Kumar, Divyangi Singh, Kathiravan Srinivasan, Yuh-Chung Hu
Summary: Blockchain technology has experienced substantial growth in the past decade, finding applications in various fields for its security and privacy features. In the healthcare industry, blockchain has been used for secure data logging, transactions, and maintenance with smart contracts. This review explores the integration of artificial intelligence (AI) with blockchain and discusses its applications in healthcare, including EHR management, telemedicine, genomics, drug research, specialized imaging, and outbreak prediction.