Review
Clinical Neurology
Jonathan Huang, Nathan A. Shlobin, Michael DeCuypere, Sandi K. Lam
Summary: This study systematically reviewed DL studies in neurosurgical outcome prediction and found that lack of model interpretability and data quality pose significant barriers to the validity and reproducibility of DL models. The study highlights the need for greater transparency and reproducibility in model development and reporting to facilitate validation and clinical use.
Article
Public, Environmental & Occupational Health
Akshaya Karthikeyan, Akshit Garg, P. K. Vinod, U. Deva Priyakumar
Summary: This study introduces machine learning methods based on blood test data for predicting COVID-19 mortality risk, achieving high accuracy. By analyzing key biomarkers, it provides decision-making solutions for healthcare systems to expedite the decision-making process.
FRONTIERS IN PUBLIC HEALTH
(2021)
Article
Computer Science, Artificial Intelligence
Guilherme Palumbo, Davide Carneiro, Miguel Guimares, Victor Alves, Paulo Novais
Summary: In recent years, there has been a significant increase in the number of machine learning algorithms and their parameters. This presents both opportunities and challenges in training models. Traditional search-based methods become computationally expensive and time-consuming as datasets grow, especially in data streaming scenarios. This paper proposes a meta-learning approach that can predict performance indicators and recommend the best algorithm/configuration for training models. The proposed approach is up to 130 times faster than a state-of-the-art method and only slightly worse in terms of model quality, making it suitable for scenarios that require regular model updates with shorter training time.
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
(2023)
Review
Biotechnology & Applied Microbiology
Ayleen Bertini, Rodrigo Salas, Steren Chabert, Luis Sobrevia, Fabian Pardo
Summary: Through analyzing 98 articles, 31 were selected for predicting perinatal complications with main features of electronic medical records, medical images, and biological markers. The studies mainly focus on pre-eclampsia and prematurity, using AUC as the main precision metric for accuracy measurement.
FRONTIERS IN BIOENGINEERING AND BIOTECHNOLOGY
(2022)
Review
Computer Science, Information Systems
Khadijeh Moulaei, Hamid Sharifi, Kambiz Bahaadinbeigy, Ali Akbar Haghdoost, Naser Nasiri
Summary: This study conducted a systematic review and meta-analysis to investigate the performance of machine learning algorithms in predicting viral hepatitis. The results showed that SVM and KNN algorithms demonstrated superior performance in predicting hepatitis.
INTERNATIONAL JOURNAL OF MEDICAL INFORMATICS
(2023)
Article
Oncology
Siteng Chen, Ning Zhang, Liren Jiang, Feng Gao, Jialiang Shao, Tao Wang, Encheng Zhang, Hong Yu, Xiang Wang, Junhua Zheng
Summary: This study developed a novel computational recognition technology using machine learning methods for accurate diagnosis and prognosis prediction of ccRCC patients. The results showed high diagnostic accuracy in the training, test, and external validation cohorts, demonstrating the potential clinical use of this machine learning histopathological image signature for ccRCC.
INTERNATIONAL JOURNAL OF CANCER
(2021)
Review
Behavioral Sciences
Federica Colombo, Federico Calesella, Mario Gennaro Mazza, Elisa Maria Teresa Melloni, Marco J. Morelli, Giulia Maria Scotti, Francesco Benedetti, Irene Bollettini, Benedetta Vai
Summary: Applying machine learning to objective markers can improve the accuracy of bipolar disorder diagnosis and overcome prognosis uncertainty caused by subjectivity. However, future studies should adopt best practices in methodology for further advancements in this field.
NEUROSCIENCE AND BIOBEHAVIORAL REVIEWS
(2022)
Review
Health Care Sciences & Services
Paula Dhiman, Jie Ma, Constanza L. Andaur Navarro, Benjamin Speich, Garrett Bullock, Johanna A. A. Damen, Lotty Hooft, Shona Kirtley, Richard D. Riley, Ben Van Calster, Karel G. M. Moons, Gary S. Collins
Summary: This article conducted a systematic review on oncology-related studies that developed and validated prognostic models using machine learning. The findings revealed the presence of spin, i.e., overinterpretation of findings, in these studies. The inconsistent reporting and use of overly strong or leading words in the publications indicate the need for caution when reading and using prognostic models in oncology.
JOURNAL OF CLINICAL EPIDEMIOLOGY
(2023)
Review
Geriatrics & Gerontology
Qi Xie, Xinglei Wang, Juhong Pei, Yinping Wu, Qiang Guo, Yujie Su, Hui Yan, Ruiling Nan, Haixia Chen, Xinman Dou
Summary: This article critically appraises and quantifies studies using machine learning to predict delirium through a systematic review and meta-analysis. The findings show that machine learning models perform well in predicting delirium. However, there are potential shortcomings in the current approaches, including low comparability and reproducibility.
JOURNAL OF THE AMERICAN MEDICAL DIRECTORS ASSOCIATION
(2022)
Review
Neurosciences
Yujia Yang, Li Tang, Yiting Deng, Xuzi Li, Anling Luo, Zhao Zhang, Li He, Cairong Zhu, Muke Zhou
Summary: This study aimed to evaluate the accuracy of artificial intelligence models in predicting the prognosis of stroke. The results showed that AI models performed well in predicting the outcomes of ischemic stroke, which could assist physicians in assessing stroke patients' outcomes.
FRONTIERS IN NEUROSCIENCE
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Jianhong Cheng, John Sollee, Celina Hsieh, Hailin Yue, Nicholas Vandal, Justin Shanahan, Ji Whae Choi, Thi My Linh Tran, Kasey Halsey, Franklin Iheanacho, James Warren, Abdullah Ahmed, Carsten Eickhoff, Michael Feldman, Eduardo Mortani Barbosa, Ihab Kamel, Cheng Ting Lin, Thomas Yi, Terrance Healey, Paul Zhang, Jing Wu, Michael Atalay, Harrison X. Bai, Zhicheng Jiao, Jianxin Wang
Summary: Deep learning models using longitudinal CXRs and clinical data were developed to predict in-hospital mortality for COVID-19 patients in the ICU. Models based on pre-ICU CXRs achieved an AUC of 0.632 and an accuracy of 0.593, models based on ICU CXRs achieved an AUC of 0.697 and an accuracy of 0.657, models based on all longitudinal CXRs achieved an AUC of 0.702 and an accuracy of 0.694, and models based on clinical data alone achieved an AUC of 0.653 and an accuracy of 0.657. The addition of longitudinal imaging to clinical data significantly improved mortality prediction, reaching an AUC of 0.727 and an accuracy of 0.732.
EUROPEAN RADIOLOGY
(2022)
Review
Dermatology
Juhong Pei, Xiaojing Guo, Hongxia Tao, Yuting Wei, Hongyan Zhang, Yuxia Ma, Lin Han
Summary: This study aims to systematically evaluate the performance of machine learning models in predicting pressure injury. A total of 18 relevant studies were included for narrative review, and 14 of them were eligible for meta-analysis. The results showed that these models demonstrated excellent performance in predicting pressure injury, with high sensitivity and specificity. However, further high-quality studies are needed to validate these findings.
INTERNATIONAL WOUND JOURNAL
(2023)
Article
Biophysics
Xinxi Lu, Jikai Wang, Junxia Cai, Zhihuan Xing, Jian Huang
Summary: This study aims to establish a prediction model for gestational diabetes and hypertension using machine learning methods and real pregnancy examination data. The results show that the model can accurately predict the occurrence of diabetes and hypertension based on the examination data. This study is beneficial for early prevention and treatment of these pregnancy complications.
JOURNAL OF MECHANICS IN MEDICINE AND BIOLOGY
(2022)
Review
Computer Science, Information Systems
Ruiyao Chen, Jiayuan Chen, Sen Yang, Shuqing Luo, Zhongzhou Xiao, Lu Lu, Bilin Liang, Sichen Liu, Huwei Shi, Jie Xu
Summary: This study systematically examined the prognostic value of machine learning (ML) in patients with COVID-19. The results demonstrated the satisfactory performance of ML in predicting prognostic outcomes, suggesting its potential value in supporting clinical decision-making.
INTERNATIONAL JOURNAL OF MEDICAL INFORMATICS
(2023)
Article
Clinical Neurology
XiaoSheng Li, Zongning Chen, Hexian Jiao, BinYang Wang, Hui Yin, LuJia Chen, Hongling Shi, Yong Yin, Dongdong Qin
Summary: This study examined the efficiency of machine learning in predicting post-stroke cognitive impairment and found that it has the potential to be used as a tool for prediction and the development of clinical scoring scales.
FRONTIERS IN NEUROLOGY
(2023)
Article
Oncology
Shu-Tai Shen Hsiao, Chao-Yang Kuo, Tsan-Hon Liou, Tai-Ton Wang, Yen-Lin Liu, Sung-Hui Tseng
Summary: This study developed an ICF core set for interviewing pediatric brain tumor survivors in Taiwan, aiming to assist healthcare professionals in implementing disability assessment and management measures.
Article
Multidisciplinary Sciences
Sri Susanty, Herdiantri A. Sufriyana, Emily Chia-Yu A. Su, Yeu-Hui A. Chuang
Summary: This study aimed to develop and validate a questionnaire-free, machine-learning model as an alternative triage test for the GDS-15 among community-dwelling older adults. The best models were the random forest (RF) and deep-insight visible neural network by internal validation, but both performances were undifferentiated by external validation. The AUROC of the RF model was 0.619 (95% CI 0.610 to 0.627) for the external validation set with a non-local ethnic group.
Article
Microbiology
Yung-Hsien Shao, Yu-Wei Wu, Muhammad Naufal, Jer-Horng Wu
Summary: In this study, a metagenomic approach was used to analyze the genomic contents of core members in a partial nitritation-anaerobic ammonium oxidation (PN-A) reactor treating low-strength ammonium wastewater. The analysis showed the metabolic traits and predicted microbial interactions of 18 core species, providing new insights into microbial adaptation to the intermittent microaeration specific to the PN-A reactor and its application to low-strength ammonium wastewater.
FRONTIERS IN MICROBIOLOGY
(2023)
Article
Biochemistry & Molecular Biology
Ming-Ren Yang, Yu-Wei Wu
Summary: Understanding the mechanisms of genes is crucial in studying antimicrobial-resistant bacteria. Genes related to antimicrobial resistance can be used as biomarkers for predicting bacterial resistance to antibiotics. We developed a Cross-Validated Feature Selection approach for selecting the most predictive gene sets from bacterial genomes. By testing this approach on bacterial pan-genome datasets, we found that it successfully identified both known and novel antimicrobial resistance genes, demonstrating its potential as a tool for expanding resistance gene databases.
COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL
(2023)
Article
Biochemistry & Molecular Biology
Bo-Siyuan Jian, Shao-Lun Chiou, Chun-Chia Hsu, Josh Ho, Yu-Wei Wu, John Chu
Summary: The traditional natural product discovery approach has only explored a small portion of nature's chemical diversity. By using bioinformatic tools, we can interpret the instructions encoded in microbial biosynthetic genes and overcome methodological limitations, thus expanding the scope of discovery. The evaluation of prediction algorithms for nonribosomal peptides (NRPs) synthesis shows that there is still a vast unexplored biosynthetic diversity in nature, suggesting the possibility of discovering unknown NRPs with functional roles other than siderophores.
ACS CHEMICAL BIOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Herdiantri Sufriyana, Yu-Wei Wu, Emily Chia-Yu Su
Summary: This study demonstrates a human-guided deep learning approach that develops, validates, and deploys prognostic prediction models for PROM and an estimator of time of delivery from non-image data using convolutional neural networks. The developed model outperforms previous models and can be explained using knowledge-based diagrams and model representation.
Article
Genetics & Heredity
Ming-Ren Yang, Shun-Feng Su, Yu-Wei Wu
Summary: By using a machine learning feature selection approach, we accurately predicted the minimum inhibitory concentration (MIC) values of antimicrobial drugs, and found that half of the selected features were annotated as hypothetical proteins with unknown functions, indicating the potential of uncovering novel genes associated with pathogenic antimicrobial resistance through feature selection on the entire gene set.
FRONTIERS IN GENETICS
(2023)
Article
Medicine, General & Internal
Chen-Chih Chung, Emily Chia-Yu Su, Jia-Hung Chen, Yi-Tui Chen, Chao-Yang Kuo
Summary: This study developed an extreme gradient boosting (XGBoost)-based model using age, fasting glucose, and National Institutes of Health Stroke Scale (NIHSS) scores as predictors to accurately predict three-month functional outcomes after acute ischemic stroke (AIS). The model showed reliable predictive power and identified factors associated with unfavorable prognoses, such as NIHSS score > 5, age over 64 years, and fasting blood glucose > 86 mg/dL. Furthermore, it demonstrated the importance of different predictors for patients receiving different AIS treatments.
Article
Health Care Sciences & Services
Yi-Tui Chen, Emily Chia-Yu Su, Fang Ming Hung, Tomoru Hiramatsu, Tzu-Jen Hung, Chao-Yang Kuo
Summary: This paper aimed to analyze and compare case fatality rates of COVID-19, examine the existence of learning curves for medical treatments, and explore the impact of vaccination on fatality rate reduction. The results showed that low registration and viral test rates led to low fatality rates, and there was a significant learning curve for all countries except China. Treatment for COVID-19 can be improved through repeated experience. Vaccinations in the U.K. and U.S.A. effectively reduced fatality rates, possibly due to higher vaccination rates. Learning curves for medical treatment were identified, explaining the effect of vaccination rates on fatalities.
Article
Computer Science, Interdisciplinary Applications
Chun-You Chen, Ya-Lin Chen, Jeremiah Scholl, Hsuan-Chia Yang, Yu-Chuan (Jack) Li
Summary: This study evaluated the overall performance of a machine learning-based CDSS (MedGuard) in triggering clinically relevant alerts and intercepting inappropriate drug errors and LASA drug errors. The results showed that MedGuard has the ability to improve patients' safety by triggering clinically valid alerts.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2024)
Article
Zoology
Tzi-Yuan Wang, King-Siang Goh, Liang-Jong Wang, Li-Ling Wu, Feng-Yu Wang, Yu-Wei Wu
Summary: Abscondita cerata is an endemic firefly species in Taiwan, which has a high abundance and wide distribution. The draft genome of Abs. cerata, sequenced using Nanopore technology, was found to be 967 Mb in size, assembled into 4,855 contigs with an N50 length of 325.269 kb. It was predicted to contain 55,206 protein-coding genes, with functional annotations for 37.78% of them. Repeat elements accounted for 47.11% of the genome, with DNA transposons being the most common type. The completeness of the genome and genes was evaluated to be 84.8% and 79%, respectively. The comparative transcriptome analysis revealed insights into the vision, humidity sensing, and luminescence of Abs. cerata.
ZOOLOGICAL STUDIES
(2023)
Article
Multidisciplinary Sciences
Zheng-Jie Huang, Brijesh Patel, Wei-Hao Lu, Tz-Yu Yang, Wei-Cheng Tung, Vytautas Bucinskas, Modris Greitans, Yu-Wei Wu, Po Ting Lin
Summary: This study presents a novel approach using deep learning techniques for automatic cell detection. By optimizing image contrast and introducing a universal contrast enhancement variable, the proposed method achieves high accuracy in yeast cell detection. Comparative experiments demonstrate the superior performance of this method in cell detection, with significant improvements compared to conventional methods.
SCIENTIFIC REPORTS
(2023)