Ferhat D Zengul1,2, Timmy Lee3,4, Dursun Delen5,6, Ammar Almehmi3, Nataliya V Ivankova7,8, Tapan Mehta7, Kazim Topuz9. 1. Department of Health Services Administration, The University of Alabama at Birmingham, Birmingham, Alabama, USA, ferhat@uab.edu. 2. School of Engineering, Center for Integrated Systems, The University of Alabama at Birmingham, Birmingham, Alabama, USA, ferhat@uab.edu. 3. Division of Nephrology and Hypertension, The University of Alabama at Birmingham, Birmingham, Alabama, USA. 4. Veterans Affairs Medical Center, Birmingham, Alabama, USA. 5. Department of Management Science and Information Systems, Spears School of Business, Oklahoma State University, Stillwater, Oklahoma, USA. 6. Center for Health Systems Innovation, Department of Management Science and Information Systems, Spears School of Business, Oklahoma State University, Stillwater, Oklahoma, USA. 7. Department of Health Services Administration, The University of Alabama at Birmingham, Birmingham, Alabama, USA. 8. Acute, Chronic, and Continuing Care, School of Nursing, The University of Alabama at Birmingham, Birmingham, Alabama, USA. 9. School of Finance, Operations Management and International Business, Collins College of Business, The University of Tulsa, Tulsa, Oklahoma, USA.
Abstract
BACKGROUND: Nephrology research is expanding, and harnessing the much-needed information and data for the practice of evidence-based medicine is becoming more challenging. In this study, we used the natural language processing and text mining approach to mitigate some of these challenges. METHODS: We analyzed 17,412 abstracts from the top-10 nephrology journals over 10 years (2007-2017) by using latent semantic analysis and topic analysis. RESULTS: The analyses revealed 10 distinct topics (T) for nephrology research ranging from basic science studies, using animal modeling (T-1), to dialysis vascular access-related issues -(T-10). The trend analyses indicated that while the majority of topics stayed relatively stable, some of the research topics experienced increasing popularity over time such as studies focusing on mortality and survival (T-4) and Patient-related Outcomes and Perspectives of Clinicians (T-5). However, some research topics such as studies focusing on animal modeling (T-1), predictors of acute kidney injury, and dialysis access (T-10) exhibited a downward trend. CONCLUSION: Stakeholders of nephrology research may use these trends further to develop priorities and enrich the research agenda for the future.
BACKGROUND: Nephrology research is expanding, and harnessing the much-needed information and data for the practice of evidence-based medicine is becoming more challenging. In this study, we used the natural language processing and text mining approach to mitigate some of these challenges. METHODS: We analyzed 17,412 abstracts from the top-10 nephrology journals over 10 years (2007-2017) by using latent semantic analysis and topic analysis. RESULTS: The analyses revealed 10 distinct topics (T) for nephrology research ranging from basic science studies, using animal modeling (T-1), to dialysis vascular access-related issues -(T-10). The trend analyses indicated that while the majority of topics stayed relatively stable, some of the research topics experienced increasing popularity over time such as studies focusing on mortality and survival (T-4) and Patient-related Outcomes and Perspectives of Clinicians (T-5). However, some research topics such as studies focusing on animal modeling (T-1), predictors of acute kidney injury, and dialysis access (T-10) exhibited a downward trend. CONCLUSION: Stakeholders of nephrology research may use these trends further to develop priorities and enrich the research agenda for the future.
Authors: Ferhat D Zengul; Ayse G Zengul; Michael J Mugavero; Nurettin Oner; Bunyamin Ozaydin; Dursun Delen; James H Willig; Kierstin C Kennedy; James Cimino Journal: Intell Based Med Date: 2021-06-17