This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Evidence-based medicine (EBM) has transformed clinical decision-making by integrating the best available research evidence with clinical expertise and patients’ values and preferences. Over the past three decades, EBM has evolved from the critical appraisal of individual studies into a broader framework encompassing evidence synthesis, clinical practice guidelines, assessment of evidence certainty, research transparency, and shared decision-making. Despite these advances, contemporary EBM faces important challenges, including the rapidly increasing volume of research, delays in evidence synthesis and implementation, limited applicability of randomized controlled trials to heterogeneous real-world populations, and difficulties in individualizing population-level evidence. Emerging approaches—including real-world evidence, living evidence, learning health systems, precision medicine, and artificial intelligence (AI)—offer opportunities to address these limitations. Together, these approaches may enable a transition from static to continuously updated evidence, from population-average to more personalized evidence, and from a linear evidence pipeline to a learning evidence ecosystem in which clinical practice both uses and generates evidence. AI may further accelerate evidence retrieval, synthesis, updating, and individualized decision support, while introducing challenges related to reliability, bias, transparency, reproducibility, and accountability. Next-generation EBM should therefore be conceptualized not as a replacement for traditional EBM but as its evolution into a digitally connected, continuously learning evidence ecosystem. In the AI era, the foundational principles of EBM—source verification, critical appraisal, uncertainty assessment, integration of patient preferences, and accountable human judgment—will become increasingly important.
Evidence-based medicine (EBM) has transformed clinical decision-making over the past three decades. EBM emerged in response to growing concerns about variation in clinical practice, the underuse of effective interventions, the overuse of unnecessary care, and the difficulty of translating rapidly expanding medical research into clinical decisions [1,2].
EBM introduced a systematic approach to clinical decision-making by integrating the best available research evidence with clinical expertise and patients’ values and preferences [3]. It promoted the formulation of answerable clinical questions, systematic identification and critical appraisal of relevant evidence, and explicit application of evidence to patient care. Over time, EBM expanded beyond the appraisal of individual studies to include systematic reviews, clinical practice guidelines, structured assessment of the certainty of evidence, and shared decision-making, thereby strengthening the transparency and consistency of clinical decisions [4,5].
Despite these advances, important challenges remain. The volume and pace of research now exceed clinicians’ capacity to identify and evaluate relevant evidence, while systematic reviews and guidelines may become outdated as new evidence emerges [6]. Moreover, average treatment effects derived from randomized controlled trials (RCTs) may not adequately inform decisions for individual patients, particularly older adults with multimorbidity, polypharmacy, or characteristics that are underrepresented in clinical trials.
These challenges have stimulated growing interest in real-world evidence (RWE), living evidence, learning health systems (LHSs), precision medicine, and artificial intelligence (AI) [7]. Rather than replacing EBM, these approaches may extend its principles by enabling evidence to be generated, synthesized, updated, and applied more continuously and efficiently. This review examines the evolution and current limitations of EBM and explores how these emerging approaches may contribute to a next-generation digital evidence ecosystem.
Emergence and Core Principles of Evidence-Based Medicine
Emergence of EBM
The development of EBM was shaped by the work of Archie Cochrane, David Sackett, Gordon Guyatt, and others who advocated the systematic use of empirical evidence in clinical decision-making. The term evidence-based medicine became widely used in the early 1990s through the work of Guyatt and colleagues, who described EBM as a new paradigm for medical practice and education [1,2]. The subsequent development of the Cochrane Collaboration, systematic reviews, evidence-based clinical practice guidelines, and the GRADE approach extended EBM from the critical appraisal of individual studies to systematic evidence synthesis and explicit recommendations [2,4,5].
Three Components and Five Steps of EBM
EBM integrates the best available research evidence with clinical expertise and patients' values and preferences [3]. Clinical decisions therefore depend not only on research findings but also on individual patient circumstances, expected benefits and harms, feasibility, and patient priorities [3,8].
EBM is commonly practiced through five steps: Ask, Acquire, Appraise, Apply, and Assess. These steps provide a practical framework for translating clinical uncertainty into evidence-informed decisions and have also become a foundation for EBP education [8]. The Sicily Statement emphasized that EBP requires not only knowledge but also the skills and attitudes necessary to formulate questions, search for and critically appraise evidence, apply it in practice, and evaluate performance [8]. Albarqouni and colleagues subsequently developed a consensus-based set of core EBP competencies encompassing question formulation, evidence searching and appraisal, interpretation of effects and uncertainty, application of evidence, and integration of patient values and preferences [9]. EBM can therefore be understood not only as an approach to using research evidence but also as a set of competencies for rational decision-making under uncertainty [8,9].
Expansion of EBM
EBM has expanded from individual clinicians' appraisal of research toward a broader system encompassing evidence generation, synthesis, recommendations, implementation, and evaluation, and from medicine toward evidence-based practice across health professions [2,8].
Systematic reviews became central to evidence synthesis, while GRADE provided a structured framework for assessing the certainty of evidence and distinguishing it from the strength of recommendations [4,10]. The subsequent development of GRADE Evidence-to-Decision frameworks further emphasized that recommendations should consider not only the certainty of evidence but also the balance of benefits and harms, patients' values and preferences, resource use, equity, acceptability, feasibility, and context [11].
EBM consequently evolved from the appraisal of individual studies into a broader system in which evidence is systematically synthesized, translated into recommendations, and incorporated into clinical and health care decisions [2,5,11].
Transparency, Reproducibility, and Patient-Centeredness
Research registration, reporting guidelines, data sharing, open science, and shared decision-making have further expanded EBM by improving the transparency, reproducibility, and patient-centeredness of research and clinical decisions. Reporting guidelines such as CONSORT, STROBE, STARD, and PRISMA promote more complete and transparent reporting, thereby facilitating critical appraisal and evidence synthesis [12]. Trial registration and protocol availability can help identify selective reporting and undisclosed changes in study outcomes or methods [13].
The FAIR principles emphasize that scientific data should be Findable, Accessible, Interoperable, and Reusable [14], an increasingly important requirement for digital and AI-based evidence systems. Shared decision-making (SDM) similarly operationalizes a core principle of EBM by integrating information about available options, benefits, harms, and uncertainty with patients' values, preferences, and individual circumstances [15].
Together, these developments have strengthened both the trustworthiness and usability of evidence and the role of patient-centered decision-making within EBM.
Limitations of Contemporary EBM
Despite these advances, contemporary EBM faces several limitations. The rapidly expanding volume of research exceeds clinicians' capacity for continuous appraisal, while delays persist between evidence generation, synthesis, and implementation [6,16]. RCTs remain essential for estimating causal effects but may not adequately represent older adults, patients with multimorbidity or polypharmacy, and other populations encountered in routine practice. Observational studies, pragmatic trials, and routinely collected health data may therefore provide important complementary evidence, particularly when RCT evidence is limited or has restricted applicability [17,18].
Patient-centeredness also remains incompletely realized. Contemporary EBM requires greater emphasis on evidence that is usable in clinical practice and can be integrated with clinical context, professional expertise, shared decision-making, and individual patient preferences [16]. Moreover, conventional EBM has largely followed a linear pathway from research to evidence synthesis, recommendations, and practice, with less systematic use of data generated during routine care to produce new knowledge. Addressing this gap requires a more continuous connection between evidence generation and clinical practice, providing an important rationale for the development of next-generation EBM.
Concepts Shaping the Next Generation of EBM
The four approaches shaping next-generation EBM are complementary rather than competing and are summarized in Table 1.
Real-World Evidence
Real-world data (RWD) derived from electronic health records, administrative claims, registries, and digital health technologies can be used to generate evidence on the use, benefits, and harms of health interventions [18]. RWE can complement evidence from RCTs by providing information on broader and more heterogeneous populations, long-term outcomes, rare adverse events, and treatment patterns in routine clinical practice [17,18]. Advances in causal inference methods, including target trial emulation, may further improve the validity of causal estimates derived from observational data [19].
Living Evidence
Traditional systematic reviews and clinical practice guidelines may rapidly become outdated as new research accumulates. Living systematic reviews address this limitation by continually incorporating relevant new evidence as it becomes available [6]. Living guidelines extend this approach from evidence synthesis to recommendations, allowing guidance to be updated when new evidence is likely to change clinical recommendations [20]. Together, these approaches represent a shift from periodic publication toward continuous evidence surveillance, synthesis, and updating.
Learning Health Systems
A learning health system (LHS) integrates routinely generated clinical data, scientific evidence, and clinical experience through a continuous feedback process in which practice generates data, data generate knowledge, and knowledge informs subsequent practice [21]. In this model, clinical care becomes not only the endpoint of evidence implementation but also an important source of new evidence, creating a continuous cycle between evidence generation and practice [21].
Precision Medicine
Precision medicine seeks to account for individual variability in genes, environment, and lifestyle to improve disease prevention and treatment [22]. More broadly, genomic, biomarker, phenotypic, environmental, and behavioral information may be used to characterize patient heterogeneity and support more individualized clinical decisions [7,22]. When integrated with conventional research evidence and RWE, precision medicine may help move clinical decision-making beyond population-average treatment effects toward better estimation of outcomes for individual patients. Such personalization, however, requires adequate validation, predictive performance, and evidence of clinical utility before implementation in routine practice [7].
Evidence Ecosystems
Traditional EBM has largely followed a linear pathway from primary research through evidence synthesis and guideline development to clinical practice. A digital evidence ecosystem extends this model by continuously connecting evidence generation, synthesis, recommendations, implementation, outcomes, and new evidence [23].
Learning health systems emphasize the use of routinely generated clinical data to support continuous learning, whereas evidence ecosystems encompass a broader infrastructure for generating, synthesizing, updating, disseminating, and applying evidence [23]. Within such a system, living systematic reviews can inform living guidelines, recommendations can be incorporated into digital decision-support systems, and clinical outcomes can generate RWD that contribute to subsequent evidence generation and updating [20,21,23]. Achieving this continuous cycle requires trustworthy evidence synthesis, interoperable data and digital systems, and effective mechanisms for translating evidence into practice [23].
How Artificial Intelligence May Transform EBM
AI, including machine learning and large language models (LLMs), has the potential to support multiple stages of the EBM process. At Ask, AI may assist in structuring clinical questions; at Acquire, it can support literature searching, study identification, and screening; and at Appraise, AI tools are increasingly being evaluated for study classification, data extraction, and other components of evidence synthesis [24]. At Apply, AI may help integrate patient-level information with external evidence to support individualized risk and benefit estimation, while at Assess, it may facilitate analysis of clinical outcomes and identification of evidence gaps.
AI may be particularly valuable for living evidence systems, in which continuous surveillance and updating require substantial human resources. Automated or semi-automated tools can assist with identifying new studies, screening citations, extracting structured information, and determining when evidence syntheses may require updating [24].
However, AI-generated outputs cannot be assumed to be reliable. Potential problems include inaccurate or fabricated information, algorithmic bias, limited explainability, model instability, and challenges in reproducibility, transparency, and accountability. AI therefore reinforces rather than replaces core EBM principles: source verification, critical appraisal, assessment of uncertainty, transparency, and accountable human judgment. These potential applications, risks, and safeguards are summarized in Table 2.
Next-Generation EBM as a Digital Evidence Ecosystem
Next-generation EBM can be conceptualized as embedding the traditional principles and processes of EBM within a digitally connected and continuously learning evidence system [23]. Its evolution may be characterized by three major transitions: from static to living evidence, from population-average to more personalized evidence, and from a linear evidence pipeline to a learning evidence ecosystem [7,20,22,23].
In such a system, evidence syntheses and recommendations are continuously updated as new knowledge emerges; population-level evidence is integrated with individual risk, characteristics, and preferences; and data generated during routine clinical practice contribute to subsequent evidence generation. RWE, living evidence, LHSs, precision medicine, and AI can therefore be viewed not as alternatives to EBM but as complementary components of a continuous evidence cycle [18,20-24].
Important challenges remain, including data interoperability, privacy and governance, and the reliability, transparency, equity, and accountability of AI-supported systems. A trustworthy digital evidence ecosystem will therefore require integration across clinical epidemiology, evidence synthesis, health informatics, implementation science, patient participation, and data and AI governance. The goal is not to replace the foundational principles of EBM, but to enable those principles to operate within a more timely, adaptive, personalized, and continuously learning health care system [25].
Conclusion
EBM has evolved from individual appraisal of research into a broader framework encompassing evidence generation, synthesis, recommendations, implementation, and evaluation. Rapidly expanding evidence, implementation delays, limited generalizability, and the need for individualized decisions now require further evolution.
Next-generation EBM should integrate RWE, living evidence, LHSs, precision medicine, shared decision-making, and AI within a continuously learning evidence ecosystem. The AI era therefore does not signal the end of EBM; rather, it makes its central tasks—determining which evidence can be trusted, how uncertainty should be evaluated, and how evidence should be applied to individual patients—increasingly important.
Notes
Conflict of Interest
Soo Young Kim has been a member of the editorial board of the Journal of Evidence-Based Practice since 2025. However, he was not involved in the peer reviewer selection, evaluation, or decision process of this article. No other potential conflicts of interest relevant to this article were reported.
Funding
None.
Data Availability Statement
Data sharing is not applicable to this article as no new data were created or analyzed in this study.
Ethics Approval and Consent to Participate
Not applicable.
Authors' Contributions
Conceptualization: Kim SY. Methodology: Kim SY. Writing - original draft: Kim SY. Writing - review & editing: Kim SY.
Acknowledgments
None.
Table 1.
Complementary Approaches Shaping the Next Generation of Evidence-Based Medicine
Approach
Primary purpose
Main data sources
Temporal orientation
Contribution to EBM and key challenge
Real-world evidence
Evaluate effectiveness and safety in routine practice
Electronic health records, claims, registries, and patient-generated data
Historical or prospective
Improves generalizability beyond RCTs; vulnerable to confounding, bias, and variable data quality
Living evidence
Keep syntheses and guidance current
Continuous literature surveillance and newly available studies
Continuous or frequent updating
Reduces the delay between evidence generation and guidance; requires sustainable surveillance and updating workflows
Learning health systems
Convert practice data into knowledge and feed knowledge back into care
Clinical, operational, and patient-outcome data
Continuous feedback cycles
Connects evidence generation with implementation; depends on interoperability, governance, and organizational capacity
Precision medicine
Optimize decisions for individual patients
Genomic, biomarker, clinical, environmental, and lifestyle data
Individual-level, real-time, or repeated assessment
Extends decisions beyond average treatment effects; requires robust validation and attention to equity and access
4. Institute of Medicine. Finding what works in health care: standards for systematic reviews. Washington (DC), National Academies Press. 2011.
5. Institute of Medicine. Clinical practice guidelines we can trust. Washington (DC), National Academies Press. 2011.
6. Elliott JH, Turner T, Clavisi O, Thomas J, Higgins JPT, Mavergames C, et al. Living systematic reviews: an emerging opportunity to narrow the evidence-practice gap. PLoS Med 2014; 11: e1001603.
7. Beckmann JS, Lew D. Reconciling evidence-based medicine and precision medicine in the era of big data: challenges and opportunities. Genome Med 2016; 8: 134.
8. Dawes M, Summerskill W, Glasziou P, Cartabellotta A, Martin J, Hopayian K, et al. Sicily statement on evidence-based practice. BMC Med Educ 2005; 5: 1.
9. Albarqouni L, Hoffmann T, Straus S, Olsen NR, Young T, Ilic D, et al. Core competencies in evidence-based practice for health professionals: consensus statement based on a systematic review and Delphi survey. JAMA Netw Open 2018; 1: e180281.
10. Guyatt GH, Oxman AD, Vist GE, Kunz R, Falck-Ytter Y, Alonso-Coello P, et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ 2008; 336: 924-6.
11. Alonso-Coello P, Schunemann HJ, Moberg J, Brignardello-Petersen R, Akl EA, Davoli M, et al. GRADE Evidence to Decision (EtD) frameworks: a systematic and transparent approach to making well informed healthcare choices. 1: introduction. BMJ 2016; 353: i2016.
12. Kim SY. Reporting guidelines. Korean J Fam Med 2009; 30: 62.
13. De Angelis C, Drazen JM, Frizelle FA, Haug C, Hoey J, Horton R, et al. Clinical trial registration: a statement from the International Committee of Medical Journal Editors. N Engl J Med 2004; 351: 1250-1.
14. Wilkinson MD, Dumontier M, Aalbersberg IJJ, Appleton G, Axton M, Baak A, et al. The FAIR Guiding Principles for scientific data management and stewardship. Sci Data 2016; 3: 160018.
15. Elwyn G, Frosch D, Thomson R, Joseph-Williams N, Lloyd A, Kinnersley P, et al. Shared decision making: a model for clinical practice. J Gen Intern Med 2012; 27: 1361-7.
18. Sherman RE, Anderson SA, Dal Pan GJ, Gray GW, Gross T, Hunter NL, et al. Real-world evidence-what is it and what can it tell us? N Engl J Med 2016; 375: 2293-7.
21. Friedman CP, Rubin JC, Brown JS, Buntin MB, Corn M, Etheredge LM, et al. Toward a science of learning systems: a research agenda for the high-functioning Learning Health System. J Am Med Inform Assoc 2015; 22: 43-50.
23. Rajit D, Johnson A, Callander E, Teede H, Enticott J. Learning health systems and evidence ecosystems: a perspective on the future of evidence-based medicine and evidence-based guideline development. Health Res Policy Syst 2024; 22: 4.
25. Institute of Medicine. Best care at lower cost: the path to continuously learning health care in America. Washington (DC), National Academies Press. 2013.
