[2026 CPhI Korea Insight] RWD-Driven Evidence Life Cycle: From Drug Development to RMP-Based Safety Management
페이지 정보
본문
SeltaSquare identifies industry shifts early and delivers meaningful insights to our clients.
At CPhI Korea 2026, held August 25–27 at COEX Seoul, Hyejung Lee, Head of SeltaSquare's Integrated Evidence (IE) Strategy Division, presented a session titled "RWD-Driven Evidence Life Cycle: From Drug Development to RMP-Based Safety Management." The session covered how Real-World Data (RWD) research is applied across the entire drug development lifecycle, the conditions required for Real-World Evidence (RWE) derived from RWD to support regulatory decision-making, and the direction of SeltaSquare's RMP-based RWD safety studies and AI applications.
RWD/RWE: A Core Asset Across the Entire Drug Development Lifecycle
RWD refers to health-related data routinely collected from a variety of sources, including electronic health records (EHR), health insurance claims data, and patient registries. RWE is the clinical evidence on a drug's potential benefits or risks derived from analyzing this RWD. Simply accumulating large volumes of data does not, by itself, produce reliable evidence — RWE that can genuinely support decision-making only emerges when a well-formulated research question is followed by appropriate study design and analysis.
The scope of RWD spans the entire drug development lifecycle: from identifying disease epidemiology and unmet needs in early development, to clinical trial design, external control arm construction, and generating evidence for post-marketing safety and efficacy evaluation and indication expansion. The presentation used the case of Sotorasib to illustrate RWD's role in complementing the interpretation of single-arm trial results — showing an Evidence Life Cycle in which evidence generated at each stage, from clinical trials through the post-marketing phase, carries forward seamlessly into the next.
On the study design side, a range of observational study designs — cohort studies, case-control studies, nested case-control studies, and case series — are widely used. For safety studies in particular, an active-comparator new-user design can be considered depending on the research question and data characteristics, helping to reduce confounding by indication and bias arising from the inclusion of prevalent users. The ICH M14 guideline sets out principles for non-interventional studies using RWD for drug safety evaluation, covering research question specification, feasibility assessment, protocol and statistical analysis plan development, and data extraction, analysis, and reporting. Notably, it emphasizes clearly defining the research question, design, and analytic methods in advance — rather than selecting favorable definitions or methods after seeing the primary results. When selecting a data source, it is essential to weigh the strengths and limitations of each source and choose the one that is genuinely fit-for-purpose for the study objective.
Global Shifts in RWD and RMP Regulation
Global regulators have continued to sharpen their standards for RWD use over the past decade. In the United States, following the 21st Century Cures Act in 2016, the FDA has expanded its RWE framework through the RWE Program Framework (2018) and the Advancing RWE Program (2023). Sponsors selected for this program can discuss their RWE approach with the FDA before developing a protocol or initiating a study. The FDA has also announced plans to revise existing RWE-related guidance by December 2026, incorporating lessons learned from the program. In Europe, the implementation of ICH M14 and the EMA's ENCePP methodology are similarly clarifying the principles governing non-interventional pharmacoepidemiological studies.
In Japan, the 2018 revision of GPSP explicitly defined Post-Marketing Database Studies (PMDS) as a pharmacovigilance activity. Following an MHLW notification in July 2024, Japan shifted from uniformly conducting PMS for all approved products to a model in which the need for and method of post-marketing studies are determined based on the specific research question. As a result, the number of approved products for which PMS was not conducted rose from an annualized 17.8 to 62.4, and the proportion of new active substances undergoing database studies increased from 12.8% to 27.3%. In Korea, the re-examination (재심사) system has likewise been abolished in favor of an RMP-centered post-marketing safety management framework — making it increasingly important to define researchable questions based on product-specific safety concerns and to select data and study designs suited to that purpose.
SeltaSquare's RMP-Based RWD Safety Studies and In-House AI Capability
SeltaSquare is currently conducting two RMP-based RWD safety studies. Both are retrospective, non-interventional studies using multi-institutional electronic health records. The first study targets adults with metabolic conditions — hypertension, type 2 diabetes, dyslipidemia, and obesity — using multi-institutional EMR data from 2016–2025 to develop a technology that maps institution-specific diagnosis, lab, and prescription data to standardized terminologies (MedDRA, LOINC, and others). It also aims to integrate unstructured clinical records to jointly detect RMP safety concerns and Adverse Events of Special Interest (AESI).
The second study focuses on patients prescribed neuropathic pain medications in rheumatology departments, using unstructured clinical narratives and natural language processing (NLP) to detect symptom-based adverse events that are difficult to capture through diagnosis codes alone. The study will ultimately evaluate the comparative safety between gabapentinoid-class and antidepressant-class medications.
These two studies share a common direction, characterized by four elements: ① Beyond Structured Data — integrating unstructured clinical records to detect adverse events actually experienced by patients; ② Standardized & Interoperable — achieving interoperability across multi-institutional data through standardized terminology mapping; ③ Detect More, Earlier — enabling additional and earlier detection of adverse events compared to diagnosis-code-only approaches; and ④ extending beyond simple detection into Comparative Safety and Causal Inference research — together forming an "RMP-to-RWE research case."
SeltaSquare treats AI not merely as an automation tool but as a research capability for generating clinically meaningful safety evidence. Its AI engineers and data scientists, working alongside drug safety experts with clinical and pharmacovigilance experience, are turning safety information that structured data alone might miss into evidence that can inform actual research and regulatory decisions.
SeltaSquare Insight: Beyond Data, Toward Reproducible Evidence
RWD is a core asset used across the entire drug development lifecycle — from understanding disease and clinical development through post-marketing safety management and follow-on development. Using RWD for decision-making requires a clearly defined research question, data that is fit for the purpose, and a study design and outcome variables validated in advance. In RMP-based safety research, AI and NLP serve as tools to detect adverse events that would be difficult to identify from structured data alone and to standardize multi-institutional data — completing an Evidence Life Cycle that connects evidence from one stage to decision-making in the next, rather than treating development and post-marketing research as separate.
"SeltaSquare is an Evidence Partner that combines RWD, pharmacovigilance, and AI capabilities to build the Evidence Life Cycle together."
