rtificial intelligence (AI) and real-world evidence (RWE) are no longer emerging concepts; they are becoming the connective tissue across the entire life-science ecosystem, and they reflect a growing urgency: how do we scale data-driven innovation responsibly while keeping patients, quality, and societal trust at the center?
The Expanding Promise: AI as a Force Multiplier in Healthcare
These themes surfaced across plenary sessions and in multiple scientific tracks at the DIA India Annual Meeting 2025. AI, machine learning, and data science are rapidly shifting from experimental tools to operational necessities. From early research to bedside care, AI is accelerating pattern recognition, decision support, and process optimization. AI can democratize advanced therapeutics (including “living drugs”) by reducing development timelines, enabling more precise patient identification, and scaling access strategies for large populations.
What RWD/RWE Must Deliver Today
RWE must answer meaningful clinical and scientific questions, not simply produce more data, which challenges the field to define:
- What we collect
- Why it matters
- How we ensure quality
- Where RWE fits in the decision-making continuum.
It remains critical to separate the two often conflated concepts of Real-World Data (RWD) and Real-World Evidence (RWE). Elevating the quality of data to the standard of evidence requires a structured approach that focuses on clear research questions, fit-for-purpose data curation, transparent analytical frameworks, and strong governance and documentation. Better data alone does not yield better evidence; methodological rigor does.
Several recent applications illustrate how structured approaches are translating RWD into actionable evidence. During the COVID-19 pandemic, one company’s mHealth app used machine-learning models to analyze symptom data from millions of participants, generating near-real-time insights and demonstrating the scalability of remote health research. Analytical frameworks such as Trial Pathfinder have also used real-world data sets to evaluate clinical trial eligibility criteria, showing that many commonly used criteria unnecessarily exclude patients who could benefit from treatment. Beyond research design, RWE is increasingly informing regulatory and scientific decisions—for example, disease-progression modeling using the Alzheimer’s Disease Neuroimaging Initiative database, drug-drug interaction analyses using large EHR data sets, and registry- and EHR-based studies supporting comparative effectiveness and label-expansion decisions in areas such as rheumatoid arthritis and oncology.
Regulatory and Operational Challenges: A Call for Harmonization
Regulators globally are increasingly receptive to real-world data methods, yet variation in expectations remains a challenge. Practical hurdles exist in study approvals, data quality assessment, and cross-functional alignment; industry and regulators must continue to collaborate to build mutual confidence in real-world data methodologies.
Transparency and governance are also non-negotiable aspects of data science. Responsible AI frameworks encompass explainability, bias mitigation, auditability, and robust accountability structures (see appended list). These principles will shape how AI-generated insights are evaluated by regulators and adopted in healthcare decision-making.
Regulators are increasingly open to incorporating real-world evidence alongside traditional trial data. For example, the US Food and Drug Administration accepted real-world data from an electronic health record database to contextualize results from the single-arm JAVELIN Merkel 200 trial, supporting the accelerated approval of avelumab for metastatic Merkel cell carcinoma. In this case, retrospective observational data on chemotherapy-treated patients helped benchmark trial outcomes in a rare cancer setting where randomized trials were difficult to conduct.
Pharmacovigilance: From Automation to Augmentation
Safety functions are experiencing some of the fastest AI adoption across the pharmaceutical value chain across three major themes:
Early Signal Detection with RWD: Well-curated real-world data sets enable earlier, more sensitive identification of safety signals than traditional reporting alone. These include patterns emerging from electronic health records, insurance claims databases, patient communities, and digital platforms/databases. Historical cases illustrate the value of such approaches. Analyses of electronic health record data detected a signal linking rofecoxib with acute myocardial infarction several years before strong signals emerged in traditional spontaneous reporting systems such as WHO VigiBase. Similarly, long-term observational analyses helped evaluate the potential association between pioglitazone and bladder cancer, demonstrating how real-world data sets can help clarify emerging safety concerns.
Workforce Transformation: The future PV workforce must blend domain expertise with data literacy. AI will not replace safety professionals, but it will redefine roles and move teams toward interpretation, contextual judgment, and cross-functional decision-making. Emerging applications already illustrate this shift. Generative AI tools are being deployed for personalized pharmacovigilance training through chatbot-enabled learning platforms, delivering adaptive learning modules and automated compliance tracking. Similar systems are also being used to triage social media and conduct web-based queries on adverse event reporting, providing consistent responses based on internal guidelines while routing complex safety issues to pharmacovigilance specialists.
Real-World AI, Not “AI Theatre”: As a recently published article heralding “the end of AI theatre” explains, use of AI must now shift from pilots to fully operationalized, value-producing systems that provide scalable automation, embedded quality controls, and continuous monitoring of model performance. Skilled augmentation, not blind automation, is the sustainable path forward.
Clinical Trials: Smarter, Faster, More Innovative, More Inclusive
AI and RWE are also reshaping trial design, site selection, and patient engagement.
AI for Patient-Centric Recruitment can address enrollment bottlenecks by narrowing eligibility criteria, predicting site performance, and identifying underrepresented patient groups. These approaches are making trials more inclusive while reducing recruitment delays. Emerging case studies highlight the potential impact of AI-enabled recruitment strategies. In a cardiovascular phase 3 trial, AI-supported patient identification and site optimization accelerated enrollment and improved demographic diversity while reducing operational costs (see below). Similar approaches in rare neurological and pediatric immunology studies have enabled rapid identification of eligible patients across multiple countries, improved consent and engagement rates through personalized communication, and in some cases shortened overall study timelines.
- Reduce development time by ~10% or more, as seen in analyses of decentralized and adaptive clinical trials.
- Support external control arms using registry and EHR data sets in rare diseases and oncology.
- Enable early decision-making through adaptive interim analyses and predictive analytics.
- Minimize patient burden through decentralized components, which have been associated with 7%–10% reductions in screen failures and fewer protocol amendments.
All the above align strongly with global trends in evidence modernization.
The Path Forward: Responsible Acceleration in India
- AI and RWE are complementary, and their integration is unlocking new efficiencies across clinical development, PV, and medical affairs.
- Governance and quality remain the backbone. Innovation cannot come at the cost of reliability and safety.
- Cross-functional (departmental) collaboration is essential to fully realize the benefits of data-driven methods throughout the evidence-generation lifecycle.
- Human judgment remains central, especially as roles evolve in AI-augmented environments.
- Regulators and industry are converging, but sustained dialogue is essential to maintain trust and consistency.
Conclusion
The DIA India Annual Meeting 2025 demonstrated that the life-sciences ecosystem is entering a new phase that combines real-world evidence, AI, and responsible innovation to accelerate access, improve safety, and strengthen patient-centricity in India. Rigor, transparency, and shared learning must remain essential as national and ultimately global adoption grows.