Workshop 1 was held at Novo Nordisk headquarters in Copenhagen, Denmark, in March 2023. The workshop opened with a plenary introduction to the study and theme 1, followed by a breakout discussion in which facilitators guided structured brainstorming and professional writers documented participants’ written and verbal contributions. Participants then reconvened for plenary introductions to themes 2 and 3, with a corresponding breakout discussion after each introduction. The same participants remained in their assigned groups across all 3 breakout sessions. The workshop concluded with a plenary review of the notes from all sessions and collection of participant feedback. Its primary objective was to initiate discussion and evaluation of the 3 core themes guiding the study’s examination of digital transformation. The participants were selected through targeted invitations sent via professional networks, industry associations, and direct outreach to pharmaceutical research and development leaders, consultants, and digital health experts. Eligibility required active involvement in clinical research, digital transformation, or regulatory innovation. Workshop 1 recruitment involved issuing 60 invitations, of which 47 participants attended. Random assignment to 3 working groups (8‐9 participants each) ensured diversity of perspectives and balanced group composition.

Workshop 1 comprised 3 parallel working groups, all addressing the 3 themes:

Facilitators guided structured brainstorming sessions, capturing all ideas and experiences in digital worksheets [38]. Each group presented their findings in a plenary session, generating 110 distinct transformation topics. Participants then provided feedback on the presentations.

Written workshop notes, session transcripts, and participant feedback were documented and archived for traceability.

After workshop 1, the researchers conducted a targeted literature search in the Web of Science Core Collection to triangulate theme 1 findings with external evidence (Multimedia Appendix 3) [39]. The search covered publications from 2012 to 2023 and used a high-precision topic query comprising “FAIR data,” “machine-readable data,” “discoverable data,” “interoperable data,” and “democratic data” [40]. The query was combined with the predefined Web of Science life-sciences categories to prioritize precision over recall.

In April 2024, researcher-prepared worksheet notes, plenary summaries, and consolidated outputs from workshop 1 were entered into ChatGPT-4.0 (OpenAI) to generate concise summaries of ideas and provisional groupings of both common and unique categories. These outputs were used only as starting points for the manual review and coding described below.

The follow-up survey was administered online between March 2023 and April 2023, 4 weeks after workshop 1. It was sent to all 47 workshop 1 participants to preserve a shared frame of reference for prioritizing the categories generated during that workshop and to permit direct within-cohort comparison. The survey was not distributed to new stakeholders; additional stakeholder perspectives were incorporated during workshop 2. Weekly reminder emails were sent to nonrespondents, and the survey closed after the planned 2-week collection window.

The online survey instrument was developed from workshop 1 outputs and structured into sections covering participant information, theme prioritization, collaboration preferences, and additional comments. Respondents ranked 17 survey topics (8 in theme 1, 4 in theme 2, and 5 in theme 3) using a ranking scale and provided their name, company affiliation, and role (Multimedia Appendix 4).

Survey responses were collected through an online platform, reviewed for completeness, and exported to Microsoft Excel for analysis. Quality-control procedures included logical consistency checks, verification of participant identities against the invited sample, and data-integrity validation.

Participants ranked the stakeholder-derived topics independently within each of the 3 themes by assigning a unique rank to each topic according to its relative priority within that theme.

The rankings assigned by all participants were tabulated by theme and aggregated for each survey topic to generate an overall ranking score. The aggregated ranking scores were then ordered from highest to lowest within each theme to produce the final prioritized ranking of survey topics. Descriptive summaries of the ranking distributions and overall ranking scores were generated to characterize patterns of stakeholder prioritization across respondents. No inferential statistical analyses were performed because the survey was designed to prioritize stakeholder-derived topics rather than to test hypotheses.

The survey rankings were subsequently incorporated into the cross-phase, mixed methods integration as the quantitative component for comparison with the qualitative findings from workshop 1 and with the implementation tensions identified during workshop 2.

Workshop 2 was held at the DPHARM (Disruptive Innovations to Modernize Clinical Research) conference in Boston, United States, in October 2024. The workshop opened with an introductory plenary session, followed by panel discussions on each of the 3 themes, with 5 digital-transformation experts participating in each panel. Each panel discussion was followed by an interactive breakout discussion involving the panelists and all participants. A closing plenary presentation summarized the findings across the 3 theme discussions. All workshop content was audio-recorded. The objective was to build on the initial discussions from workshop 1 and the results of the survey through a more refined examination of the study’s core themes and to support the comprehensive implementation-focused analysis described below. Invitations were sent to workshop 1 and survey participants and additional subject-matter experts (conference participants) to preserve continuity with earlier phases while broadening the discussion with implementation, regulatory, and digital health expertise relevant to implementation barriers. A total of 35 participants attended, including 26 from prior phases.

Discussions were audio-recorded and automatically transcribed using Zoom’s audio transcription functionality. The researchers reviewed the full transcripts for accuracy before thematic coding. Each panel corresponded to 1 of the 3 themes so that implementation barriers could be examined using the same outcome, pathway, and measurement structures that were used in workshop 1 and the survey [42].

Workshop 2 data were analyzed across themes using the analytic procedures described in the Integrated Data Analysis subsection. These procedures included identification and organization of statements relating to implementation barriers, trade-offs, and competing demands; classification and consolidation of related findings; assessment of their prominence; integration with workshop 1 and survey findings through tension, mixed methods, gap, convergence, and evolution analyses.

The researchers derived implementation tensions [41] from the workshop 2 transcripts and survey outputs to identify and characterize pressures, frictions, trade-offs, and competing forces affecting the translation of evidence-based digital transformation into routine operational use. A 5-step analytic procedure was followed:

This process supported the synthesis and prioritization of dominant implementation tension domains.

ChatGPT-4.0 supported the analysis by using deidentified workshop 2 transcript excerpts, survey outputs, and structured researcher-developed prompts as inputs. AI assistance was used to organize relevant excerpts, provisionally classify statements within the predefined tension types, consolidate related coded statements and recurring patterns, support comparisons of their relative prominence across discussions, and organize comparisons with survey priorities for cross-phase validation. The researchers reviewed all AI-assisted outputs against the original source data and confirmed, modified, or rejected the proposed classifications and groupings through verification and consensus before determining the final tension domains. The model and version, inputs, prompts, and outputs for each AI-assisted step are provided in Multimedia Appendix 1.

The researchers integrated evidence across the 3 study phases to determine whether priorities identified in earlier phases were sustained, refined, or challenged in later phases and to connect stakeholder priorities with implementation realities. Findings were interpreted sequentially and comparatively, rather than as isolated phase-specific results.

Gap analysis was subsequently used to identify discrepancies between consensus priorities and implementation readiness. Workshop 1 aspirations were compared with workshop 2 barrier discussions; missing enablers such as regulatory frameworks and interoperability were mapped; and identified gaps were classified as structural, procedural, or cultural.

ChatGPT-4.0 supported the mixed methods integration using the complete, researcher-prepared analytical outputs from all 3 study phases: workshop 1 qualitative findings, survey results, and workshop 2 qualitative findings, together with structured, researcher-developed prompts. These inputs were provided to the model to generate preliminary cross-phase comparison matrices, convergence summaries, and gap classifications. The AI-generated outputs were reviewed and refined by the researchers through manual comparison with the complete input data to verify accurate representations and calculated values, followed by data adjudication and consensus. Representative prompts, inputs, and outputs are provided in Multimedia Appendix 1

The researchers used convergence analysis to compare findings across workshop 1, the follow-up survey, and workshop 2. Workshop 1 categories were mapped to the 17 survey topics, survey rankings were normalized within each theme, and workshop 2 transcripts were coded for thematic emphasis, contextual relevance, and semantic alignment with the earlier phases.

The researchers reviewed areas of agreement and discrepancy through structured adjudication meetings, with decisions documented for auditability. Integrated comparison matrices were used to classify findings as high convergence, partial convergence, or divergence. This process identified which priorities remained stable across phases, which evolved, and which revealed implementation tensions.

The ChatGPT-4.0 tool outcomes from mixed methods integration and gap analysis were used in this section. Representative prompts, inputs, and outputs are provided in the mixed methods section of Multimedia Appendix 1.

For workshop 1, 60 invitations were issued and 47 participants attended. Participants represented large pharmaceutical companies (n=32, 68.1%), midsized pharmaceutical companies (n=3, 6.4%), consulting firms (n=5, 10.6%), and technology or other organizations (n=7, 14.9%; Table S8 in Multimedia Appendix 5).

The follow-up survey was completed by 26 of 47 (55.3%) participants, with no missing participant identification or company affiliation data. Respondents were from Denmark (n=9), the United States (n=8), the United Kingdom (n=3), Switzerland (n=2), Belgium (n=2), Germany (n=1), and other regions (n=1). By organization type, respondents represented large pharmaceutical companies (n=17, 65.4%), midsized pharmaceutical companies (n=4, 15.4%), small-sized pharmaceutical companies (n=1, 3.8%), and other (academic/standards; n=2, 7.7%; Table S9 in Multimedia Appendix 5).

Workshop 2 included 35 participants, of whom 26 had participated in prior phases. Participants represented large pharmaceutical companies (18/35, 51.4%), midsized pharmaceutical companies (4/35, 11.4%), clinical research organizations or consulting organizations (3/35, 8.6%), technology or digital health organizations (7/35, 20%), and patient advocacy or other organizations (3/35, 8.6%; Table S10 in Multimedia Appendix 5).

Participants represented 8 geographic groupings from Europe and North America and multiple pharmaceutical, academic, consulting, regulatory, and technology organizations. Workshop 1 included 47 participants from 15 companies (large to midsized), 26 (55.3%) participants completed the follow-up survey, and workshop 2 included 35 participants from 25 companies. Because participants could contribute to more than one phase, attendance totals are phase-specific and are not additive. Detailed geographic distribution and participant continuity across phases are provided in Tables S6 and S7 in Multimedia Appendix 5; detailed company representation and phase-specific engagement are provided in Tables S8 to S10 in Multimedia Appendix 5.

Cross-phase synthesis identified recurring priorities involving patient-centered outcomes, access and equity, FAIR and democratic data, collaboration, innovation, efficiency, and governance-related measurement needs. These priorities were accompanied by implementation barriers involving governance, interoperability, change management, safety standards, competition, and privacy protection. The integrated findings are presented in the following sections on implementation tensions; mixed methods integration and gap analysis; and convergence and evolution.

The study identified 5 industry-wide tensions across 3 themes. Each tension represented both a pressure to advance digital transformation and a competing constraint that could limit implementation.

The first tension was between innovation speed and safety standards. Participants described pressure to adopt digital technologies quickly, while regulatory and patient-safety requirements constrained the pace of adoption. The proposed resolution was a balanced approach that maintains safety while accelerating adoption where appropriate.

The second tension was between collaboration and competition. Participants not only supported precompetitive collaboration on common challenges but also identified the need for individual company differentiation and protection of intellectual property. The proposed resolution was to establish clear boundaries between collaborative and competitive areas.

The third tension was between technology adoption and change management. New digital tools and platforms were viewed as opportunities for transformation, but their implementation depended on training, workflow integration, and cultural change. The proposed resolution was an integrated approach that addresses technology and organizational change together.

The fourth tension was between patient preferences and operational efficiency. Participants emphasized individual preferences and diverse patient needs while recognizing the pressures of operational efficiency and cost reduction. The proposed resolution was to use patient-centric approaches that maintain operational viability.

The fifth tension was between data democratization and privacy protection. Industry-wide learning and collaboration were balanced against the protection of patient data and competitive information. The proposed resolution was to apply FAIR data principles with appropriate privacy safeguards.

Table 5 presents the survey ranking scores, workshop 2 coverage percentages, and gap scores for each topic.

Theme 1 concentrated more than expected on data-and-automation outcomes (+16.16%), with additional emphasis on the patient-provider relationship (+3.31%) and a modest lift in financial outcomes (+0.74%) and collaboration (+0.47%). In contrast, FAIR and democratic data (−10.83%) and sustainable scale (−9.51%) received less attention than their survey importance would suggest, and access and equity were slightly below expectations (−1.49%), while curative/personalized/preventive sat near alignment (+1.16%). Overall, the conversation tilted toward near-term data-driven outcome narratives, while foundational stewardship and scalable operating models were comparatively underserved.

Theme 2 devoted substantially more airtime to automation and standardization than expected (+27.90%) and also gave extra focus to data services and insights (+3.90%). Meanwhile, the data-centric end-to-end view, the connective tissue across the life cycle, was underaddressed relative to its priority (−24.40%), and vendor partnerships received less emphasis than their expected share (−7.40%). The conversation favored pragmatic enablers and standards, leaving the integrative, end-to-end orchestration agenda with less coverage.

Theme 3 leaned into strategy, return on investment, and value generation beyond expectations (+12.33%), and data readiness and governance were also covered slightly more than expected (+4.77%). Collaboration and benchmarking landed near alignment (−1.40%), while organization and culture trailed expectations (−5.67%), and ethics and compliance were notably underaddressed relative to their perceived importance (−10.13%). The emphasis pattern suggests a pull toward measurable value and readiness, with ethical and compliance considerations not receiving proportionate conversation time.

Cross-phase convergence was assessed by comparing the workshop 1 categories, survey rankings, and workshop 2 discussion themes. The resulting persistent and evolving categories are described below.

The strong cross-phase convergence observed in this study aligns with prior work demonstrating that digital transformation in life sciences increasingly depends on patient-centered design, interoperable data, and cross-sector collaboration rather than technology adoption alone [15,18,46]. Comparable studies have similarly shown that sustainable digital transformation requires governance, standards, workforce capability, and organizational readiness to evolve alongside technical innovation. Recent work examining digital transformation in drug development has highlighted that implementation success is strongly influenced by organizational alignment, governance structures, and change management rather than technological capability alone [41,47].

The emphasis on access, equity, and reduction of patient burden is also consistent with emerging evidence demonstrating that digital technologies improve participation, inclusivity, and trial accessibility only when they are accompanied by intentional design, community engagement, and consideration of health equity [48-50]. These findings suggest that stakeholders increasingly view digital transformation not solely as a mechanism for operational efficiency but also as an opportunity to improve patient experience, representation, and access to innovation.

Collectively, these observations reinforce the growing consensus that digital transformation should be understood as a multidimensional change process spanning technology, governance, culture, workforce capability, and patient engagement.

An important observation was the evolution of discussions across study phases. Workshop 1 focused primarily on opportunities, desired outcomes, and future possibilities, whereas workshop 2 concentrated on implementation barriers, competing priorities, governance challenges, and organizational readiness. This progression suggests that the pharmaceutical industry may be entering a more mature phase of digital transformation in which the challenge is no longer establishing the need for change but determining how transformation can be implemented successfully and sustained over time.

This interpretation is consistent with the implementation science literature, which has increasingly demonstrated that transformation initiatives often encounter barriers during adoption, integration, scale-up, and sustainability rather than during initial innovation [41]. Similar observations have been reported in digital health implementation research and sociotechnical transformation frameworks, including the NASSS framework, which highlights how organizational complexity, governance arrangements, stakeholder interactions, and contextual factors frequently determine whether innovations are successfully embedded into routine practice [41].

The findings, therefore, suggest that pharmaceutical digital transformation should be viewed primarily as a sociotechnical challenge rather than a purely technological one. Stakeholders consistently emphasized trust, collaboration, incentives, workforce readiness, regulatory alignment, and organizational capability alongside technological innovation. These findings imply that future value realization may depend less on acquiring new technologies and more on developing the organizational conditions necessary to implement them effectively [46].

Unlike studies focused on specific digital modalities and application areas such as DCTs, AI, digital end points, digital therapeutics, or real-world evidence generation [12-14,51,52], this study provides a cross-stakeholder, end-to-end view of digital transformation priorities spanning outcomes, pathways, and measurement domains across the pharmaceutical product development life cycle.

A key contribution is the identification of a practical implementation-tension framework comprising 5 recurring tensions: innovation vs safety, collaboration vs competition, technology vs change management, patient preferences vs efficiency, and data democratization vs privacy. While aspects of these tensions have been discussed separately within the digital health and implementation literature [41], they have rarely been synthesized into a unified framework derived from multiple stakeholder groups across the pharmaceutical ecosystem.

The framework advances understanding by demonstrating that barriers to digital transformation are often systemic rather than technical. The findings suggest that progress depends not only on technology development but also on governance structures, incentive alignment, organizational readiness, leadership commitment, workforce capability, and effective change management. In this sense, the study extends existing literature by providing an implementation-focused perspective that complements technology-focused digital transformation research [41].

For industry, the findings reinforce the importance of organizing digital transformation around a small number of durable priorities rather than fragmented technology initiatives [46,53]. The convergence on patient-centered outcomes, FAIR-aligned data, and collaboration suggests that these areas may provide strategic anchors for future transformation efforts.

For regulators, the findings echo growing calls for clearer expectations regarding digital evidence generation, interoperability, data quality, and governance to reduce uncertainty and support responsible innovation [54]. Regulatory clarity may play an increasingly important role in enabling adoption while maintaining appropriate standards for safety, quality, and trust [50,54].

For technology providers, the findings highlight the importance of developing interoperable, low-burden solutions that align with FAIR principles, support integration into existing workflows, and address implementation challenges alongside technical functionality [16,49]. Providers that support organizational adoption and change management may be better positioned to contribute to sustainable transformation than those focused solely on technology deployment [55].