WF-15 — Research Concept, Ethics Review and Project Registration

This workflow's real function is the launch-year register. KPIs 4.3, 4.5 and 5.4 are cumulative over five years and each project or impact unit is counted once, in the year it is launched, and never again. Without a launch-year register MBRU will either double count, which is an audit finding, or under count, which is lost points, and will not be able to tell which of the two it has done. Registering a project at conception, with its launch year, its partner attributes and its Linked Programs value, is the control that prevents both.

1. Purpose

To ensure that every scholarly activity IoL undertakes is registered at conception with a stable identifier, a launch year and the partner, funding and programme attributes that OBEF requires, that it has passed scientific and ethics review before any data is collected, and that the people involved have settled authorship, data management and conflict of interest in advance, so that the project can be counted exactly once across a five-year window and can be evidenced without anybody having to remember what happened.

2. Scope

Scope statement. This process manages an IoL research or scholarly project from the point at which a concept is proposed, through scientific and ethics review and registration, to the point at which the project record is complete and released to the portfolio for execution and reporting.

Applies to. All scholarly activity led by or hosted within IoL, whether funded or unfunded, whether staff-led or student-led, curricular or extra-curricular, including educational research, health professions education scholarship, curriculum evaluation studies, and creative or developmental scholarly work of the kind the Frascati definition admits in disciplines where traditional research is not the primary form of scholarship. It applies to projects with no external partner as much as to those with one, because the denominator of KPIs 4.3 and 5.4 is all research projects.

Does not apply to. Routine programme monitoring, internal quality assurance activity and service evaluation that does not meet the Frascati criteria; these are WF-04. It does not manage the conduct of the research itself, publication (WF-18), impact capture (WF-19), IP disclosure (WF-20), or the negotiation of partnership agreements (WF-24). It does not replace the MBRU institutional research ethics process; it is the IoL-side workflow that feeds into it and records its outcome. [IoL to confirm] the name, standing and turnaround of the MBRU committee that grants ethics approval, and whether IoL has any delegated authority for minimal-risk educational research.

Applicable requirements. OBEF guide v11.5 KPI 4.3, 4.4, 4.5 and 5.4 counting rules; the Frascati Manual (OECD 2015) five-criteria definition of research; CHEDS/HEDB Institute - Research Projects.xlsx; MBRU research ethics and research integrity policy; UAE personal data protection obligations for research data; funder and partner terms where these exist.

What counts as a research project, quoted. Any scholarly activity meeting the five Frascati criteria: novel, creative, uncertain in outcome, systematic, and transferable or reproducible. All five, not a majority. Creative activities count in disciplines where traditional research is not the primary form of scholarly activity, which is directly relevant to health professions education, where curriculum development, assessment instrument design and educational innovation can meet all five criteria and are routinely not registered because they do not look like a laboratory study.

Why the denominator matters as much as the numerator. KPIs 4.3 and 5.4 both divide qualifying collaborations by total research projects. Every project registered here without an industry or international partner lowers those two ratios. That is not a reason to under-register: an incomplete denominator is a misstatement, and cross-KPI assertion C2 requires the 4.3 and 5.4 denominators to be identical. It is a reason to apply the Frascati test honestly in both directions, and to record the test that was applied.

3. Trigger, boundary and endpoint

4. SIPOC

5. Accountability

Process owner. IoL Director of Research, or the role able to require registration before data collection and to resolve disputes about whether an activity qualifies.

Escalation.

6. Process steps

1. Submit the concept note. A short structured note: question, rationale, design, population, partners contemplated, resources, and the programmes the work relates to. One page is enough. The purpose of this step is to catch the project at conception, because everything this workflow captures becomes harder to establish with every month that passes. [CAPTURE] proposer, date of concept, working title. Serves the launch-year determination.

2. Screen against the Frascati criteria. Apply all five: novel, creative, uncertain in outcome, systematic, transferable or reproducible. Record the screening outcome and its reasoning, not just the verdict. [CAPTURE] Frascati screening decision with reasoning and date. Serves the 4.3 and 5.4 denominator and the 4.4 qualification. Decision point. An activity failing the screen is not registered as a research project and is routed to WF-04 as programme monitoring or service evaluation. An activity passing it must be registered even if it has no partner, no funding and no publication ambition, because it belongs in the denominator. Note for health professions education specifically. Curriculum innovation, assessment instrument development and educational technology evaluation frequently satisfy all five criteria. The Frascati definition's explicit accommodation of creative activity exists for exactly this case. Screen these in on their merits rather than out by habit.

3. Scientific review. A proportionate internal review of design, feasibility and methodological soundness before the project consumes resource or reaches the ethics committee. Record reviewers, comments and the response. [CAPTURE] review record and date. [IoL to confirm] whether IoL operates its own scientific review panel or relies on the MBRU research office, and what the standing turnaround is.

4. Ethics review and decision. Submit to the MBRU research ethics committee, or record the formal exemption decision where the activity qualifies for one. No data collection begins before this step closes. [CAPTURE] ethics reference number, decision, decision date, expiry or review date, conditions attached. Exception route. Where a study is exempt or falls below the threshold for full review, the exemption decision itself is the record, and it is recorded with the same discipline as an approval. An undocumented assumption of exemption is the commonest research integrity finding in educational research and is entirely avoidable.

5. Data management plan. What data will be collected, in what form, where it will live, who has access, how it is backed up, how it is shared with partners, what happens at the end of the project, and who holds it after the investigators have moved on. [CAPTURE] the plan, its version and its owner. This is workflow content in its own right, not an OBEF artefact, and it is the thing most often skipped in small educational studies where the data feels informal until somebody leaves.

6. Privacy assessment and retention schedule. Where the study involves personal data of students, patients, staff or partners, record the lawful basis, the consent mechanism, the minimisation applied, the retention period and the disposal method. [CAPTURE] privacy assessment reference and retention period. Note the interaction with Appendix B. OBEF evidence provisions permit the Ministry to demand contact data for students counted under KPI 4.4 and for research partners under 4.3 and 5.4. Retention schedules set here must not dispose of data the institution may be required to produce, and consent language must not preclude that disclosure. Settle both at registration.

7. Authorship agreement. Record the expected author list, the order principle, the criteria to be applied, and the process for resolving change. [CAPTURE] authorship agreement, dated, with the parties recorded. Doing this at conception rather than at submission is the single cheapest dispute-prevention measure available to a research group, and it costs one paragraph.

8. Conflict of interest declarations. Collected from every investigator, including students, and from partner-affiliated contributors. [CAPTURE] declarations with dates, and any management plan agreed.

9. Set the partner attributes, which cannot be reconstructed later. For every partner contemplated: identity, type from the approved vocabulary, whether the partner is an industry partner, an international partner, or both, and the reference to the agreement or MoU that evidences the partner organisation's awareness of the collaboration. [CAPTURE] Industrial partner name, Partner Type, International partner name, International Partner Type, Agreement/MOU Number. Serves 4.3 and 5.4 through WF-17. The awareness test, stated plainly. Working with an individual who happens to be affiliated to an organisation, without that organisation's demonstrable knowledge or approval, does not qualify as a partnership for either KPI. The organisation must be aware of the collaboration. An agreement, an MoU, a countersigned letter or a documented institutional approval is what demonstrates it. A colleague's goodwill is not.

10. Set the funding attributes. Total project budget, funding entity, funding source, and the partner's share of funding where a partner contributes. [CAPTURE] Funding Amount, Funding Entity, Funding Source, Partner_funding, International_Partner_funding. Serves the 4.3 and 5.4 financial route thresholds, and feeds 3.4 through WF-17 and WF-24. Record the budget even where it is zero or notional. The 4.3 financial route requires a minimum project budget of AED 50,000 and a partner share of at least 20%, and a blank budget field cannot be tested against either.

11. Assign the Project ID and fix the launch year. One identifier per project, issued once, never reused, never revised. The launch year is the year the project launched and is fixed at registration. [CAPTURE] Project ID, launch year, Fiscal Year, Start Date, planned End Date, Status, Type of Scholarly Activity, Field of R&D, Keywords, Principal Investigator, ORC ID. Serves 4.3, 4.4, 4.5, 5.4 and cross-KPI assertion C6. This is the step the whole workflow exists for. Everything downstream refers to this identifier and this year. A project whose launch year is decided at submission time, three years later, from a start date that was never recorded, is a project that will be counted twice or not at all.

12. Set Linked Programs. Record every IoL programme the project relates to, using Program Codes. A project linked to several programmes is counted for all of them at programme level, expanded rather than split. [CAPTURE] Linked Programs. Serves programme-level 4.3, 4.4 and 5.4. An unlinked project scores at institution level only. For a channel C4 workflow feeding two programme rows, that is a straightforward and recoverable loss of points, and the field takes ten seconds.

13. Register and release. Write the row to Institute - Research Projects.xlsx and the entry to the Launch-Year Register, then release the project to the portfolio. [CAPTURE] registration date and the completeness check result. Completeness check before release: Frascati screening recorded; ethics decision or exemption recorded; data management plan lodged; authorship agreement dated; conflicts declared; partner type and awareness evidence set where a partner exists; funding amount present; Project ID unique; launch year fixed; Linked Programs populated. A project failing the check is not released.

14. Maintain the register through the life of the project. Status and End Date change; Project ID and launch year do not. Amendments to ethics approval, changes of principal investigator and the addition of a partner after launch are recorded as amendments against the same record. Decision point. Where a partner joins after launch, the collaboration is still counted in the project's launch year, not in the year the partner joined, because the counting unit is the project. Record the date the partner joined so the decision is visible and reproducible, and refer the treatment to WF-17, which owns the 4.3 and 5.4 numerators.

7. OBEF data generated

Capture rule. Six attributes cannot be reconstructed after the fact and must be recorded at registration. The launch year, because a start date inferred later from a first publication or a first data file is a guess, and the whole five-year counting rule rests on it. The partner's identity and type, because a health-system collaborator recorded as a free-text name three years on cannot be classified against the approved partner-type vocabulary. Whether the partner organisation was aware of the collaboration, because awareness is evidenced by a document that exists at the time or does not exist at all. The funding amount and the partner's share, because the 20% and AED 50,000 tests are arithmetic on numbers nobody records retrospectively. Linked Programs, because the person who knows which programme the work serves is the investigator, at the moment of conception. The Frascati screening reasoning, because a denominator that cannot be justified is a denominator that will be challenged.

Reproducibility test. With this register in place, a second analyst can construct the 4.3 and 5.4 denominators, apply both qualification routes, and produce identical numerators, working from records alone. Without it, they cannot produce any of them, and the failure is silent: an incomplete project list looks exactly like a small research portfolio. That silence is the specific reason this workflow is worth building before anybody needs its output.

8. Max-score design

This workflow produces no KPI of its own. Its targets are therefore expressed as the conditions the other four KPIs need, and the honest way to read this table is that the register does not score points, it makes points scoreable.

What this workflow must do to reach the target.

1. Register every qualifying project, including the unfunded and unpartnered ones. The temptation runs the other way, because each unpartnered project lowers the 4.3 and 5.4 ratios. Resist it. An understated denominator is a misstatement, it fails assertion C2 the moment the two KPIs are compared, and it will not survive an evidence request that asks for the full project list.
2. Fix the launch year at registration and never revise it. This is the single control that satisfies cross-KPI assertion C6, that every project counted in 4.3, 4.5 or 5.4 exists once in the launch-year register.
3. Record partner type against the approved vocabulary at the moment the partner is named, not as free text to be classified later. WF-17 depends on this and cannot recover it.
4. Record the awareness evidence reference, or record that there is none. A blank field is ambiguous; an explicit "none" is a decision, and it stops a project being claimed on an assumption three years later.
5. Populate Linked Programs on every project. Two Program Codes, ten seconds, and the difference between an institution-only claim and a programme-level one on three KPIs.
6. Screen honestly in both directions. The Frascati criteria admit educational and creative scholarship that IoL habitually does not think of as research, and they exclude routine evaluation that IoL might be tempted to count.

[REDESIGN] actions.

Sequencing note. R3 is the action with a deadline attached that nobody has set. The 4.3, 4.5 and 5.4 windows are five years cumulative, so every year that passes without a backfill permanently removes a year of countable projects from the first submission that uses the register. The evidence needed for a backfill also decays: ethics approvals, agreements and award letters from four years ago are findable now and will be harder each year. R1 and R2 protect the future; R3 is the only action that recovers the past.

9. Indirect strategy where data cannot be collected

Category 4, the activity exists but is invisible, and this workflow is the archetype of it.

IoL undertakes scholarly activity. Curriculum evaluation, assessment design research, educational technology studies, health professions education scholarship with health-system colleagues, supervision of student projects: it happens, and much of it meets the Frascati criteria on any honest reading. None of it counts towards KPI 4.3, 4.5 or 5.4 unless it exists as a registered project with a launch year, a partner type and a programme linkage. The response to Category 4 is capture, not tactics. There is no argument to be constructed here and no interpretive latitude to exploit. There is a register, and either it exists or the points do not.

Where a genuine interpretive judgement does arise, it is the Frascati screen, and it runs both ways. The five criteria are a definition, not a hurdle, and the guide's explicit statement that creative activities count in disciplines where traditional research is not the primary form of scholarly activity is a deliberate accommodation for fields exactly like health professions education. Applying it is legitimate. Stretching it to cover routine curriculum administration is not. The safeguard is the recorded reasoning at step 2: a screening decision with its rationale can be reviewed and, if wrong, corrected. A screening decision with no rationale is indistinguishable from a preference.

A second-order point about the denominator that is worth stating explicitly. Because 4.3 and 5.4 divide by total research projects, there is a structural incentive to register fewer projects. Anyone reading the formulae will notice it. It is worth naming inside the workflow rather than leaving it to be discovered privately, because an unnamed incentive is the one that operates. Three things close it off: the denominator is evidenced and can be requested under Appendix B; assertion C2 compares the two denominators against each other; and an understated project count also understates KPI 4.1's activity base and IoL's own account of what it does. The correct response to a low ratio is to register more collaborations, which is WF-17's job, not to register fewer projects.

Ambiguities in the guide, stated rather than resolved.

1. What "launch year" means for a project with a long gestation. The counting rule says a collaboration is counted once in the year it launched, but "launched" is not defined against a specific event: concept, ethics approval, funding award, first data collection and formal start date can span two or three years. Choose one definition, record it in the register's own documentation and in the Mechanism of collecting the data field, and hold it constant. IoL's working assumption, stated as an assumption: the launch year is the year of the recorded Start Date, which is the date from which work was authorised to proceed. Confirm with MoHESR.
2. How a partner joining after launch is treated. The project is counted once in its launch year, so a partner joining in year three either qualifies the project retrospectively in its launch year or does not qualify it at all. The guide does not say. The register records the date the partner joined so the choice is visible; the choice itself belongs to WF-17 and to a written question to MoHESR.
3. Whether a project with no external partner belongs in the denominator. On the plain reading of "total research projects", yes. Nothing in the guide restricts the denominator to partnered projects. It is worth noting because the alternative reading would be materially more favourable, which is exactly why it should not be adopted without an answer.
4. The status of creative and developmental educational work. The Frascati accommodation is real but its boundary is not drawn. A new assessment instrument developed systematically, evaluated and published is plainly in. A revised course handbook is plainly out. The space between them is where the screening reasoning earns its keep.

Boundary check. This workflow must never:

• register an activity as a research project where the Frascati screen was not applied, or was applied and failed;
• omit a qualifying project from the register because it has no partner and would lower the 4.3 or 5.4 ratio;
• assign or revise a launch year retrospectively to move a project into a more advantageous year of the five-year window;
• record a partner organisation as aware of a collaboration without an agreement, MoU or documented institutional approval to evidence it;
• record a partner type that the approved vocabulary does not support, or classify an organisation by what would be convenient rather than what it is;
• allow data collection to begin before the ethics decision or documented exemption is recorded;
• backfill a launch year from recollection rather than from a contemporaneous document, or fail to flag a backfilled row as backfilled;
• create a second record for a project that already exists in order to resolve a reporting inconvenience.

10. Service standards

11. Records and evidence

Appendix B readiness. The Ministry can demand partnership documents, proof of financial and intellectual contribution, and contact data for industry and international partners, and it can do so without notice. Today IoL would be assembling that from individual investigators' files and memories, which is neither fast nor complete. With the register operating, the answer to every such request is a filtered extract plus the referenced documents, and 15 working days is comfortable rather than tight. [IoL to confirm] where research agreements are currently held and whether a single index of them exists.

12. Risks and controls

13. Performance measures

The first five measures are the workflow. Everything else in this document is scaffolding around them, and a research portfolio can be genuinely productive while scoring nothing on three KPIs if those five measures sit at zero.

14. Change control