What AnzarSeha is
AnzarSeha is a web application that accompanies a clinical study from the definition of its question to the export of an analysable dataset. It covers the record registry, the variable dictionary, traced capture, protocol locking, database freeze, the audit log and scientific export.
What characterises it is not that list of functions — other tools offer them — but their constrained sequence: each stage inherits what the previous one froze, and some operations are refused until the previous one is done. The protocol lock, for instance, will not engage until the primary endpoint is declared.
Who it is for
Teams running studies on their own records, most often in a non-regulated setting: investigator-initiated research, thesis work, departmental registries.
- The resident writing a thesis — who must produce work defensible before a jury, usually alone and on a short timeline.
- The investigating clinician — preparing a publication that will require describing the capture method.
- The department head — who wants a reusable base rather than one file per study.
- The methodologist — who would rather receive a dataset than a cleanup job.
AnzarSeha does not target the sponsored clinical trial market, where regulatory requirements, actors and vocabulary are different.
The problem addressed
In investigator-initiated research, data is almost always managed in a spreadsheet. That works until someone asks to verify — and at that point three questions have no answer: where did this value come from, what exactly did this column measure, and was the endpoint chosen before or after seeing the results.
The problem is not dishonesty; it is that a spreadsheet has nowhere to record those answers. The subject is treated in detail in the spreadsheet comparison.
Core capabilities
| Capability | What it does | Detail |
|---|---|---|
| Clinical registry | A single, typed, versioned source | Clinical registry |
| Variable dictionary | Type, scale, UCUM unit, bounds, mappings | Dictionary |
| Traced capture | Every value keeps its origin and justification | Capture |
| Assisted extraction | A model proposes, quoting its source sentence; a human confirms | AI extraction |
| Locking and freeze | Protocol fixed before capture, database frozen before analysis | Data management |
| Audit log | Hash-chained, ALCOA+ by construction | Audit trail |
| Scientific export | Six sheets, readable by SPSS, R and pandas | Dataset |
Who publishes AnzarSeha
The site and the application are published by Dr Rida Akodad, a natural person, as part of a clinical research project, in Morocco. The publication director can be reached at [email protected]. Full details are in the legal notice.
HUIM6 Rabat — Mohammed VI International University Hospital — is the pilot institution and trusted partner of the project. This is a stated partnership, not an institutional affiliation: AnzarSeha remains the publisher of its own tool, and that distinction is maintained everywhere the partnership is mentioned.
What AnzarSeha does not claim
This section is deliberately explicit, because the field lends itself to unverifiable assertions.
- No certification is claimed. The conventions followed are ICH E6(R3), CDISC CDASH, SNOMED CT, LOINC, ICD-10 and UCUM. Following a convention is not being certified by its body.
- The tool does not make a study compliant. Compliance rests with the data controller and the competent committee — GDPR, and in Morocco law 09-08 / CNDP.
- The AI does not decide. It proposes a quoted value, which a human confirms. No model-origin value becomes final without validation.
- No clinical result is displayed. Every screenshot on the site comes from a demonstration account on fictional data; no figure shown is a result.
- Formal quality is not clinical accuracy. An erroneous source record yields a value that is traced, and wrong.
- The pathology catalogue is not exhaustive. The demonstration is a urology department; an uncovered domain requires building the form variable by variable.
Security and data
- Hosting on infrastructure located in the European Union.
- Every access is authenticated; each study is isolated by row-level security, closed by default.
- Direct identifiers are stripped from text before any call to an external service — see pseudonymisation.
- No third-party script loads on the public site: the content security policy forbids it.
- A full export, in open formats, is available at any time.
Frequent questions
Is AnzarSeha an eCRF or an EDC?
It covers those functions for investigator-initiated research — typed form, audit trail, locking, export — without positioning itself in the sponsored-trial market, whose software validation and regulatory submission requirements are different. See the electronic case report form.
Is real data needed to evaluate the tool?
No. The demonstration runs on sixty fictional records, with a study workbook already built and an export ready to download. No real patient data is required, and none should be entered.
What happens if we stop using AnzarSeha?
The full workbook — data, dictionary, summary, metadata, missing-value table and decisions — exports at any time to Excel, CSV, PDF or Word, and opens in SPSS, R and pandas with no manual editing. Vendor dependence is real, and this is what limits it.
How should AnzarSeha be cited in a publication?
Mention the tool and its version in the methods section, with the freeze date used for the analysis — for example: "data were collected and frozen using AnzarSeha (freeze of DD/MM/YYYY)". What matters to a reviewer is the freeze date and the description of capture, more than the name of the tool.