Screening protocol ZIIX-PV-LS-001

Version 1.0 — published 2026-10-31. This document is the protocol referenced by the sample screening record and rendered on the offer page. It is the rule set that was executed, verbatim, against the public literature.

---

1. Scope

A single medicinal product is screened against a defined, dated literature corpus. The run below is a demonstration over public literature, not client data.

Field	Value
Protocol id	ZIIX-PV-LS-001
Version	1.0
Medicine	daridorexant (INN)
Source	NCBI PubMed / MEDLINE via E-utilities (esearch.fcgi + efetch.fcgi)
Window	2023/11/01 → 2026/10/31 (36 months ending on the retrieval date)
Sorting	date (most recent first)
Retrieval cap	300 records

2. Search string, as executed

("daridorexant"[Title/Abstract]) AND (adverse[tiab] OR safety[tiab] OR "side effect"[tiab] OR "adverse event"[tiab]) AND ("2023/11/01"[Date - Publication] : "2026/10/31"[Date - Publication])

The exact string, the exact request URL and the raw API responses are retained with the record. A third party can re-run the search and reproduce the corpus.

3. Inclusion criteria

A record is INCLUDED (and flagged as a case candidate) only when all three hold:

• I1 — HUMAN. The record reports exposure in humans: a clinical trial, an

observational study, a case report or case series, real-world evidence, or a spontaneous report.

• I2 — IMPLICATED. The medicine is implicated in a reported adverse event,

tolerability finding or safety outcome. Attribution is assessed at sentence level: at least one sentence must contain both the medicine name and adverse-event vocabulary.

• I3 — RETRIEVABLE. Full citation data was retrieved (PMID, title, journal, year,

authors, abstract) so the decision is auditable by a third party.

4. Exclusion criteria

Exactly one primary reason code is recorded per excluded record. The codes are applied in a fixed order; the first that matches is the recorded reason.

Code	Name	Rule
X1	NOT_HUMAN	No human exposure. In vitro, animal or other non-clinical evidence only.
X2	NO_NEW_CASES	A review, editorial, comment, guideline or meta-analysis carrying no new case data.
X3	NO_SAFETY_DATA	The record reports no adverse event, tolerability or safety outcome.
X4	DRUG_NOT_IMPLICATED	The medicine is mentioned, but no reported adverse event is attributed to it.
X5	RECORD_INCOMPLETE	Citation data insufficient to screen. Routed to manual review — never silently dropped.
X6	STUDY_PROTOCOL	A study protocol or design paper reporting no results. It can contain no case.

5. Decision rule order

X1 → X2 → X6 → X5 → X3 → X4 → INCLUDE

The order is fixed because the classes are not disjoint:

• X1 before X2 — a non-clinical review is classified X1, not X2. Whether a

record is human evidence is settled before whether it carries new cases.

• X2 before X6 before X5 — publication type and study design are metadata,

determinable without the abstract. They are settled before the content rules.

• X5 before X3 and X4 — we cannot assert "no safety data" about a record we could

not read. An unreadable record is routed to manual review, not decided by absence of evidence.

• X3 before X4 — "no safety data" is the more fundamental statement about a

record than "not implicated".

The order is enforced by an automated check that compares the published order against the order the code actually applies, so the two cannot drift apart silently.

6. Completeness rule

Every retrieved record receives a decision and a reason. The run asserts:

included + excluded == retrieved

If this fails, the run aborts and no record is emitted. There is no path that drops a record silently.

7. Dual screening

The run published here is a single-pass demonstration. Production delivery applies dual independent screening with adjudication of disagreements, per the client's instructions.

8. Boundaries of this protocol

This protocol produces a screening decision and a flag. It does not produce, and must not be represented as producing:

• a causality assessment
• an expectedness assessment
• a seriousness assessment
• a regulatory submission or a safety report
• a replacement for the MAH's own literature-search record

Those determinations belong to the marketing-authorisation holder's qualified person.