actor-1-clinicaltrials-protocol-diff

Apify actor that monitors ClinicalTrials.gov protocol-change events (status updates, enrollment milestones, primary-endpoint edits, sponsor changes) and emits per-NCT-ID JSONL diff records.

Who this is for

• Pharma competitive-intel analysts at small BI shops (buy substitute: Evaluate Pharma $35k/yr, Citeline $50k+/yr, Informa BDCI).
• AI-agent builders embedding drug-discovery workflows who need structured protocol-change streams.
• Medical journalists at STAT / Endpoints / FiercePharma tracking trial-design changes.

What it does

1. Accepts a watchlist config (sponsor / indication / phase / NCT ID list).
2. Fetches current studies via ClinicalTrials.gov REST API v2.
3. Diffs against the previous snapshot in Apify Dataset (.actor/input_schema.json picks the key-value store).
4. Emits one dataset record per detected change with:
   • nct_id
   • change_type (status / enrollment / endpoint / sponsor / eligibility / locations)
   • before / after values
   • detected_at timestamp
   • standard BaseRecord fields (actor, schema_version, extracted_at, source_url)

Pricing (Apify Store)

Pay-per-result (PPR):

• $0.02 per new diff record emitted.
• Typical buyer: watchlist of ~500 studies, weekly cron → ~50-200 changes/week → ~$1-4/week Apify charge → $4-16/month.
• Mid-market buyer: watchlist of 5000 studies, daily cron → ~100-500 changes/day → $2-10/day → $60-300/month.

Inputs (see .actor/input_schema.json)

Field	Type	Default	Description
watchlist	array of NCT IDs	[]	Explicit study list
sponsors	array of strings	[]	Filter by sponsor name (partial match)
conditions	array of strings	[]	MeSH condition terms
phases	array of strings	[]	e.g. ["Phase 2", "Phase 3"]
statuses	array of strings	[]	e.g. ["Recruiting", "Active, not recruiting"]
change_types	array of strings	all six	Which diff types to emit
snapshot_key	string	auto	Key-value store key for previous snapshot
max_studies	int	10000	Hard cap for cost ceiling

Output

JSONL dataset; one record per detected change. Example::

{
  "actor": "actor-1-clinicaltrials-protocol-diff",
  "schema_version": "1.0.0",
  "extracted_at": "2026-05-11T18:00:00Z",
  "source_url": "https://clinicaltrials.gov/study/NCT01234567",
  "nct_id": "NCT01234567",
  "change_type": "status",
  "before": {"overall_status": "Not yet recruiting"},
  "after":  {"overall_status": "Recruiting"},
  "detected_at": "2026-05-11T18:00:00Z"
}

Running locally

# from repo root after `pip install -e .[dev]`:
cd actors/actor-1-clinicaltrials-protocol-diff
apify run --purge  # uses .actor/input_schema.json defaults

For a targeted run, pass explicit input::

apify run --input '{"watchlist": ["NCT01234567"]}'

Known gotchas

• CT.gov v2 API returns null for many optional fields where v1 returned empty strings. extract.py normalises both to None.
• conditions input uses partial-match on normalized MeSH term — "Alzheimer" will match "Alzheimer Disease" and "Alzheimer's Disease". Buyers sometimes assume exact-match; README + listing must be explicit.
• Studies in "Withdrawn" status are frequently re-activated. Treat status transitions as bidirectional, NOT a one-way flow.

Store listing (draft — edit before publish)

Title

ClinicalTrials.gov Protocol Change Monitor

Tagline

Daily diff of NIH-registered trial protocols. Drop Evaluate Pharma and Citeline — get the same change stream for 1/100th the cost.

Features

• ✅ Full ClinicalTrials.gov REST API v2 coverage (500k+ studies)
• ✅ Six change types (status, enrollment, endpoint, sponsor, eligibility, locations)
• ✅ Watchlist OR sponsor OR condition OR phase filters
• ✅ JSONL dataset output, ready for your BI pipeline
• ✅ Incremental: first run snapshots, later runs diff

Example use cases

1. Competitive intel: Track all Phase 2/3 trials by Roche / AstraZeneca / Novartis for oncology. Alert on enrollment changes.
2. Indication tracking: Monitor every NCT study in Alzheimer's disease — know when any trial changes primary endpoints.
3. Drug-discovery AI agent: Feed change events into your agent's context as a retrieval-augmented fact stream.

Legal

ClinicalTrials.gov data is US government public-domain (42 USC §282(j)). Actor respects CT.gov rate limits (2 rps default). No PII in output.