Global Health Intelligence MCP Server

Global Health Intelligence MCP is agent-native applied health intelligence infrastructure for AI agents, pharma intelligence teams, and global health analysts.

It wraps five free authoritative health data sources (ClinicalTrials.gov, PubMed, WHO Global Health Observatory, EMA medicines, and Open Food Facts) behind one MCP endpoint and adds a composite drug pipeline research tool that queries trials, literature, and EU approvals in parallel. No API keys required. Built for pharma competitive intelligence, biotech business development, regulatory affairs, clinical research, global health programme planning, and food-safety audits.

The category

Global Health Intelligence MCP is agent-native applied health intelligence infrastructure. Unlike $40,000 to $100,000 per seat per year platforms like Cortellis, Citeline Pharma Intelligence, Evaluate Pharma, or IQVIA, which are built for human analysts juggling dashboards, this MCP exposes the same underlying public health data (trials + EU approvals + global indicators + literature + food safety) as MCP tools an AI agent can call directly. It is the drug-pipeline + regulatory + population-health side of the fleet, not academic literature search (see Academic Research Intelligence MCP for that). Pay $0.05 per single-source call or $0.15 per composite drug research call instead of an annual subscription.

In one sentence

Run a single MCP call to research a drug or condition across ClinicalTrials.gov, PubMed, and the EMA medicines register in parallel, with computed pipeline signals (Phase 3 trial counts, EMA authorisations, orphan designations, recruiting trials) returned as structured JSON.

Category: Pharma intelligence MCP. Clinical trial search. Global health data. EU drug approval lookup. Primary use case: Give an AI agent one tool that covers applied health intelligence: drug pipelines, EU regulatory status, global health indicators, and food safety. Can also be used for biotech licensing due diligence, rare-disease landscape mapping, and global immunization tracking.

Also known as: pharma intelligence MCP, clinical trials MCP server, EMA medicines MCP, WHO health indicators MCP, drug pipeline research API, ClinicalTrials.gov agent tool, Cortellis alternative, Evaluate Pharma alternative.

What this actor does

• What it is: A standby-mode MCP server exposing 7 tools that wrap 5 free authoritative health data sources.
• What it checks: Clinical trial registry data, biomedical literature, WHO global health indicators by country, EU-authorised medicines with orphan and biosimilar flags, and crowdsourced food product nutrition.
• What it returns: Structured JSON with trial phase and recruitment breakdowns, PubMed bibliographic records with DOI and PMC links, WHO indicator values per country and year, EMA medicine records with authorisation status and therapeutic area, Open Food Facts product records with Nutri-Score and NOVA, and a composite drug pipeline report with computed signals.
• What it does NOT do: No FDA Orange Book or PMDA approval data, no individual medical advice, no full-text PDF download, no real-time pricing or formulary data, no manuscript-level peer-review verdicts.
• Who it's for: Pharma competitive intelligence analysts, biotech business development and corporate development teams, clinical research coordinators, regulatory affairs teams, global health researchers, food-safety auditors, AI agents needing applied health context.

What you get from one call

health_drug_research fans out to ClinicalTrials.gov, PubMed, and EMA in parallel and returns:

• pipelineSignals[] computed flags such as "1 EMA-authorized medicine(s) found", "47 Phase 3 clinical trial(s)", "12 actively recruiting trial(s)", "3 orphan medicine designation(s)", and "Extensive published literature (25+ articles found)"
• clinicalTrials.byPhase trial count per phase (PHASE1, PHASE2, PHASE3, PHASE4)
• clinicalTrials.recruiting active recruiting count
• clinicalTrials.topSponsors[] the five most active sponsors with trial counts
• literature.recentArticles[] recent PubMed records with PMID, journal, publication date, DOI, and article type
• approvedMedicines.authorized count of currently EMA-authorised medicines for the query
• approvedMedicines.orphan count of orphan-designated medicines
• approvedMedicines.medicines[] medicine records with active substance, status, therapeutic area, orphan and biosimilar flags, authorisation date

What makes this different

• Live trials + EU approvals + literature in one call. health_drug_research queries ClinicalTrials.gov, PubMed, and EMA in parallel. One MCP tool replaces three separate platform tabs and the manual cross-referencing step.
• Pipeline signals computed against real data. Phase 3 counts, active recruiting counts, orphan designations, and EMA authorisations are derived from the actual records returned in the same call, not from a static curated database that may be months out of date.
• Headline aggregates for Nutri-Score, NOVA, orphan, and biosimilar. Food and medicine searches surface population-level breakdowns (Nutri-Score A-E distribution, orphan and biosimilar counts) up front, so an agent does not need a second pass over the raw items to answer "how many of these are orphan drugs".

Before vs after

Without this MCP	With this MCP
Open ClinicalTrials.gov, EMA, PubMed, and WHO GHO in four tabs and cross-reference manually	One tool call, three sources queried in parallel with computed signals
Pay $40,000 to $100,000 per year per seat for Cortellis, Citeline, or Evaluate Pharma	Pay $0.05 per single-source call or $0.15 per composite drug research call
Eyeball trial registry pages to count Phase 3 trials and recruiting status	pipelineSignals[] and byPhase computed automatically
Track EMA orphan and biosimilar designations by reading PDFs	orphanMedicines and biosimilars counts in the response envelope
Look up food nutrition data product by product	byNutriScore distribution across the result set, plus barcode lookup

Architecture

agent prompt
   ↓
MCP /mcp endpoint (StreamableHTTP)
   ↓
7 registered tools (5 single-source + 1 composite + 1 free list)
   ↓
5 sub-actors called via apify-client (memory 256MB, 120s timeout)
   ↓
ClinicalTrials.gov │ PubMed │ WHO GHO │ EMA │ Open Food Facts
   ↓
composite (health_drug_research): Trials + PubMed + EMA in parallel
                                  → pipelineSignals[] computation
                                  → phase/status/sponsor breakdowns
                                  → orphan + biosimilar flag aggregates
   ↓
structured JSON response

WHO and Open Food Facts are standalone single-source tools, not part of the composite. The composite is drug-pipeline focused (trials + approvals + literature), not population-health or nutrition. The MCP runs in Apify Standby mode with a configurable idle-shutdown window (default 300s) so platform compute stops billing when no tools are firing.

Built for

Pharma competitive intelligence analysts tracking competitor drug pipelines; biotech BD and corporate development teams running licensing due diligence; clinical research coordinators surveying the trial landscape for a disease or intervention; regulatory affairs teams monitoring EU authorisation status, orphan designations, and biosimilar approvals; global health researchers querying WHO indicators by country and year; epidemiologists and programme planners tracking immunization, mortality, and SDG health targets; food-safety auditors looking up Nutri-Score, NOVA processing level, allergens, and per-100g nutrition; AI agents embedded in Claude Desktop, Cursor, Windsurf, LangChain, or custom workflows that need an applied health intelligence tool.

This server runs in Standby mode on the Apify platform. It orchestrates ClinicalTrials.gov, PubMed, WHO GHO, EMA medicines, and Open Food Facts through five sibling Apify actors, computes pipeline signals from the trial, literature, and approval data, and returns structured JSON your AI agent reasons over directly.

Questions this MCP answers

• "What's the trial landscape for pembrolizumab? Show Phase 3 counts and top sponsors."
• "Which EU medicines are authorised for rare diseases? Filter for orphan designation."
• "What's the life-expectancy trend for India from 2000 to 2022?"
• "Are there active Phase 3 trials recruiting for Alzheimer's disease?"
• "What's the Nutri-Score of this product (barcode 3017620422003)?"
• "Compare immunization rates across the United States, the United Kingdom, and China."
• "Which sponsors are running the most CAR-T cell therapy trials?"
• "Has the EMA authorised any biosimilars for adalimumab?"
• "What's the published literature on semaglutide weight loss outcomes?"
• "Find food products in the United States with Nutri-Score E for an audit."

For AI agents

• First tool to reach for when a prompt mentions clinical trials, drug pipelines, EU drug approvals, EMA, ClinicalTrials.gov, WHO health indicators, global health data, or food safety.
• Call health_list_sources first (it is free). It enumerates the 7 tools and 5 sources so the agent can decide which single-source tool to call before paying for the composite drug research.
• Use health_drug_research for "research this drug" or "research this condition" prompts. It queries ClinicalTrials.gov + PubMed + EMA in parallel for $0.15. One call replaces three single-source calls plus manual cross-referencing.
• Set is_condition correctly on the composite. false (default) searches trials by intervention (drug name); true searches by condition (disease). The flag changes which ClinicalTrials.gov field the trial search runs against and produces different results for the same query string.
• Branch on pipelineSignals[], not raw counts. The signals collapse the trial, approval, and literature blocks into actionable flags ("4 Phase 3 clinical trial(s)", "1 EMA-authorized medicine(s) found") an agent can route on directly.
• WHO and Open Food Facts are standalone. The composite covers drugs + approvals + literature only. Use health_who_indicators for population health and health_food_safety for nutrition queries.
• PubMed is rate-limited at the source (3 req/sec). Large batches of health_pubmed calls in parallel will queue at the upstream API. Single calls inside the composite are fine.

Use this MCP when an AI agent needs to:

• research a drug pipeline (trials + EU approvals + recent literature)
• assess the EU regulatory status of a medicine or active substance
• count active Phase 3 trials, recruiting trials, or orphan designations for a target
• query WHO indicators by country, year, or indicator code
• look up food product nutrition, Nutri-Score, NOVA, allergens, or ingredients
• map a rare-disease landscape across trials, EU approvals, and literature
• run global health programme planning queries (immunization, mortality, life expectancy)

What data can you access?

Data Point	Source	Example
🧪 Clinical trial records: condition, intervention, phase, status, sponsor, enrollment	ClinicalTrials.gov v2	NCT04612972, KEYNOTE-868, PHASE3, RECRUITING, Merck Sharp & Dohme
📚 Biomedical literature with PMID, DOI, PMC, article type, MeSH	PubMed / NCBI E-utilities	"Pembrolizumab plus chemotherapy in advanced endometrial cancer", NEJM 2023, DOI: 10.1056/NEJMoa2302312
🌍 Global health indicators by country and year	WHO GHO OData	Life expectancy at birth, GBR 2020: 80.4 years
💊 EU-authorised medicines with status, therapeutic area, orphan and biosimilar flags	EMA	KEYTRUDA (pembrolizumab), Authorised, antineoplastic agents, MAH: Merck Sharp & Dohme
🥫 Food product nutrition, Nutri-Score, NOVA, allergens, ingredients per 100g	Open Food Facts	Nutella 350g, Nutri-Score E, NOVA 4, allergens: milk, soya, nuts
🎯 Trial phase breakdown computed across result set	ClinicalTrials.gov composite	byPhase: { PHASE1: 6, PHASE2: 14, PHASE3: 47, PHASE4: 3 }
🚦 Trial recruitment status breakdown	ClinicalTrials.gov composite	byStatus: { RECRUITING: 12, ACTIVE_NOT_RECRUITING: 18, COMPLETED: 31 }
🏥 Top sponsors ranked by trial count	ClinicalTrials.gov composite	topSponsors: Merck (47), AstraZeneca (12), Roche (9)
🟢 Nutri-Score distribution across food query result	Open Food Facts	byNutriScore: { a: 14, b: 8, c: 19, d: 31, e: 12 }
🚩 Orphan and biosimilar designation counts	EMA composite	orphanMedicines: 3, biosimilars: 2

Why use Global Health Intelligence MCP?

Pharma and global health intelligence usually involves either:

• a $40,000 to $100,000 per seat per year subscription to Cortellis, Citeline, Evaluate Pharma, or IQVIA (built for human analysts, hard to wire into agent workflows)
• juggling four to five free public portals (ClinicalTrials.gov, EMA, PubMed, WHO GHO, Open Food Facts) in browser tabs and copy-pasting between them
• building custom integrations against each source's API, normalising the response shapes, and maintaining them as upstream APIs evolve

This MCP turns that into one agent-native surface. A single composite call queries trials, literature, and EU approvals in parallel, computes pipeline signals from the raw records, and returns structured JSON your AI agent acts on directly. Single-source tools stay available when the question is scoped (e.g. "just give me the WHO indicator for life expectancy in Japan").

• Scheduling: run periodic drug-pipeline sweeps on Apify Scheduler; pipe new Phase 3 trial counts or EMA status changes to Slack via webhooks
• API access: trigger searches from Python, JavaScript, or any HTTP client using standard MCP protocol
• Parallel fan-out: composite drug research queries three sources simultaneously, not sequentially
• No API keys: all five data sources are free public health APIs, no credentials to provision
• Integrations: pipe results into Notion, Airtable, Google Sheets, or any webhook-compatible competitive intelligence workflow

Features

Five free authoritative health data sources, one MCP toolset

• ClinicalTrials.gov v2 with filters for condition, intervention, sponsor, location, phase (EARLY_PHASE1 / PHASE1 / PHASE2 / PHASE3 / PHASE4), and recruitment status (RECRUITING / NOT_YET_RECRUITING / ACTIVE_NOT_RECRUITING / COMPLETED). Sort by start date, last update, or enrollment count.
• PubMed / NCBI E-utilities with field-tag syntax ([Title], [MeSH Terms], [Author]), boolean operators, article-type filter (Review, Clinical Trial, RCT, Meta-Analysis, Systematic Review, Case Reports), date range, DOI and PMC links. Rate-limited at the source to 3 req/sec.
• WHO Global Health Observatory with indicator codes (e.g. "WHOSIS_000001") or keyword search ("mortality", "life expectancy", "immunization"), ISO 3-letter country filter, year range. Returns indicator and country lists for cross-country comparison.
• EMA medicines register searchable across name, active substance, INN, therapeutic area, ATC code, and MAH. Category filter (Human / Veterinary), status filter (Authorised / Withdrawn / Refused / Suspended / Revoked / Expired), partial-match therapeutic area. Includes orphan and biosimilar flags.
• Open Food Facts with product name, direct barcode lookup, brand, category, Nutri-Score grade (a / b / c / d / e), and country-of-sale filter. Returns per-100g nutrition, ingredients, allergens, and NOVA processing level.

Composite drug research tool

• health_drug_research queries ClinicalTrials.gov + PubMed + EMA in parallel via Promise.all.
• is_condition flag routes the trial search to condition (disease) or intervention (drug name) so the same composite handles "research pembrolizumab" and "research Alzheimer's disease".
• pipelineSignals[] computed automatically: EMA-authorised count, Phase 3 trial count, recruiting trial count, orphan designation count, extensive-literature flag (25+ articles found).
• Trial response includes byPhase, recruiting, and topSponsors[] (top 5 ranked by trial count).
• EMA response includes authorized and orphan counts plus medicine records with status, therapeutic area, biosimilar flag, and authorisation date.

Operational layer

• Apify Standby mode with configurable idle-shutdown (default 300s, env var STANDBY_IDLE_TIMEOUT_SECS, minimum 60s).
• Failure-webhook registration on every container start; customer-side failures push to the operator's webhook handler automatically.
• Per-sub-actor memory: 256 and waitSecs: 120 so slow sources degrade the response instead of blocking it. Empty arrays come back for failed sources, the composite still returns with available data.

Quickstart workflows

Drug pipeline competitive scan

competitor's lead asset name (e.g. "datopotamab deruxtecan")
 → health_drug_research with is_condition: false
 → read pipelineSignals[], topSponsors[], byPhase
 → flag if "X Phase 3 clinical trial(s)" or "X EMA-authorized medicine(s) found"
 → export trials and approvedMedicines to competitive intelligence tracker

Rare-disease landscape mapping

indication (e.g. "primary biliary cholangitis")
 → health_drug_research with is_condition: true
 → check approvedMedicines.orphan count for existing orphan designations
 → check clinicalTrials.recruiting for active development
 → cross-reference with literature.recentArticles for clinical context

Global immunization tracking

indicator and countries (e.g. "DTP3 coverage" across USA, IND, NGA)
 → health_who_indicators with indicator keyword and country (run once per country)
 → compare values across countries.indicators returned
 → schedule weekly to detect coverage drops

Use cases for applied health intelligence

Drug pipeline competitive analysis

Pharma competitive intelligence analysts tracking a competitor's lead asset need to know in one place: how many Phase 3 trials are running, who the top sponsors are, whether the EMA has already authorised the drug, whether orphan designations exist, and what the recent literature says. A single health_drug_research call returns the trial phase breakdown, the top 5 sponsors with trial counts, the EMA authorisation status with active substance and therapeutic area, and the most recent PubMed articles with DOIs. Replaces a manual workflow of opening ClinicalTrials.gov, the EMA register, and PubMed in three tabs and cross-referencing the IDs.

Biotech licensing and BD due diligence

Business development teams running licensing due diligence on a target asset need rapid pipeline signals: is the drug already EU-authorised, what trial phase is it in, who is running it, and is there a strong literature base. The pipelineSignals[] block surfaces these as actionable flags so a BD analyst can triage assets without manually reading 30 trial records and 50 PubMed abstracts. Pair with the Company Deep Research actor for the sponsor-company financials side.

Clinical research feasibility and landscape

Clinical research coordinators planning a new trial need to assess what is already in development for a condition: how many trials are recruiting, how many are in Phase 3, which sponsors dominate, and what comparator arms are typical. health_clinical_trials with status_filter: ["RECRUITING"] and phase_filter: ["PHASE3"] returns exactly the relevant set, with byPhase and byStatus breakdowns computed automatically.

Regulatory affairs and EU approval monitoring

Regulatory affairs teams monitoring EU drug approvals can run health_ema_medicines filtered by therapeutic area or ATC code to track new authorisations, withdrawals, refusals, and suspensions. The orphanMedicines and biosimilars aggregate counts surface portfolio-level signals (e.g. "12 orphan medicines authorised for rare blood disorders") without per-record review. Schedule weekly to detect status changes.

Global health programme planning

Global health researchers, epidemiologists, and programme planners use health_who_indicators to query the WHO GHO for mortality rates, life expectancy, immunization coverage, disease prevalence, and SDG health targets by country and year. Cross-country comparisons (e.g. "DTP3 immunization across USA, IND, NGA from 2010 to 2022") are returned as structured JSON ready to chart or feed into a programme report.

Food safety and product audit

Food safety auditors and consumer-protection analysts use health_food_safety to look up products by barcode, name, brand, or category. The response surfaces Nutri-Score (A-E), NOVA processing level (1-4), allergens, ingredients, and per-100g nutrition. Filter by nutriscore_grade: "e" and country: "United States" to surface the most-processed, lowest-grade products on the US market for an audit shortlist.

Evidence-based medicine literature triage

Clinical researchers and evidence synthesis teams use health_pubmed with article-type filters (Systematic Review, Meta-Analysis, RCT) and date ranges to triage the evidence base for a clinical question. The PubMed field-tag syntax ([MeSH Terms], [Title], [Author]) and boolean operators (AND, OR, NOT) give precision a free-text web search cannot match.

How to connect this global health intelligence MCP

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "global-health-intelligence": {
      "url": "https://global-health-intelligence-mcp.apify.actor/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
      }
    }
  }
}

Cursor, Windsurf, or Cline

Use the same URL and token in your MCP server settings panel. The server communicates via standard MCP protocol over HTTP POST to /mcp.

Python (via requests)

import requests

response = requests.post(
    "https://global-health-intelligence-mcp.apify.actor/mcp",
    headers={
        "Content-Type": "application/json",
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
    },
    json={
        "jsonrpc": "2.0",
        "method": "tools/call",
        "params": {
            "name": "health_drug_research",
            "arguments": {
                "drug_or_condition": "pembrolizumab",
                "is_condition": False
            }
        },
        "id": 1
    }
)
result = response.json()
report = result["result"]["content"][0]["text"]
print(report)

JavaScript

const response = await fetch(
  "https://global-health-intelligence-mcp.apify.actor/mcp",
  {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "Authorization": "Bearer YOUR_APIFY_TOKEN"
    },
    body: JSON.stringify({
      jsonrpc: "2.0",
      method: "tools/call",
      params: {
        name: "health_clinical_trials",
        arguments: {
          condition: "Alzheimer disease",
          status_filter: ["RECRUITING"],
          phase_filter: ["PHASE3"],
          max_results: 50
        }
      },
      id: 1
    })
  }
);
const data = await response.json();
const result = JSON.parse(data.result.content[0].text);
console.log(`Found ${result.total} trials, phases:`, result.byPhase);

cURL

# Composite drug research, runs Trials + PubMed + EMA in parallel
curl -X POST "https://global-health-intelligence-mcp.apify.actor/mcp" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_APIFY_TOKEN" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
      "name": "health_drug_research",
      "arguments": {
        "drug_or_condition": "pembrolizumab",
        "is_condition": false
      }
    },
    "id": 1
  }'

Environment variables

All five data sources are free public health APIs and need no key.

Variable	Required	Purpose
STANDBY_IDLE_TIMEOUT_SECS	Optional	Standby idle-shutdown window in seconds (default 300, minimum 60). The instance exits after this idle period to release platform compute; the next request cold-starts a fresh one.

MCP tools

Tool	PPE event	Price	What it returns
health_clinical_trials	clinical-trials	$0.05	ClinicalTrials.gov v2 search by condition, intervention, sponsor, location, status, and phase. Returns trial records plus computed byPhase and byStatus breakdowns. Max 500 results.
health_pubmed	pubmed-search	$0.05	PubMed biomedical literature search with field tags, boolean operators, article-type filter, and date range. Returns bibliographic records plus uniqueJournals count. Max 200 results. Source rate limit 3 req/sec.
health_who_indicators	who-indicators	$0.05	WHO Global Health Observatory data by indicator code or keyword, ISO 3-letter country, and year range. Returns indicator values plus deduplicated indicators[] and countries[] for cross-country comparison. Max 1000 results.
health_ema_medicines	ema-medicines	$0.05	European Medicines Agency authorised medicines by name, active substance, INN, therapeutic area, ATC code, or MAH. Filters by Human/Veterinary category and authorisation status. Returns medicine records plus orphanMedicines and biosimilars counts. Max 200 results.
health_food_safety	food-safety	$0.05	Open Food Facts product search by name, barcode, brand, or category. Filters by Nutri-Score grade and country of sale. Returns product records (per-100g nutrition, ingredients, allergens, NOVA, Nutri-Score) plus byNutriScore distribution. Max 100 results.
health_drug_research	drug-research	$0.15	Composite drug/condition research. Queries ClinicalTrials.gov + PubMed + EMA in parallel. Returns pipelineSignals[], clinicalTrials block (byPhase, recruiting, topSponsors[], trial records), literature block (recent articles with DOI), and approvedMedicines block (authorised, orphan, medicine records). Set is_condition: true for diseases, false for drugs.
health_list_sources	(none, free)	Free	Enumerates the 7 tools and 5 sources with credential and capability info. No upstream fetch, no charge. Useful for agent planning before paying for a search.

Tool input reference

Tool	Parameter	Type	Required	Description
health_clinical_trials	condition	string	One of four required	Disease or condition (e.g. "lung cancer", "Alzheimer")
health_clinical_trials	intervention	string	One of four	Drug or device name (e.g. "pembrolizumab", "insulin")
health_clinical_trials	sponsor	string	One of four	Sponsor organisation (e.g. "Pfizer", "NIH", "Novartis")
health_clinical_trials	location	string	One of four	Country or city
health_clinical_trials	status_filter	string[]	No	RECRUITING / NOT_YET_RECRUITING / ACTIVE_NOT_RECRUITING / COMPLETED etc.
health_clinical_trials	phase_filter	string[]	No	EARLY_PHASE1 / PHASE1 / PHASE2 / PHASE3 / PHASE4
health_clinical_trials	sort_by	enum	No	StartDate:desc (default), StartDate:asc, LastUpdatePostDate:desc, EnrollmentCount:desc
health_clinical_trials	max_results	number	No	1 to 500, default 25
health_pubmed	query	string	One of three	Search query with optional PubMed syntax ([Title], [MeSH Terms], boolean)
health_pubmed	author	string	One of three	Author name
health_pubmed	journal	string	One of three	Journal name
health_pubmed	date_from / date_to	string	No	YYYY or YYYY/MM/DD
health_pubmed	article_type	enum	No	Review / Clinical Trial / Randomized Controlled Trial / Meta-Analysis / Systematic Review / Case Reports
health_pubmed	sort_by	enum	No	relevance (default) or pub_date
health_pubmed	max_results	number	No	1 to 200, default 25
health_who_indicators	indicator	string	Yes	Indicator code (e.g. "WHOSIS_000001") or keyword (e.g. "mortality", "life expectancy")
health_who_indicators	country	string	No	ISO 3-letter country code (e.g. "USA", "GBR", "CHN")
health_who_indicators	year_from / year_to	number	No	Year range
health_who_indicators	max_results	number	No	1 to 1000, default 100
health_ema_medicines	query	string	Yes	Search across medicine name, active substance, INN, therapeutic area, ATC code, MAH
health_ema_medicines	category	enum	No	All (default) / Human / Veterinary
health_ema_medicines	status	enum	No	All (default) / Authorised / Withdrawn / Refused / Suspended / Revoked / Expired
health_ema_medicines	therapeutic_area	string	No	Therapeutic area, partial match
health_ema_medicines	max_results	number	No	1 to 200, default 25
health_food_safety	query	string	One of four	Product name keyword
health_food_safety	barcode	string	One of four	Product barcode / EAN for direct lookup
health_food_safety	brands	string	One of four	Brand name, partial match
health_food_safety	categories	string	One of four	Category, partial match
health_food_safety	nutriscore_grade	enum	No	all (default) / a / b / c / d / e
health_food_safety	country	string	No	Country where sold (e.g. "France", "United States")
health_food_safety	max_results	number	No	1 to 100, default 25
health_drug_research	drug_or_condition	string	Yes	Drug name, active substance, or disease/condition to research
health_drug_research	is_condition	boolean	No	true = search as condition/disease, false (default) = search as drug/intervention

Output example

health_drug_research for { "drug_or_condition": "pembrolizumab", "is_condition": false }:

{
  "query": "pembrolizumab",
  "searchType": "drug",
  "sources": ["ClinicalTrials.gov", "PubMed", "EMA"],
  "pipelineSignals": [
    "1 EMA-authorized medicine(s) found",
    "47 Phase 3 clinical trial(s)",
    "12 actively recruiting trial(s)",
    "Extensive published literature (25+ articles found)"
  ],
  "clinicalTrials": {
    "total": 25,
    "byPhase": {
      "PHASE1": 1,
      "PHASE2": 6,
      "PHASE3": 47,
      "PHASE4": 3
    },
    "recruiting": 12,
    "topSponsors": [
      { "sponsor": "Merck Sharp & Dohme LLC", "trials": 14 },
      { "sponsor": "National Cancer Institute (NCI)", "trials": 3 },
      { "sponsor": "AstraZeneca", "trials": 2 },
      { "sponsor": "Roche", "trials": 2 },
      { "sponsor": "Memorial Sloan Kettering Cancer Center", "trials": 2 }
    ],
    "trials": [
      {
        "nctId": "NCT04613596",
        "title": "Pembrolizumab Plus Chemotherapy in Advanced Endometrial Cancer (KEYNOTE-868)",
        "status": "ACTIVE_NOT_RECRUITING",
        "phases": ["PHASE3"],
        "sponsor": "Merck Sharp & Dohme LLC",
        "enrollment": 819
      },
      {
        "nctId": "NCT04267848",
        "title": "Pembrolizumab With or Without Chemotherapy in Cervical Cancer (KEYNOTE-826)",
        "status": "ACTIVE_NOT_RECRUITING",
        "phases": ["PHASE3"],
        "sponsor": "Merck Sharp & Dohme LLC",
        "enrollment": 617
      }
    ]
  },
  "literature": {
    "total": 25,
    "recentArticles": [
      {
        "pmid": "37870949",
        "title": "Pembrolizumab plus chemotherapy in advanced endometrial cancer",
        "journal": "New England Journal of Medicine",
        "pubDate": "2023-11-30",
        "doi": "10.1056/NEJMoa2302312",
        "articleType": "Randomized Controlled Trial"
      },
      {
        "pmid": "37467012",
        "title": "Five-year outcomes with pembrolizumab plus chemotherapy in NSCLC",
        "journal": "Journal of Clinical Oncology",
        "pubDate": "2023-09-01",
        "doi": "10.1200/JCO.23.00747",
        "articleType": "Clinical Trial"
      }
    ]
  },
  "approvedMedicines": {
    "total": 1,
    "authorized": 1,
    "orphan": 0,
    "medicines": [
      {
        "name": "KEYTRUDA",
        "activeSubstance": "pembrolizumab",
        "status": "Authorised",
        "therapeuticArea": "Antineoplastic Agents",
        "orphan": false,
        "biosimilar": false,
        "authorisationDate": "2015-07-17"
      }
    ]
  }
}

Single-source tools (health_clinical_trials, health_pubmed, health_who_indicators, health_ema_medicines, health_food_safety) return a simpler { total, ...aggregates, data: [...] } shape with the source-native field set on each item.

Output fields

pipelineSignals[] (composite drug research)

Signal pattern	Source	Triggered when
"X EMA-authorized medicine(s) found"	EMA	One or more results have authorisationStatus === "Authorised"
"X Phase 3 clinical trial(s)"	ClinicalTrials.gov	One or more trials include "PHASE3" in phases[]
"X actively recruiting trial(s)"	ClinicalTrials.gov	One or more trials have status === "RECRUITING"
"X orphan medicine designation(s)"	EMA	One or more results have orphanMedicine === true
"Extensive published literature (25+ articles found)"	PubMed	PubMed returned 25 or more articles

Clinical trials block

Field	Type	Description
clinicalTrials.total	number	Trial record count (capped at 25 in composite, up to 500 in single-source)
clinicalTrials.byPhase	object	Trial count per phase (PHASE1, PHASE2, PHASE3, PHASE4, Unknown)
clinicalTrials.recruiting	number	Count of trials with status RECRUITING
clinicalTrials.topSponsors[]	object[]	Top 5 sponsors ranked by trial count, with sponsor and trials fields
clinicalTrials.trials[]	object[]	First 10 trial records with nctId, title, status, phases, sponsor, enrollment

Literature block

Field	Type	Description
literature.total	number	PubMed article count
literature.recentArticles[]	object[]	First 10 articles with pmid, title, journal, pubDate, doi, articleType

Approved medicines block

Field	Type	Description
approvedMedicines.total	number	EMA record count
approvedMedicines.authorized	number	Count with status "Authorised"
approvedMedicines.orphan	number	Count with orphanMedicine flag
approvedMedicines.medicines[]	object[]	First 10 records with name, activeSubstance, status, therapeuticArea, orphan, biosimilar, authorisationDate

Single-source envelopes

Tool	Headline fields
health_clinical_trials	total, byPhase, byStatus, data[]
health_pubmed	total, uniqueJournals, data[]
health_who_indicators	total, indicators[], countries[], data[]
health_ema_medicines	total, orphanMedicines, biosimilars, data[]
health_food_safety	total, byNutriScore, data[]

How much does it cost to research a drug?

Global Health Intelligence MCP uses pay-per-event pricing: $0.05 per single-source call, $0.15 per composite drug research call, free for health_list_sources. Platform compute is included.

Scenario	Tool calls	Cost per call	Total cost
Quick test, single trial search	1	$0.05	$0.05
Full drug research (Trials + PubMed + EMA)	1	$0.15	$0.15
Rare-disease landscape (composite + WHO + extra EMA)	3	mixed	$0.25
25-drug competitive pipeline scan	25	$0.15	$3.75
Monthly 500-search competitive intelligence cycle	500	$0.15	$75.00
Weekly EMA status check across 50 medicines for a year	2,600	$0.05	$130.00
1,000 single-source food safety lookups	1,000	$0.05	$50.00

You can set a maximum spending limit per run to control costs. The actor stops when your budget is reached, returning a structured error your pipeline can handle gracefully. Apify's free tier includes $5 of monthly platform credits.

Compared to Cortellis, Citeline Pharma Intelligence, or Evaluate Pharma seats at $40,000 to $100,000 per year, an agent running 500 composite drug research calls per month through this MCP costs about $900 per year. The two are complementary rather than substitutable for full-coverage human-analyst workflows, but for AI-agent pipelines and targeted research the per-call model removes the seat-licensing barrier.

How it works

1. Standby request received. Apify routes the MCP POST to /mcp on the standby instance. The activity timer resets, so the idle-shutdown countdown restarts.
2. MCP tool dispatch. The McpServer matches the tool name (health_drug_research, health_clinical_trials, etc.) and validates input against the Zod schema. Invalid inputs return a structured { error: ... } response without charging.
3. PPE charge. Actor.charge({ eventName }) fires before any upstream call, matching Apify PPE semantics. The free tool (health_list_sources) skips the charge.
4. Sub-actor call via apify-client. Each tool calls one or more sibling actors (ryanclinton/clinical-trial-tracker, ryanclinton/pubmed-research-search, ryanclinton/who-gho-search, ryanclinton/ema-medicines-search, ryanclinton/open-food-facts) with memory: 256 and waitSecs: 120.
5. Composite fan-out. health_drug_research runs ClinicalTrials.gov + PubMed + EMA in parallel via Promise.all. Each sub-actor returns a dataset; the MCP iterates items, computes phase counts, sponsor counts, recruiting count, EMA approved count, orphan count, and builds the pipelineSignals[] array.
6. Sub-actor failure handling. A failed sub-actor returns an empty array (the apify-client wrapper catches errors and logs them). The composite still returns with whatever the other sources delivered, so a slow or failing source degrades the result instead of blocking it.
7. Idle shutdown. A 30-second interval checks Date.now() - lastRequestAt. If the gap exceeds STANDBY_IDLE_TIMEOUT_SECS (default 300), the actor calls Actor.exit() to release platform compute. The next request cold-starts a fresh instance.

Tips for best results

1. Set is_condition correctly on the composite. false (default) searches ClinicalTrials.gov by intervention (drug name); true searches by condition (disease). The same query string ("Alzheimer" vs "donanemab") returns very different trial sets depending on the flag, and the wrong flag produces empty or off-target results.

2. Use ATC codes in EMA queries for therapeutic-class searches. "L01XC18" returns all pembrolizumab-class anti-PD-1 antibodies more precisely than free-text "anti-PD-1". The EMA register search accepts ATC codes alongside name, active substance, and INN.

3. Use PubMed field tags for precision. "semaglutide[Title] AND weight loss[MeSH Terms]" returns far fewer false positives than "semaglutide weight loss". The PubMed source supports the full field-tag and boolean syntax.

4. WHO indicator codes (WHOSIS_xxxxxx) are more precise than keywords. The WHO GHO has hundreds of indicators with overlapping keywords. If you know the code (e.g. "WHOSIS_000001" for life expectancy at birth), use it; otherwise start with a keyword and use the returned indicators[] list to pick the exact code for the follow-up query.

5. Barcode lookup beats name search in Open Food Facts. EAN / GTIN barcodes return the exact product record. Name search is fuzzy and returns many variants of the same product across different regional packagings.

6. Combine status_filter and phase_filter for actionable trial queries. status_filter: ["RECRUITING"] + phase_filter: ["PHASE3"] returns only currently-enrolling late-stage trials, far more useful for competitive intelligence than the full result set.

7. Use health_who_indicators once per country for cross-country comparison. The WHO sub-actor returns the indicator for one country at a time. To compare USA, GBR, IND, run three calls and combine the results. The countries[] field confirms which country each record covers.

8. Tune STANDBY_IDLE_TIMEOUT_SECS for traffic pattern. Bursty agent traffic benefits from a longer idle window (600-900s) to avoid cold starts. Always-on workloads can use the 300s default.

Combine with other Apify actors

Actor	How to combine
Clinical Trial Tracker	The ClinicalTrials.gov sub-actor. Call directly for high-volume trial indexing or batch exports where the MCP overhead is not needed.
PubMed Research Search	The PubMed sub-actor. Call directly for large bibliographic pulls that exceed the 200-result cap inside the MCP.
WHO GHO Search	The WHO Global Health Observatory sub-actor. Call directly for batch cross-country comparisons or scheduled indicator monitoring.
EMA Medicines Search	The EMA medicines sub-actor. Call directly for regulatory portfolio sweeps or scheduled status-change monitoring.
Open Food Facts	The Open Food Facts sub-actor. Call directly for batch barcode lookups or category-wide audits.
Academic Research Intelligence MCP	The academic-literature companion. Use this MCP for the applied health + regulatory side; use Academic Research for multi-database literature reviews (PubMed + Semantic Scholar + Crossref + OpenAlex + ArXiv + ORCID).
Research Integrity Screening MCP	Cross-screen trial sponsors and PubMed authors for retraction, paper-mill, and citation-anomaly flags before citing them in a competitive intelligence report.
Company Deep Research	Pair when researching pharma or biotech sponsors. This MCP returns the pipeline, Company Deep Research returns the corporate-level intelligence.

Limitations

• EMA covers EU-authorised medicines only. No FDA Orange Book, no PMDA (Japan), no MHRA (UK post-Brexit standalone), no Health Canada. A US-only drug will show 0 results in approvedMedicines even when it is widely prescribed in the United States. Use the EMA result as the EU regulatory signal, not as a global approval indicator.
• ClinicalTrials.gov is the U.S. NIH registry. Many international trials also register in EU CTR (Clinical Trials Register), ISRCTN (UK), or national registries that are not covered. Trials registered exclusively in non-US registries will not appear.
• WHO GHO indicator coverage is uneven. Some indicators are well-populated across countries and years; others have gaps for specific regions or older years. The response surfaces what is actually available, but absence of data does not mean absence of the underlying health condition.
• Open Food Facts is crowdsourced. Coverage is strong in France and Western Europe, weaker in the United States, weaker still outside the EU and North America. Some product records have incomplete nutrition data, missing allergens, or stale brand information. Treat as informational, not regulatory.
• PubMed source rate limit (3 req/sec). Large parallel batches of health_pubmed calls will queue at the upstream. The composite drug research tool runs PubMed once per call, so the limit only bites for sustained parallel single-source workloads.
• Composite covers drugs + approvals + literature only. WHO indicators and Open Food Facts are not part of health_drug_research. Use the standalone tools for population-health or nutrition queries.
• No full-text PDF download. The MCP returns trial records, bibliographic metadata, WHO indicator values, medicine register entries, and food product records. Follow the doi, url, or nctId field for full content.
• Sub-actor timeout is 120 seconds. A slow source returns an empty array and the composite still returns with available data. pipelineSignals[] reflects only the sources that delivered.
• Composite caps trial records at 25, EMA at 15. Inside health_drug_research, the trial sub-actor is called with maxResults: 25 and EMA with maxResults: 15 to keep latency predictable. For deeper sets, call the single-source tool directly with a higher max_results.

Integrations

• Apify API, trigger health intelligence searches programmatically from competitive intelligence dashboards, regulatory monitoring tools, or clinical research planning software.
• Webhooks, push new Phase 3 trial counts, EMA status changes, or orphan designations to Slack, email, or competitive intelligence trackers the moment a scheduled search completes.
• Zapier, connect to Airtable or Google Sheets pipeline trackers; auto-log drug research results when new targets are added.
• Make, build weekly drug-pipeline monitoring workflows that diff new trials and EMA approvals against the prior run.
• LangChain / LlamaIndex, embed the MCP as a tool in agent pipelines for automated pharma intelligence, regulatory monitoring, and global health research workflows.

Troubleshooting

health_drug_research returned 0 trials for a known drug. Check the is_condition flag. If set to true, the trial search runs against the disease field, which will return 0 when the query is a drug name. Set is_condition: false (the default) for drug queries. The same applies in reverse: a disease query like "Alzheimer" with is_condition: false searches trials by intervention and may return 0.

health_who_indicators returned nothing for a keyword. The WHO GHO has hundreds of indicators with subtle naming. Try a broader keyword first ("mortality", "life expectancy", "immunization"), use the returned indicators[] list to identify the exact indicator name, then re-query with the indicator code (e.g. "WHOSIS_000001"). Indicator codes give deterministic results, keywords are fuzzy.

health_ema_medicines returned 0 for a known drug. The EMA register is searchable by name, active substance, INN, therapeutic area, ATC code, and MAH. Try the active substance INN (e.g. "pembrolizumab") rather than the brand name (e.g. "KEYTRUDA") if the brand search returns nothing. EMA records use INNs as the primary index.

health_food_safety barcode not found. Open Food Facts is crowdsourced, so coverage is uneven outside the EU. Try a name search if the barcode lookup fails. Products sold only in non-EU markets may not have records.

health_pubmed is slow on large queries. PubMed enforces 3 req/sec at the source. Large max_results values (100-200) take noticeably longer than small ones. For batch workflows, schedule rather than burst.

Cold-start delay on first call. Standby mode shuts the instance down after idle (default 300s) to release platform compute. First request after idle takes ~10-20 seconds to spin up. Subsequent calls in the same window are instant. Increase STANDBY_IDLE_TIMEOUT_SECS if you need longer warm windows.

Tool returns { "error": true, "message": "Spending limit reached" }. Your Apify run has hit the maximum charge limit configured for the run. Increase maxTotalChargeUsd in your run configuration, or purchase additional platform credits.

Responsible use

• All data accessed by this server comes from publicly available health data sources: ClinicalTrials.gov (U.S. NIH), PubMed (U.S. NIH / NLM), WHO Global Health Observatory, the European Medicines Agency public register, and Open Food Facts (crowdsourced open database).
• ClinicalTrials.gov records are sponsor-submitted and not independently validated. Trial registration is not the same as trial completion, peer-reviewed publication, or regulatory approval.
• EMA authorisation status reflects EU regulatory decisions only. It is not equivalent to U.S. FDA, UK MHRA, Japanese PMDA, or any other national regulator's decision.
• Open Food Facts data is crowdsourced and informational, not regulatory. Do not use Nutri-Score or NOVA grades as a substitute for formal nutritional analysis or regulatory food-safety determinations.
• This MCP is not a medical advice service. Do not use its output for individual medical, dietary, prescribing, or treatment decisions. Pipeline signals, trial recruitment status, and EMA authorisation flags are research inputs, not clinical guidance.
• Comply with applicable data-protection regulations in your jurisdiction when storing or sharing health-related search outputs.
• For guidance on web scraping and data use legality, see Apify's guide.

FAQ

How is this MCP different from Cortellis, Citeline Pharma Intelligence, or Evaluate Pharma? Cortellis, Citeline, and Evaluate Pharma are curated pharma intelligence platforms built for human analysts, priced at $40,000 to $100,000 per seat per year, and reached through web dashboards. This MCP exposes the underlying public health data (ClinicalTrials.gov, EMA, PubMed) plus computed pipeline signals as MCP tools an AI agent can call directly, with per-call pricing ($0.05 single-source, $0.15 composite). The two are complementary rather than substitutable: Cortellis adds curated analysis, conference intelligence, and licensing-deal data this MCP does not have. This MCP wins on AI-agent integration, per-call economics, and direct access to the raw registry and approval data without a seat license.

What is Nutri-Score? Nutri-Score is a front-of-pack nutritional rating system used across France, Belgium, Germany, the Netherlands, Spain, and Luxembourg, ranking products from A (best nutritional profile) to E (worst). It is computed from per-100g nutrient content (sugar, saturated fat, sodium, fibre, protein, fruit / vegetable / nut content). The nutriscore_grade filter in health_food_safety lets you filter products by grade for audits or comparisons.

What is NOVA processing level? NOVA is a classification system that groups foods by the extent of industrial processing: NOVA 1 (unprocessed or minimally processed), NOVA 2 (processed culinary ingredients), NOVA 3 (processed foods), NOVA 4 (ultra-processed foods). Open Food Facts records carry a NOVA value where it has been computed, useful for ultra-processed food audits and consumer-protection research.

Why does the composite drug research tool only query three sources (Trials + PubMed + EMA), not all five? The composite is drug-pipeline focused. WHO Global Health Observatory is population-level indicator data (mortality, immunization, life expectancy by country) and Open Food Facts is food product data. Neither fits the "research this drug or condition" workflow the composite is designed for. They remain available as standalone tools for population-health and nutrition queries.

Does this MCP cover FDA-approved drugs? Not directly. The EMA sub-actor covers EU-authorised medicines only. FDA-approved drugs that are also EU-authorised will appear in EMA results; FDA-only drugs will not. For US regulatory coverage, you currently need a separate FDA Orange Book or DailyMed integration.

Can this MCP replace a Cortellis or Citeline subscription? For full-coverage human-analyst workflows that need curated competitive intelligence, conference reporting, and licensing-deal databases, no. For AI-agent pipelines, targeted drug-pipeline queries, and per-call economics, yes. Many teams use both: Cortellis or Citeline for the human-analyst surface, this MCP for the agent-callable raw-data surface and for cost-controlled spot queries that do not justify a seat.

How fresh is the trial data? ClinicalTrials.gov is updated by sponsors and curated by the U.S. NIH on an ongoing basis. New trials typically appear within days of registration. Status changes (RECRUITING to ACTIVE_NOT_RECRUITING, etc.) reflect what sponsors have reported. The MCP queries the live API on every call, so results are as fresh as ClinicalTrials.gov itself.

How fresh is the EMA data? The EMA public medicines register reflects EU regulatory decisions and is updated by the agency. New authorisations, withdrawals, refusals, and suspensions typically appear within days to weeks of the decision. Authorisation date and status fields reflect the official EMA record at query time.

Do I need any API keys to use this MCP? No. All five data sources (ClinicalTrials.gov, PubMed, WHO GHO, EMA, Open Food Facts) are free public health APIs with no authentication required for the access patterns this MCP uses. You only need an Apify token (for billing) configured in your MCP client.

Can I schedule searches to run periodically? Yes. Use the Apify Scheduler to trigger the actor on a daily, weekly, or monthly cadence. Configure a webhook to push new Phase 3 trial counts, EMA status changes, or orphan designations to your notification system. This is useful for competitive pipeline monitoring, regulatory affairs tracking, and global health programme reporting.

Is it legal to use this tool for pharma competitive intelligence and global health research? All five underlying data sources (ClinicalTrials.gov, PubMed, WHO Global Health Observatory, EMA medicines register, Open Food Facts) are publicly available health databases that explicitly support programmatic access. Accessing and analysing public health records for competitive intelligence, regulatory monitoring, clinical research planning, and food-safety audits is a standard practice. See Apify's guide on web scraping legality for broader context.

Help us improve

If you encounter issues, you can help us debug faster by enabling run sharing in your Apify account:

1. Go to Account Settings > Privacy
2. Enable Share runs with public Actor creators

This lets us see your run details when something goes wrong, so we can fix issues faster. Your data is only visible to the actor developer, not publicly.

Support

Found a bug or have a feature request? Open an issue in the Issues tab on this actor's page. For custom solutions or enterprise integrations, reach out through the Apify platform.