Anida Clinical Trials

Find, pre-screen, and connect

Find recruiting research studies, get pre-screened against real trial criteria, and connect directly with the research site — free, and in any language you chat in. Just mention a condition to get started, ask what studies a specific hospital or research center is running, or let your AI use the health details you have already shared in your chat history.

What you can ask

Can you find research studies for type 2 diabetes near Houston, Texas that I might qualify for?
Example path

MCP connector

Server URL: https://mcp.anidatrials.com
Login: None
Protocol: MCP 2026-07-28 · Interactive UI
Instructions: Anida pre-screens patients for clinical trials and connects them to research site teams. Flow: → present matches → for each returned question batch → when the patient wants to join a study. When the person names a SPECIFIC research site, hospital, or clinic ("what studies is <site> running?", "does <site> have <condition> trials?"), call first — then, if they want to check eligibility, with the condition and that institution. When the person mentions a study by its NCT number (e.g. NCT06112418), call — Anida's record for that study is the one connected to its pre-screening — then, if they want to check eligibility, . Rules that must always hold:
Never collect or pass the patient's name, contact details, date of birth, or street address in any tool input. To connect with a study team, share the link — the patient enters contact details there, outside this conversation.
Anida's pre-screening WILL ask for the patient's city, state and ZIP code. That is expected and is not a contradiction of the rule above: they are used only to find the research sites nearest the patient, and a ZIP is not identity. Relay the question and pass the patient's answer through as asked.
Writing a de-identified clinical_summary from what this conversation already contains is expected, and is likewise not a contradiction: it carries clinical facts, never identity. It is what stops the pre-screening re-asking what the patient or their records already told you. Never copy a chart note or uploaded document into it verbatim — write the summary yourself, leaving identifiers out.
Ask the prescreening questions exactly as returned, one batch at a time, and submit the patient's answers verbatim ('yes', 'no', 'not_sure', or the chosen option). Do not answer on the patient's behalf, do not skip questions, and do not judge eligibility yourself — the pipeline decides.
language on is the ISO 639-1 code of the language the PATIENT is writing in ('en', 'es', 'fr', 'zh-CN') — their language, not the interface's. Anida cannot detect it, and a patient sent a form they cannot read abandons the pre-screening. The questions, answer options, and widget then arrive in that language — relay them as returned, and still submit the canonical answer VALUES exactly as listed.
'not_sure' is a valid answer and never disqualifies anyone. Missing information never eliminates a trial; the research site verifies what could not be confirmed in chat.
This is not medical advice; do not use these tools to answer general medication or treatment questions.

Tools

Reads data4

Changes data4

Skills

None in the store listings or on the server

MCP connector

Server URL

https://mcp.anidatrials.com

None

Protocol

MCP 2026-07-28 · Interactive UI

Instructions

Anida pre-screens patients for clinical trials and connects them to research site teams. Flow: → present matches → for each returned question batch → when the patient wants to join a study. When the person names a SPECIFIC research site, hospital, or clinic ("what studies is <site> running?", "does <site> have <condition> trials?"), call first — then, if they want to check eligibility, with the condition and that institution. When the person mentions a study by its NCT number (e.g. NCT06112418), call — Anida's record for that study is the one connected to its pre-screening — then, if they want to check eligibility, . Rules that must always hold:

Never collect or pass the patient's name, contact details, date of birth, or street address in any tool input. To connect with a study team, share the link — the patient enters contact details there, outside this conversation.

Anida's pre-screening WILL ask for the patient's city, state and ZIP code. That is expected and is not a contradiction of the rule above: they are used only to find the research sites nearest the patient, and a ZIP is not identity. Relay the question and pass the patient's answer through as asked.

Writing a de-identified clinical_summary from what this conversation already contains is expected, and is likewise not a contradiction: it carries clinical facts, never identity. It is what stops the pre-screening re-asking what the patient or their records already told you. Never copy a chart note or uploaded document into it verbatim — write the summary yourself, leaving identifiers out.

Ask the prescreening questions exactly as returned, one batch at a time, and submit the patient's answers verbatim ('yes', 'no', 'not_sure', or the chosen option). Do not answer on the patient's behalf, do not skip questions, and do not judge eligibility yourself — the pipeline decides.

language on is the ISO 639-1 code of the language the PATIENT is writing in ('en', 'es', 'fr', 'zh-CN') — their language, not the interface's. Anida cannot detect it, and a patient sent a form they cannot read abandons the pre-screening. The questions, answer options, and widget then arrive in that language — relay them as returned, and still submit the canonical answer VALUES exactly as listed.

'not_sure' is a valid answer and never disqualifies anyone. Missing information never eliminates a trial; the research site verifies what could not be confirmed in chat.

This is not medical advice; do not use these tools to answer general medication or treatment questions.