Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for clinical data managers

Clinical External Data Reconciliation Agent

Vendor and lab data reconciled with the study database on every transfer

Clinical External Data Reconciliation Agent: what goes in, what the agent does and what you get

What it does

Trials receive data from central labs and imaging vendors that must agree with the study database, and mismatches found late can delay lock. On each transfer this agent first checks the vendor file against the agreed format and expected record counts. Then it compares key fields with the study data: subject, visit, test and result. It applies agreed rules, such as unit conversions, before calling anything a mismatch. Where records still do not match, it classifies the cause, such as a missing visit, a unit difference or a subject not yet entered, and drafts a query to the vendor or the study team. It will not mark a transfer reconciled until every mismatch is explained or resolved. You approve vendor queries and sign-off. Edge case: a correctly converted result in different units is matched, not flagged.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueYes, continueApprovedNoNo 1 STARTS WHEN External data file received 2 USES A TOOL Check the file against the agreed format 3 CHECKS THE RESULT Does the record count match what was expected? If not: list missing or extra records by subject andvisit. Back to step 2. 4 USES A TOOL Apply agreed rules such as unit conversions 5 USES A TOOL Compare key fields with the study database 6 CHECKS THE RESULT Does every record reconcile under the agreed rules? If not: classify each mismatch and draft a query to thevendor or study team. Back to step 5. 7 DOES Track the transfer status toward reconciliation 8 YOU APPROVE Data manager approves queries and reconciliationsign-off 9 RESULT Reconciliation record for the transfer
Read the steps as a list
  1. External data file received
  2. Check the file against the agreed format
  3. Does the record count match what was expected?If not: list missing or extra records by subject and visit. Back to step 2.
  4. Apply agreed rules such as unit conversions
  5. Compare key fields with the study database
  6. Does every record reconcile under the agreed rules?If not: classify each mismatch and draft a query to the vendor or study team. Back to step 5.
  7. Track the transfer status toward reconciliation
  8. Data manager approves queries and reconciliation sign-offThe agent waits here for your OK.
  9. Reconciliation record for the transfer

How it decides

Records match only when key fields agree after agreed conversions; a transfer is reconciled only when every mismatch is explained or resolved.

  • Match only after agreed unit conversions
  • Classify mismatches by cause
  • Reconcile only when all mismatches are resolved

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Key fields to match
  • Unit conversions
  • Expected counts per transfer
  • Query templates

What keeps you in control

It always asks you first

  • Sending queries to vendors
  • Reconciliation sign-off

Hard limits

  • Never edits study or vendor data
  • No vendor query without approval

It stops when

  • Done: transfer reconciled or mismatches tracked
  • Stop: the transfer format spec is missing

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensA central lab file for study RS-330 arrived on April 4 with 480 results against 500 expected, so the count check failed. The agent found 20 belonged to a subject not yet entered and flagged it to the site. Several glucose values matched only after unit conversion. It drafted a vendor query for three genuinely missing visits, which the data manager approved. The rerun on April 11 reconciled.

More agents for clinical data managers