medicationexplorer data-backed, AI-assisted code set creation
About ▾
Overview what Medication Explorer does, and how to connect your AI Data and definitions the sources, what a code set covers, and the words its page uses How to read a code set what each part of a code set page and its workbook shows Disclaimer for research use; not medical advice or billing guidance Privacy what is kept, and who can see it Sources and notices where the data comes from, and the notices its terms ask for

How to read a code set

Every saved code set has its own page and an Excel workbook. This page tells what each part shows. The words they use are defined on Data and definitions.

Opening and following

  • Text with a chevron (›) opens in place. Click it again to close it.
  • Only blue underlined text is a link to another page.
  • Links out of the site, such as DailyMed, PubMed and DOIs, open in a new tab.

The top box

  • The request, and its description as the user's AI paraphrased it.
  • Cautions for the whole set, if the AI gave any.
  • How the request was refined by AI: each question the AI asked, and the user's response.
  • The history: when the code set was built, reviewed or revised, and with which model. Earlier versions are linked.
  • Download the workbook.

The medications

  • Included (N) lists the medications in the code set. Not Included (N) lists those the AI examined and left out, with any left out by the rules without a decision of the AI's.
  • Medications are grouped by FDA class or by ATC class. Switch with "Grouped by".
  • Each medication is one card. Closed, a card shows the medication, its tags, and how many NDC and HCPCS codes it has. Click it to open it.

An open card

  • Tags tell why the medication is included: a label states the use, or it is an off-label use with a source, or from the AI's knowledge. An orange tag marks what the AI supplied from AI knowledge. Check it.
  • Associated: acts on marks a medication that treats something that goes with the condition, not the condition itself (a complication, a symptom, an effect of its treatment), and says what it acts on.
  • AI Explanation: what the AI wrote when it decided.
  • Indication statements: what the labels state. Each label's Indications section opens in place, beside a link to the label on DailyMed.
  • Other sources for this use: guidelines, payer policies and publications the AI examined. A tag tells what each source did for the decision.
  • Products: the medication's products, included and not included, with their forms and routes.
  • Other indications: other uses its labels state. The medication's codes are used for these too.
  • FDA class and ATC class.

The workbook

The workbook lists every code. Its sheets:

  • Contents: each sheet and what it holds. Click a sheet's name to open it.
  • Column guide: what each column on each sheet means.
  • Summary: the request, the counts and the cautions.
  • Included and Not Included: each decision, with its reason and the label evidence.
  • Labels cited: each label behind an included medication's indication statements, with a link to it on DailyMed.
  • NDC codes and HCPCS codes: the codes, each with its medication and product.
  • Codes left out: codes of included medications that were left out, and why.
  • Issues: anything the data or the AI flagged.
  • AI Interaction: the request, its description, the questions the AI asked and the user's responses.
  • Other sources for this use: each source on its own row, linked where it can be, and what it did for the decision. A source that names several PubMed articles has a row for each.

A drug or class lookup has no Labels cited or Other sources for this use sheet unless it cites them, and shows its NDC and HCPCS codes right after the Summary. A sheet with nothing in it says so in its first row. Otherwise, a column with nothing in it for a code set is left out of its workbook. A medication's RxCUI is its RxNorm code. Every row has an internal id: the RxCUI, or ours for a product, a kit or a combination that RxNorm has no code for. In the workbook, as on the page, only blue underlined text is a link.