Medical Research Data Director

Medical research data rarely speaks one language.

MRI, laboratory measurements, PET, pathology, genomics and research databases each describe part of a biomedical picture in their own way. This guide shows how the prototype makes those connections visible without asking you to become a medical terminologist first.

Take the narrated tourResearch & backgroundOpen the app

The idea in one paragraph

Different instruments can describe the same research subject from very different angles. The difficult part is making sure their measurements can be connected without losing identity, time, units, provenance or local meaning. The Medical Research Data Director records those correspondences: the original field, the shared concept it may map to, the evidence for the connection and whether that mapping has been reviewed.

1. Meet the data
What is it?
2. Connect meanings
How do terms line up?
3. Try the translation
Apply mappings
4. Check the work
What needs review?
5. Take it with you
Export results

The easiest way to try it

  1. Open Crosswalk Studio.
  2. Find psa_ng_ml. Notice that a local term is connected to a shared standard; you do not need to memorize the code.
  3. Open Transform Lab and click Load synthetic sample.
  4. Open Quality & Provenance to see what still needs checking.
  5. Open Exports to see how the work can be saved and reused.
You do not need to create a medical mapping. The examples are already in the prototype. The point is to understand the metadata workflow and the places where expert judgment remains necessary.

One example in everyday language

The local dataset says: psa_ng_ml
It means: total prostate-specific antigen
A shared standard calls it: LOINC 2857-1
The unit is: ng/mL

The app keeps both sides of the connection. It preserves the original language and records the bridge to a shared standard, along with confidence, review status and provenance.

What the unfamiliar words mean

TermPlain-English meaning
CrosswalkA documented translation between one data language and another.
Source fieldThe original column or variable name.
Canonical targetThe shared concept that different local terms can point to.
LOINCA common naming system for laboratory tests and observations.
DICOMA standard for medical images and the information that describes them.
SNOMED CTA large clinical vocabulary for findings, anatomy and procedures.
FHIRA common way for health-data systems to package and exchange information.
ProvenanceThe paper trail showing where a mapping came from and what supports it.
ConfidenceHow strong the mapping appears to be; it is not the same as expert approval.

Where AI fits

If Gemini is enabled, it can suggest possible mappings for one de-identified field. Think of it as a first-pass assistant rather than an authority. Its suggestions remain drafts until a knowledgeable person checks them against the appropriate terminology source.

The rest of the app works without Gemini.

Where this app stops

This prototype asks, “What does this data mean, and how can its meaning connect to other data?” It does not decide who may access sensitive data or what institutional controls are required. Those are separate privacy, security and governance questions.

Where the research comes from

The working prototype is the implementation layer of a wider research argument. Diagnostic Synthesis asks what becomes possible when the partial views produced by biomedical instruments can be joined reliably. The Research Data Director — Biomedical Variant turns that problem into a metadata, crosswalk, provenance and governed-publication architecture.

Explore Research & BackgroundBiomedical blueprintDiagnostic Synthesis

The sentence to remember

Medical instruments create measurements; metadata creates the connections that let those measurements become a coherent, reviewable research record.

Take the guided tourOpen the application