Diabeteris Foundry
From research data to a result you can review.
Foundry is the diabetes research platform we are building. It connects data preparation, document review and analysis, with Studio as the workspace researchers use to see the whole study.
One study.
A clear record of the work.
A researcher should be able to open a study and find its approved data, analysis settings, source documents and review notes together.
Foundry is designed to keep those connections visible. Our internal research team will use it first; partner access and dedicated care applications are later steps.
Foundry in developmentFour steps from source to review.
Find the right evidence.
Find a dataset or paper relevant to the study. Check who collected it, what it contains and whether the proposed use is permitted.
The tools inside Foundry.
Lens: review documents
Propose fields from a report or paper, then show the source passage so a reviewer can check the interpretation.
Bench: compare methods
Compare an analysis or model with a simpler reference method. Keep errors, test conditions and reviewer notes with the result.
Conduit: prepare data
Import an approved file, identify missing or repeated records, and preserve the original values when a mapping changes.
Studio: see the study
Bring data checks, analysis runs, source documents and review requests into one workspace. Controlled partner APIs are planned later.
More than
a stream of numbers.
Our research explores glucose observations alongside available information about meals, recorded insulin, activity and everyday context.
Explore the researchSee what a researcher would see.
Inspect a short glucose record, find a gap and check the original rows. This interactive Studio concept shows the kind of review we are building into Foundry.
A closer look at the record.
- Source rows
- 12in this example
- Plotted readings
- 10after exclusions
- Review flags
- 2one gap, one repeated row
Observed sample values Missing sample, not interpolated
A concept for a product in development. The displayed counts describe these 12 invented rows; they do not measure clinical performance or the size of our data library.
Authorize
Start with the permitted use.
Check source rights, study purpose and access before importing data. A public download is not permission for every use.
- Planned artifact
- Source and permission record
- Example stop condition
- Access denied → stop before import
The Metra research program
Models built around diabetes questions.
Metra is our planned family of computational models for diabetes research. Metra-1 establishes statistical and machine-learning comparisons. Metra-2 explores temporal neural networks and deep learning when the data and benchmarks justify the added complexity.
Alongside Metra, Lens connects OCR, natural language processing and evidence retrieval. Later LLM-assisted workflows would help researchers read and compare authorized sources, with citations and human review.
These are development programs. We have not released a validated clinical model, and the research workflow does not determine insulin doses.
Explore model researchA global diabetes evidence library
Bring the world’s diabetes knowledge into reach.
Our ambition is to build the world’s largest diabetes-focused dataset and evidence repository: connecting eligible records, published studies, laboratory knowledge and molecular references in one research foundation.
That is a long-term target. Today, we are building the tools to qualify sources, preserve permissions and make each connection useful. Coverage, quality and independent usefulness will define the scale we can substantiate.
Explore the data foundation
Compute & security roadmap
Serious computing. Defined boundaries.
We plan to scale from controlled local research to high-memory systems, GPU-accelerated training and distributed experiments as the data, funding and evaluation justify them.
Planned controlled research environment
- AccessIdentity · study purpose · permitted use
- DataEncrypted stores · versioning · traceable transformations
- ComputeCPU preparation · GPU training · isolated model evaluation
- ReleaseReviewer approval · scoped APIs · audit records
Infrastructure as code · Separate research and production environments · Tested recovery
High-performance computing with a purpose.
Our planned workloads span machine learning, deep neural networks, computational models and later large language models. We will select hardware against memory needs, reproducibility, training time and cost, with capacity and benchmarks documented as infrastructure is commissioned.
AWS GovCloud (US), where appropriate.
AWS GovCloud (US) is a deployment option we intend to evaluate for eligible workloads. Account eligibility and regional service availability must be confirmed. This is an infrastructure plan, not a claim of an existing GovCloud deployment or certification.
Our planned controls include least-privilege access, encryption in transit and at rest, isolated workloads, audit trails, restriction propagation and tested backups. Cloud hosting does not replace application security or governance.
Different data. Different questions.
A measurement, a document and a molecular finding each contribute a different kind of evidence. Our goal is to connect them without losing what makes them meaningful.
Glucose & daily context
Time-stamped readings, device context and recorded meals or activity.
Which patterns deserve a closer look?
Timing · Units · Missing observations
Laboratory & clinical records
Reported test values, dates and documented clinical context.
How does an observation change over time?
Assay · Reference context · Record date
Language & literature
Scientific papers, study descriptions and authorized reports.
What was studied, in whom, and with what result?
Source · Population · Study design
Food & molecular evidence
Ingredient composition, biological pathways and measured molecular findings.
Which connections justify further investigation?
Identity · Preparation · Tissue or assay
Illustrative data categories, not an inventory of acquired records. Access, permissions and fitness for a specific study must be established separately.
A wider view of the evidence.
Our research direction spans clinical measurements, scientific literature, population studies and molecular knowledge. Each source needs its own access and suitability review.
Public scientific resources
Literature indexes, research repositories and food or molecular reference resources help identify evidence worth examining. Public availability alone does not establish permission to reuse a dataset.
Purposeful collaboration
Future work with laboratories, care teams and volunteers would use defined research questions, appropriate permissions and an agreed data-use process.
Built by people.
Reviewed through evidence.
Our intended engineering culture brings software, data and scientific review together. The work must remain understandable to the people who rely on it.
Engineering the foundation
Developing research tools that people can inspect, question and improve.
Understanding populations
Studying population patterns starts with understanding the source, context and limits of the data.
Research partnerships
Help build better
diabetes research.
Discuss a study, a data source or a practical need with us.
Contact us