Discipline 02 · Data & AI
Detecting Signal: Data & Artificial Intelligence
Value tries to answer one question: what is something worth? The weights assigned to the answer — experience, cost, alternatives, long-term sustainability, and more — are complex, inherently hard to answer, and often subjective.
We take data and work to make it “promptable” based on the values defined in the contract.
Orchard builds on Medicare and other research data. We draw on data sets from ResDAC, the CMS Research Data Assistance Center, and other sources that are public or available to researchers.
Data engineering
Bringing claims, enrollment, provider and reference files into consistent, documented, analysis-ready structures.
Modeling
Statistical and machine learning methods for cost, utilization and risk in Medicare populations.
Review
We model contracts and flag necessary data to make contracts frictionless to execute with clear data sources.
Sources
Examples of Relevant Data Sources
| # | Source group | What it covers |
|---|---|---|
| 01 | ResDAC & CMS research identifiable files | Beneficiary-level Medicare files requested through ResDAC, the CMS Research Data Assistance Center, under a data use agreement. |
| 02 | Research environments | Where research identifiable and limited data are accessed and analyzed. |
| 03 | CMS public use files | Aggregated Medicare files published on data.cms.gov. |
| 04 | Medicare Advantage & Part D | Enrollment, payment and risk adjustment publications from CMS. |
| 05 | AHRQ | All-payer and survey data from the Agency for Healthcare Research and Quality. |
| 06 | Other public and researcher sources | Surveys, reference files and context data used alongside claims. |