From transfer packet to defensible code set
Four stages. Every output carries the page and sentence that supports it, so the answer to "why is this code on the claim?" is already written.
1. Secure upload
The MDS coordinator uploads the scanned hospital transfer packet, any length. It is encrypted the moment it lands and never leaves the covered environment. Duplicate packets are detected by content and skipped.
2. Extraction with citations
Every candidate condition, medication, lab and imaging finding is pulled from the packet with its page number and verbatim quote, and classified by where it came from: discharge diagnosis list, narrative, medication inference, lab or imaging, problem list.
3. Judgment under a frozen rubric
Each candidate gets one of three verdicts: supported, insufficient (with exactly what is missing and a compliant, non-leading physician query), or not supported. The rubric was ratified by a physiatrist over 288 physician-graded charts and carries a published fingerprint, so the same standard applies to chart 1 and chart 1,000.
4. Deterministic PDPM mapping and the revenue report
Supported codes map to the CMS FY2026 ICD-10 to PDPM tables with no model in the loop. The NTA comorbidity points they carry are priced against your facility's wage index and Medicare average length of stay, producing a before-and-after report per admission, per patient day, and per year. See the NTA reference table.
What the pilot involves
Send your next 15 admissions. We return the evidence-cited code set for each, the physician queries, and the revenue report for your facility. You decide what it is worth. Request a Pilot