The Population Panel lets you ask a question about every patient in your practice at once — in plain English — and get back a list where every row shows the evidence behind it. It runs on the coded clinical facts that DeepCura's ambient extraction stages from recorded visits and a clinician approves onto the chart. It is the engine behind DeepCura's healthcare CRM software that fills itself.
The three query modes
Every panel question runs over one coded concept (a LOINC lab or vital, an ICD-10 diagnosis, an RxNorm medication, a CVX immunization, or a custom trackable) or one curated registry family — diabetes, hypertension, or chronic kidney disease.
- Threshold — patients whose latest value crosses a line. Example: "every diabetic with a latest A1c above 9." Results show the match count next to the honest denominator (e.g., 28 matched · 216 scanned).
- Exists — patients with any qualifying fact on record inside a time window. Example: "everyone with a hypertension diagnosis on file."
- Missing — the care-gap and recall mode: patients in a registry with no qualifying value inside the window you choose (1–120 months). Example: "diabetic patients with no A1c recorded in the last 6 months." Each row shows the last known value and its date so your team can prioritize the most overdue first.
Every row carries its evidence
Panel results aren't a black box. Each row links to the fact behind it — including the verbatim quote it came from and, for spoken facts, the exact second of visit audio. Verification badges show how each fact entered the chart (clinician verified, EHR record, or AI extracted awaiting review). Denominators are honest: if part of the panel couldn't be read, the result is flagged rather than silently under-counted.
From list to outreach
A missing-mode result exports to CSV in one click, formatted for front-desk use. That CSV is a care-gap list for insurance-based practices and a patient recall list for cash-pay and wellness practices — same query, same chart, two vocabularies.
What DeepCura does not do: there are no bulk texts, drip sequences, or automated outreach campaigns. Your staff works the list person to person, and when patients call back, the AI receptionist answers 24/7, verifies identity, and books the appointment.
How to run your first panel query
- Open any patient chart and press P (or select Population Panel in the left rail).
- Pick a starter preset — for example, Diabetes: no A1c in 6 months — or build your own query with a concept, a mode, and a window.
- Review the results, click any row to see its evidence, and select Export CSV for outreach.
You can also ask out loud: the same queries run from the command bar by typing or dictating the question in plain English.