Pogosh CDS — Structured Clinical Decision Support

What it does

Pogosh CDS is a structured, rule-based decision support engine. Given a condition and patient context it surfaces the high-yield labs that change the decision, flags the ones that should have been ordered and were not, identifies escalation criteria from current vitals, recommends imaging with patient-specific alternatives, and names the specialist to call along with the indication to give them.

The problem is cognitive load, not knowledge. A clinician running six patients at once misses steps because guideline lookup does not fit inside every encounter and results arrive between handoffs. Pogosh is built to be called on every chart interaction rather than once per encounter.

What the engine returns

• High-yield labs with LOINC codes, rationale and critical thresholds — procalcitonin to separate bacterial from viral pneumonia, for instance, or blood cultures required before antibiotics at higher pneumonia severity index classes.
• Imaging guidance with patient-specific substitution: a shellfish allergy flags CTPA and surfaces V/Q; CKD flags contrast exposure.
• Specialist routing with actual criteria — when to call pulmonology versus infectious disease versus the ICU, and the specific indication to give them, instead of "consider specialist referral".
• Escalation triggers tied to vitals and lab thresholds that indicate deterioration.
• Missing-order alerts that cross-reference what should be ordered against what is already ordered or resulted.
• Patient-specific adjustment for comorbidities, allergies and existing orders — a first PE in a 30-year-old is not the same problem as a prior DVT on anticoagulation in a 70-year-old.

How it is delivered

A REST API on Cloudflare edge returns structured JSON. The EMR, clinical app, or AI assistant passes a condition ID plus patient context and gets recommendations back. OpenEMR deployments get a drop-in module that adds a copilot panel to the chart view.

It is a deterministic rule engine — there is no generative text in the recommendation path, and every recommendation traces back to a condition and a stated clinical rationale rather than a model output.

Coverage, program and pricing

• 413 conditions across 40+ specialties — emergency medicine, internal medicine, interventional radiology, gastroenterology, pulmonology, nephrology, neurology, cardiology, endocrinology, infectious disease and more.
• Clinic: $149/month, up to 5 providers. Practice: $299/month, up to 15 providers. Enterprise: custom.
• Every tier includes the full condition library, unlimited API calls and the OpenEMR module, with no per-call fees.
• Clinical Validation Partner Program: three months of full access in exchange for structured monthly feedback. It is framed as a collaboration, not a free trial.

Privacy and scope

The engine processes condition IDs and clinical parameters only — no names, no MRNs, no other identifiers. OpenEMR deployments stay inside the customer's own infrastructure.

It provides decision support. It does not substitute for clinical judgment, and the accountable clinician is still the accountable clinician.

For educational reference only. Clinical decisions require the physician's judgment of the individual patient. This tool supports that judgment; it does not replace it.

Written and reviewed by Pouyan Golshani, MD — March 2026