Back to All Features
AI & IntelligencePlan: Clinic

RAG Knowledge Base

Clinical drug information powered by your own documents + global medical database

Overview

A 3-tier query system: Tier 1 searches the clinic's own uploaded documents (protocols, fee lists, formulary), Tier 2 searches the global clinical database (BNF, WHO guidelines, DRAP drug list, Pakistan National Formulary uploaded by EliteHMS), and Tier 3 falls back to Groq Llama AI. Responses are labeled by source. Store owners can upload up to 25 documents (15 MB total).

Standard Operating Workflow

1

Upload Documents (Admin)

Pharmacy uploads own protocols, fee schedules, medicine list. Supported: PDF/DOCX/TXT/MD (up to 5 MB each).

2

Processing

System splits text into 600-char chunks → Gemini embedding (768-dim) → stored in Qdrant Cloud with store ID tag. Status: Ready.

3

Staff Query

Doctor or pharmacist types question: 'Paracetamol dose for a 5-year-old?'

4

3-Tier Search

Tier 1: store docs (no match) → Tier 2: global DB (Paracetamol monograph found) → Groq generates answer with source citation.

5

Labeled Response

'📋 Your Knowledge' / '🌐 Clinical Database' / '🤖 AI Knowledge'. User knows exactly how confident to be.

Key Operational Benefits

Pharmacists answer drug queries in seconds — no manual book lookup
Store-specific documents stay private — isolated by store ID in Qdrant
Source labeling shows whether answer is from clinic's own docs or global DB
Global clinical DB maintained by EliteHMS — updated regularly

Real-World Healthcare Scenarios

Pediatric Dose Query

Pharmacist asks: 'Paracetamol syrup dose for 3-year-old?' System searches global DB, finds monograph chunk, returns: '10–15 mg/kg every 4–6 hours. For 120mg/5ml syrup: 2.5–5ml per dose.' Source cited.

Target Search Keywords

medical knowledge base softwareclinical decision support Pakistandrug information system hospitalRAG knowledge base pharmacyQdrant vector database medical

Related System Modules