---
license: mit
task_categories:
- question-answering
- feature-extraction
- text-retrieval
- table-question-answering
language:
- en
tags:
- healthcare
- pharmacy
- medical
- drugs
- rag
- biomedical
- pharmacology
- india
- drug-interactions
- clinical-nlp
- healthtech
size_categories:
- 100K<n<1M
---

# PharmaLens: 200k Medicine Catalog (Salts, Prices, Interactions and Reviews)

> I spent weeks compiling and cleaning this retrieval database for a project. Instead of letting 300MB+ of structured pharmaceutical data sit idle on my hard drive, I am open-sourcing it. Use it for your RAG pipelines, chatbots, pricing tools, or whatever else you are building.

---

## Overview

Finding clean, structured pharmaceutical datasets with commercial brand names, active salt compositions, Indian market pricing, and clinical context in a single format is difficult. Most existing datasets are either US-only NDC codes or unlinked raw review text.

This repository provides 195,605 Indian medicine records linked with active chemical ingredients, real retail pricing in INR, side effect profiles, drug-drug interaction warnings, and aggregated patient review data across 986 manufacturers.

---

## Dataset Summary

| Feature | Coverage / Stats | Details |
| :--- | :--- | :--- |
| Total Medicines | 195,605 | Master branded drug catalog |
| Salt Compositions | 100% (195,605) | Active chemical ingredients and dosage strengths |
| Descriptions | 100% (195,605) | Product usage, mechanism, and indications |
| Side Effects | 100% (195,605) | Common and adverse reaction lists |
| Drug-Drug Interactions | 100% (195,605) | Interacting drugs and severity ratings |
| Market Pricing (INR) | 89.9% (176,169) | Retail market pricing across 986 manufacturers |
| Patient Reviews | 60.7% (118,803) | Review counts, average ratings, and treated conditions |
| Therapeutic Categories | 252 | Clinical classifications |

---

## Files Included

### 1. base_catalog.json (Main Dataset, ~315 MB)

Contains the full 195,605 medicine records. Includes pre-normalized search tokens `_clean_salt` and `_clean_product` for fast BM25 or vector retrieval.

Sample Record:

```json
{
  "product_name": "Human Insulatard 40IU/ml Suspension for Injection",
  "salt_composition": "Insulin Isophane (40IU)",
  "manufacturer": "Novo Nordisk India Pvt Ltd",
  "price": "₹133.93",
  "category": "Human Insulin Basal",
  "description": "Human Insulatard 40IU/ml Suspension for Injection is used to improve blood sugar control in adults and children with type 1 and type 2 diabetes mellitus...",
  "side_effects": "Hypoglycemia (low blood glucose level), Injection site allergic reaction, Lipodystrophy...",
  "drug_interactions": "{\"drug\": [\"Benazepril\", \"Captopril\"], \"effect\": [\"MODERATE\", \"MODERATE\"]}",
  "_clean_salt": "insulin isophane",
  "_clean_product": "human insulatard ml for",
  "layer1_reviews": {
    "review_count": 3,
    "avg_rating": 9.0,
    "conditions": [
      "Diabetes, Type 2",
      "Diabetes, Type 1"
    ],
    "sample_reviews": [
      "I am terrified of needles and was very upset when I found out..."
    ]
  }
}
```

### 2. shortages.json (1,850 records, ~1.7 MB)

Drug shortages database covering generic names, manufacturer contact information, availability status, and reasons for shortage.

### 3. bad_actors.json (34 records, ~16.7 KB)

Official regulatory quality inspection notices (CDSCO) covering failed product batches, failing quality parameters (e.g., dissolution, particulate matter), and reporting labs.

---

## Suggested Use Cases

- Healthcare RAG and Chatbots: Ground LLMs on structured pharmacology data, avoiding hallucinations on brand names, salts, and indications.
- Generic Substitution Engines: Match brand names by `salt_composition` to find lower-cost bio-equivalent alternatives.
- Drug Interaction Validators: Check potential adverse interactions across multi-drug regimens.
- Health Economics and Pricing Analysis: Compare brand pricing trends across 986 manufacturers.

---

## Quick Start (Python)

```python
import json

with open("base_catalog.json", "r", encoding="utf-8") as f:
    catalog = json.load(f)

print(f"Loaded {len(catalog):,} medicine records.")

# Example: Filter by salt
paracetamol_drugs = [
    item for item in catalog 
    if "paracetamol" in item.get("_clean_salt", "")
]
print(f"Found {len(paracetamol_drugs)} Paracetamol products.")
```

---

## License and Disclaimer

- License: MIT
- Purpose: Research, education, and development of biomedical search and RAG systems.
- Disclaimer: For informational and computational research purposes only. Not intended as medical advice.