ekacare/indian_drug_mcqa

Indian Drug MCQA Dataset

A multiple-choice question answering dataset for evaluating knowledge of Indian branded medications — specifically the ability to identify generic names / salt compositions from brand names.

Dataset Description

This dataset contains 1512 questions covering Indian pharmaceutical brand names across 20+ therapeutic classes. Each question presents a branded medication and asks the model to identify its correct generic composition from 3-5 options. Includes variant questions for hard-difficulty drugs using different distractor strategies and phrasing styles to increase evaluation robustness.

Use Case

Evaluate medical language models on their knowledge of the Indian pharmaceutical market, including:

• Brand name → generic name mapping
• Salt composition identification
• Therapeutic class awareness

Dataset Structure

Column	Type	Description
question_id	string	Unique identifier (e.g., "Q001")
system_prompt	string	Instruction prompt for the model
question	string	The MCQ question stem
option_a	string	Option A
option_b	string	Option B
option_c	string	Option C
option_d	string	Option D
option_e	string	Option E (may be empty for 4-option questions)
correct_answer	string	Correct answer letter (A-E)
source_ekaid	string	Source medication identifier
medication_name	string	Branded medication name
generic_name	string	Generic name / salt composition
manufacturer_name	string	Manufacturer
therapeutic_class	string	Therapeutic category
action_class_name	string	Pharmacological action class
stem_format	string	Question stem format (e.g., "brand_only", "brand_form")
phrasing_style	string	Phrasing style (e.g., "pharmacist", "clinical", "direct")
distractor_strategy	string	How distractors were chosen (e.g., "close_combo")
difficulty_tier	string	Difficulty level ("easy", "medium", "hard")
num_options	int	Number of answer options (4 or 5)

Usage

With KARMA Evaluation Framework

karma eval \
  --model "Qwen/Qwen3-0.6B" \
  --datasets "ekacare/indian_drug_mcqa" \
  --format table

Direct Loading

from datasets import load_dataset

ds = load_dataset("ekacare/indian_drug_mcqa", split="test")

# Example question
print(ds[0]["question"])
print(ds[0]["correct_answer"])

Metrics

Recommended evaluation metric: Exact Match on the predicted answer letter (A-E).

Citation

If you use this dataset, please cite:

@dataset{indian_drug_mcqa_2025,
  title={Indian Drug MCQA Dataset},
  author={Eka.Care},
  year={2025},
  publisher={HuggingFace},
  url={https://huggingface.co/datasets/ekacare/indian_drug_mcqa}
}

License

This dataset is released under the MIT License.