my-mm-med-synth: Burmese Synthetic Medical Reviews Dataset

Dataset Summary

my-mm-med-synth dataset contains synthetic Burmese (Myanmar) medical reviews for sentiment classification, created by translating English medical review corpora into Burmese using both large language models (LLMs) and machine translation models (MTMs). It is designed to support research in low-resource Natural Language Processing (NLP), particularly in the healthcare domain where annotated Burmese datasets are scarce.

The dataset consists of two subsets translated from English medical review dataset introducted in "Medical Sentiment Analysis using Social Media: Towards building a Patient Assisted System (Yadav et al., LREC 2018)":

Medical Condition Dataset (MCD)

Labels:

• Recover denoted as label 1
• Exist denoted as label 0
• Deteriorate denoted as label 2

Medication Review Dataset (MRD)

Labels:

• Effective denoted as label 0
• Ineffective denoted as label 1
• Serious Adverse Effectdenoted as label 2

my-mm-med-synth dataset is introduced in the paper:

Classifying Burmese (Myanmar) Medical Reviews Using Synthetic Data Approach : https://ieeexplore.ieee.org/document/11459830

Dataset Structure

Each example include the following fields:

• tranlated: Burmese translated review
• translated_word: Burmese translated review segmented by word-level myWord tokenizer
• translated_word: Burmese translated review segmented by syllabel-level myWord tokenizer

Ethical Considerations

• The dataset contains synthetic translations only
• No real Burmese patient data is included
• Original English datasets must be obtained separately
• Synthetic text may contain translation errors or hallucinations

Tokenization Tool

This project uses the myWord Burmese tokenizer developed by Ye Kyaw Thu for preprocessing translated burmese text.

Repository: https://github.com/ye-kyaw-thu/myWord

If you use the preprocessing pipeline, please cite: "myWord: Syllable, Word and Phrase Segmenter for Burmese, Ye Kyaw Thu, Sept 2021, GitHub Link: https://github.com/ye-kyaw-thu/myWord"

Ethical Use

This dataset is intended for research purposes only and must not be used for clinical decision-making or deployed in real-world medical systems without proper validation.

If you use my-mm-med-synth dataset, please cite the following paper.

@inproceedings{thu2026burmese_medical_reviews, title={Classifying Burmese (Myanmar) Medical Reviews Using Synthetic Data Approach},

author={Thu, Hein Minn and Moe, Pyae Bhone and Ko Ko, Nyein Chan and Khine, Aye Hninn},

booktitle={ECTI DAMT & NCON},

year={2026} }