A Clinically Curated Haematology Dataset from Routine Laboratory Practice in Bangladesh

This dataset presents a structured collection of 12,618 de-identified haematology laboratory records obtained from routine clinical testing at Aalok Healthcare, Dhaka, Bangladesh, between April and June 2026. It provides a clinically curated resource for investigating haematological variation, exploring relationships among laboratory measurements, and supporting reproducible computational analysis of routine blood-test data.

The dataset contains 23 numerical laboratory variables and one categorical screening-label column, incorporating complete blood count (CBC) parameters, differential white blood cell measurements, red blood cell indices, platelet indices, and erythrocyte sedimentation rate (ESR). CBC measurements were obtained using the Mindray BC-5130 haematology analyser, with ESR values extracted from the corresponding laboratory reports.

Records are organised into five clinician-reviewed haematological screening categories:

1. Microcytic Hypochromic Anaemia

2. Normal / Non-specific

3. Mild-Moderate Leukocytosis

4. Thrombocytopaenia

5. Severe Leukocytosis Flag

The resource preserves the original numerical measurements and recorded data characteristics, enabling investigations of feature distributions, inter-variable associations, screening-category patterns, and missing-data behaviour. Its structured, labelled format also supports the development and evaluation of reproducible statistical and machine-learning approaches for haematological screening research.

The deposit includes the de-identified CSV dataset, a variable-level data dictionary, and a README describing the dataset structure and usage. The accompanying Data Descriptor manuscript presents the collection methodology, dataset characteristics, and technical validation.