Early-Stage Diabetes Risk Prediction Data from Patient Questionnaires in Bangladesh

A tabular dataset for predicting early-stage diabetes risk, collected via direct questionnaires from patients at Sylhet Diabetes Hospital in Sylhet, Bangladesh. The data includes attributes such as age, sex, and symptoms like polyuria and polydipsia, and was approved by a doctor. It is associated with a research paper titled 'Likelihood prediction of diabetes at early stage using data mining techniques'.

Use Cases

• Train a binary classifier to predict diabetes risk based on patient-reported symptoms like polyuria and polydipsia.
• Analyze the correlation between demographic factors (age, sex) and early diabetes indicators.
• Build a feature importance model to identify which symptoms (e.g., sudden weight loss, visual blurring) are most predictive of a positive diagnosis.
• Develop a prototype screening tool for non-invasive, questionnaire-based early diabetes detection.

Strengths

• Data collection was approved by a doctor, suggesting clinical relevance.
• Includes 16 distinct attributes covering demographic, symptomatic, and diagnostic information.
• The dataset is associated with a peer-reviewed research paper published by Springer in 2020.

Limitations

• Row count is unknown, which may limit suitability assessment for model training.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect geographic bias inherent to its collection from a single hospital in Bangladesh.

Provenance

Source

Sylhet Diabetes Hospital in Sylhet, Bangladesh.

Collection Method

Collected using direct questionnaires from patients.

Geography

Sylhet, Bangladesh.