Performance of Individual Models for Diabetes Prediction and Management

5.5 KB Excel file containing performance metrics for a hybrid diabetes management system. The dataset was created by Muhammad Saleem and last updated on May 22, 2026. It validates a BiLSTM model against seven traditional ML and standard DL models using a fused PIMA and Type 2 Diabetes dataset.

Use Cases

• Benchmarking deep learning models against traditional machine learning models based on the comparative testing described.
• Developing integrated clinical tools for diabetes management based on the described hybrid system architecture.
• Analyzing model generalization performance based on the leakage-safe preprocessing protocol mentioned.
• Evaluating the trade-offs between prediction accuracy and personalized recommendation generation based on the system's two-component design.

Strengths

• Model performance validated with rigorous 10-fold cross-validation, achieving 84.02% accuracy, 87.89% precision, and 80.50% recall.
• Dataset was preprocessed following a leakage-safe protocol to increase generalizability.
• Comparative analysis includes seven traditional ML models and standard DL models against the proposed BiLSTM model.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Muhammad Saleem via figshare

Collection Method

Fused dataset (PIMA and Type 2 Diabetes) preprocessed with a leakage-safe protocol.

Time Range

null

Freshness

Last updated 2026-05-22 17:42:14; freshness should be verified.

Geography

null