pone.0348670.t025: Clinical Benchmark Datasets for Disease Prediction

pone.0348670.t025 is a collection of three benchmark clinical datasets used to evaluate a unified deep learning framework for disease prediction. The data includes the UCI Heart Disease dataset (303 samples, 13 features), the PIMA Indians Diabetes dataset (768 samples, 8 features), and a Parkinson's disease voice dataset (195 recordings, 22 features). It was authored by Vijay U. Rathod and last updated on 2026-05-08.

Use Cases

• Benchmarking disease prediction models based on the three heterogeneous clinical datasets described.
• Evaluating model performance on datasets with varying feature dimensionality, sample size, and class imbalance as mentioned in the description.
• Training and testing modular deep learning architectures, including MLP, CNN, and transformer-based ensembles, for clinical tasks.
• Studying the application of fold-safe feature selection and nested cross-validation for reliable performance evaluation in healthcare.

Strengths

• Includes three established benchmark datasets with specific sample sizes: 303, 768, and 195.
• Provides concrete performance metrics (AUC values) for different model architectures on the tasks, such as 0.8980 for heart disease prediction.
• Dataset is small (5.5 KB), facilitating quick download and initial exploration.

Limitations

• Row count for the combined collection is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The description focuses on a research framework; the actual data files may contain only the raw benchmark datasets without model outputs.

Provenance

Source

Benchmark datasets sourced from UCI and other repositories, aggregated by the author.

Collection Method

Likely compiled from public repositories for research evaluation.

Freshness

Last updated 2026-05-08 17:34:42; freshness should be verified.