Clinical Disease Prediction Performance Benchmarks Across Three Datasets

Vijay U. Rathod's study, last updated in 2026, provides performance benchmarks for a unified deep learning framework on three clinical datasets. The framework was evaluated on the UCI Heart Disease dataset (303 samples), the PIMA Indians Diabetes dataset (768 samples), and a Parkinson's disease voice dataset (195 recordings). Results include AUC scores for various model ensembles, such as 0.8980 for heart disease prediction.

Use Cases

• Benchmarking deep learning models against classical ML baselines based on the reported AUC scores for heart disease, diabetes, and Parkinson's prediction.
• Developing a dataset-aware model selection framework based on described dataset characteristics like feature dimensionality and sample size.
• Evaluating ensemble strategies (e.g., FT-Transformer + autoencoder) for clinical tabular data based on the described modular architecture.
• Implementing robust evaluation pipelines using concepts like fold-safe feature selection and nested cross-validation as described in the study.

Strengths

• Provides specific performance metrics (AUC) for model ensembles, such as 0.8980 for heart disease and 0.8451 for diabetes.
• Benchmarks on three established clinical datasets with known sample sizes (303, 768, and 195).
• Framework incorporates specific techniques like robust preprocessing and nested cross-validation for reliable evaluation.
• Dataset is small (5.5 KB) and in a common format (XLS) for quick inspection.

Limitations

• Row count and column-level documentation are unknown, limiting suitability assessment for specific modeling tasks.
• The dataset appears to contain benchmark results, not the raw clinical data used to generate them.
• Last updated 2026-05-08 17:34:33; freshness should be verified as this is a future date.

Provenance

Source

figshare, author Vijay U. Rathod

Collection Method

Likely contains aggregated experimental results from a study proposing a deep learning framework.

Freshness

Last updated 2026-05-08 17:34:33