Explainable Feature Engineering in Health Data Science: Empirical Comparison of ChatGPT-4o and Classical Machine Learning Methods
Bibliographic record
Abstract
Machine learning (ML) is demonstrating remarkable success in various healthcare applications. The success of ML in healthcare is inherently linked to the rigorous process of feature engineering and feature selection, which truly forms the backbone of ML model development. This study investigates the role of a well-known large language model (LLM), the ChatGPT-4o, in feature selection and classification processes for healthcare data, focusing on the explainability of ML. The performance of ChatGPT-4o is evaluated and compared to traditional ML methods—such as information gain (IG), correlation-based feature selection (CFS), and principal component analysis (PCA) for identifying relevant features in predictive modeling. This comparison is conducted using two widely recognized healthcare datasets, SEER and NSQIP. After evaluating the features selected by classical ML methods and LLMs through expert review, the results indicate that while ChatGPT-4o aligns closely with expert evaluations and effectively provides contextual information on healthcare datasets, traditional ML methods such as IG, CFS, and PCA outperform in systematic feature ranking due to their structured and data-driven nature. Furthermore, anonymization did not significantly affect the feature selection process, highlighting the robustness of ChatGPT-4o under privacy-preserving conditions. ChatGPT-4o's strength lies in complementing these methods by providing interpretability and facilitating exploratory analysis, rather than serving as a standalone solution for precise feature ranking.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.023 | 0.132 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.003 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.002 | 0.004 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".