Identifying intervention strategies from machine learning models with COALA: a counterfactual optimization framework
Bibliographic record
Abstract
Abstract Motivation Machine learning (ML) models have become increasingly complex, often functioning as black boxes that limit our understanding of the features contributing to ML predictions. Common explainable AI (XAI) methods such as SHapley Additive exPlanations (SHAP) focus on feature importance but fall short in identifying interactions among features and informing targeted, personalized interventions. Counterfactuals are hypothetical events where specific variables are altered to cause a change in outcome. These causal statements can be applied to AI models to identify actionable interventions for different subjects in a population. Results We propose the framework Counterfactual Optimization for Actionable interpretabiLity in AI (COALA). COALA interprets models by identifying optimal counterfactuals for each subject, which are defined as actionable changes that lead to the most positive change in predicted outcome. When applied to a gradient boosted tree model trained on the National Health and Nutrition Examination Survey (NHANES) dataset, COALA identifies different profiles of optimal counterfactuals across subjects. Features that remain constrained were able to predict the optimal counterfactual changes for a subject at 85.4%, revealing specific features that drive what a subject’s optimal counterfactual is. Availability and Implementation Code for COALA implementation, synthetic data, models trained on synthetic data, and code to replicate results and figures are available at https://github.com/brt-solo/COALA . The NHANES 2017–2018 dataset is publicly available from the National Center for Health Statistics (NCHS). The Framingham Heart Study dataset used in this study is a publicly available, Framingham-derived dataset distributed through the Massachusetts Institute of Technology OpenCourseWare (MIT OCW) repository.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.003 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".