A generative adversarial optimization strategy for predicting counterfactual trajectories of grey matter atrophy
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
BACKGROUND AND OBJECTIVE: Counterfactual explanations offer valuable insights into the behavior of machine learning models by describing hypothetical scenarios that would lead to different outcomes. In the biomedical domain, such as neuroimaging for Multiple Sclerosis (MS), counterfactual reasoning has the potential to enhance understanding of disease mechanisms and treatment effects. However, generating anatomically plausible counterfactuals that generalize well to unseen data remains a major challenge. METHODS: We propose an optimization-based adversarial framework for generating realistic counterfactual trajectories of cortical grey matter (GM) thickness in MS patients. The method uses the gradients of a pre-trained MS classifier to guide the generation process towards a desired disease state while enforcing anatomical constraints and disentangling disease-relevant signals from confounding factors such as age and sex. RESULTS: Our approach successfully produces plausible counterfactual GM thickness maps that reflect known anatomical patterns of MS progression. The generated trajectories maintain consistency with biological structure and improve interpretability of model decisions. On held-out test data, our method achieves a classification AUC of 0.893 and demonstrates strong confounder preservation, with a Mean Absolute Deviation Error (MADE) of 8.72 years for age and 0.14 for sex, and a cosine distance of 0.11 when comparing original and counterfactual instances. The ability to alter the predicted disease state while preserving the confounding variables highlights the strong disentanglement capability of our model. These results confirm the method's effectiveness in generating realistic and anatomically coherent counterfactuals, outperforming state-of-the-art baselines across multiple metrics. CONCLUSIONS: This study introduces a novel counterfactual generation method that provides interpretable, anatomically grounded explanations of MS progression. The framework serves as a powerful tool for hypothesis generation and model validation in biomedical imaging studies, particularly where understanding disease mechanisms is crucial.
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.
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) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| 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 it