Symile Mimic Eval

---
name: symile-mimic-eval
description: Tests clinical cross-modal prediction by evaluating whether ECG and blood lab measurements can jointly predict a subsequent chest X-ray in a zero-shot retrieval setting. It probes the model's ability to learn from incomplete training data and generalize to full modality combinations. Use when the user wants to benchmark on Symile-MIMIC, or asks about evaluating this task. Reports mean accuracy.
metadata:
  skill_kind: dataset_eval
  source_arxiv: 2411.01053
  bibtex_key: saporta2024symile
  confidence: high
---

# symile-mimic-eval

> Contrasting with Symile: Simple Model-Agnostic Representation Learning for Unlimited Modalities — Saporta et al. (2024) (arXiv:2411.01053, 2024)

## What this evaluates

Tests clinical cross-modal prediction by evaluating whether ECG and blood lab measurements can jointly predict a subsequent chest X-ray in a zero-shot retrieval setting. It probes the model's ability to learn from incomplete training data and generalize to full modality combinations.

## Datasets

- **Symile-MIMIC** — total 11622; splits: train/val (11041), test (581); repo https://github.com/rajesh-lab/symile

## Metrics

- `mean accuracy` **(primary)** — range: [0, 1]
  - Top-1 retrieval accuracy among 10 candidates (1 positive CXR, 9 sampled negatives) for each query. Reported as mean across 10 bootstrap samples of the test set.

## Input / output format

**Input**: Paired electrocardiogram (ECG) reading and blood laboratory measurements (50 most common labs) collected within 24 hours of hospital admission.

**Output**: Predicted chest X-ray (CXR) image from the candidate set.

## Scoring recipe

```python
candidates = [true_cxr] + sample_9_negatives(test_set_cxr)
for each query in test_set:
    scores = model.compute_similarity(query.ecg, query.labs, candidates)
    pred = argmax(scores)
    if pred == query.true_cxr:
        correct += 1
return correct / len(test_set)
```

## Common pitfalls

- Assuming full-dataset negative sampling; the evaluation explicitly uses a 10-candidate set (1 positive + 9 sampled negatives) per query to mitigate overfitting and match clinical retrieval constraints.
- Overlooking the temporal constraint; CXRs must be taken 24-72 hours post-admission, while ECGs/labs are within 24 hours of admission, which affects clinical validity and evaluation setup.

## Evidence (verbatim from paper)

> In [Figure 5]b, we report mean accuracy for Symile and CLIP over 10 bootstrap samples of the test set. While both models surpass random chance (0.1), Symile achieves an average accuracy of $0.435\pm 0.007$ (SE), outperforming CLIP’s $0.387\pm 0.003$ (SE).

## Citation

```bibtex
@misc{saporta2024symile,
  title={Contrasting with Symile: Simple Model-Agnostic Representation Learning for Unlimited Modalities},
  author={Saporta et al. (2024)},
  year={2024},
  note={arXiv:2411.01053}
}
```

- arXiv: 2411.01053