Back left table group

[rubywood]

Pull Request Template

Summary

Provide a short summary of the changes in this PR.

---

Data Modalities Covered

List the modalities included in this PR (tick all that apply):

• Image
• Text
• Signals
• Tabular
• Other

---

Dataset & DataLoader Design

• Describe the design decisions for your Dataset and DataLoader implementation.
• Explain how your design addresses scalability, flexibility, and modularity.
• Highlight any specific challenges and how you resolved them.

---

Handling Relationships Between Samples

• Explain how your DataLoader takes relationships across samples into account. For example: linking samples from different modalities that come from the same subject.
• Describe how missing modalities or unaligned samples are handled.

---

Sample Usage Instructions

Provide example code showing how to use your Dataset/DataLoader:

from load_data.dataset import CXRDataset, ECGDataset, MultimodalDataloader

cxr_dataset = CXRDataset(data_dir='path/to/cxr', ...)
ecg_dataset = ECGDataset(data_dir='path/to/ecg', ...)
dataset_list = [cxr_dataset, ecg_dataset]
loader = MultimodalDataloader(dataset_list, batch_size=8, shuffle=True)

for batch in loader:
    print(batch.keys())

Extensibility

Explain how this implementation can be extended to support additional modalities.

• What should be modified or subclassed?
• What conventions should be followed?

Checklist

• Code runs without errors
• Code is well-documented
• Tests are included and pass
• Follows project coding standards

Additional Notes

Any additional information or context that reviewers should be aware of.