Add DynamicSurvivalRNN model and MIMIC-IV Demo example

[Pinkgirl16]

Contributor: Nishi Uppuluri
NetID: nu4

Type of contribution: Model

Original paper:
Dynamic Survival Analysis for Early Event Prediction
https://arxiv.org/abs/2403.12818

High-level description:
This PR adds DynamicSurvivalRNN, a discrete-time survival model for early event prediction, inspired by the Dynamic Survival Analysis paper. The model predicts per-bin hazards over a fixed future horizon and converts them into event-within-horizon risk through the cumulative failure function. The contribution includes synthetic unit tests and a runnable example on MIMIC-IV Demo ICU mortality. The example uses the first 48 hours of ICU charted measurements to predict death within 7 days of ICU admission and includes a small hidden-dimension ablation (hidden_dim=32 vs 64). The example script runs successfully end to end and produces an ablation summary, though results are noisy because the demo dataset is very small.

Files to review:

1. pyhealth/models/dynamic_survival_rnn.py
2. tests/test_dynamic_survival_rnn.py
3. examples/mimiciv_demo_mortality_dynamic_survival_rnn.py
4. docs/api/models/pyhealth.models.dynamic_survival_rnn.rst
5. pyhealth/models/__init__.py
6. docs/api/models.rst

[Jathurshan0330]

Good PR

[github-actions]

This PR has been automatically marked as stale because it has not had recent activity. It will be closed in 7 days if no further activity occurs.