Adding ecg2age model to model zoo (#604)

* Added ECG2Age keras files

* added model files

* Added keras models for ECG2Age

* Create README.md

Author: rxrathod

.gitattributes

@@ -48,3 +48,4 @@ model_zoo/ECG_PheWAS/decoder_median.keras filter=lfs diff=lfs merge=lfs -text
 model_zoo/ECG_PheWAS/encoder_median.keras filter=lfs diff=lfs merge=lfs -text
 model_zoo/ECG_PheWAS/mgh_biosppy_median_60bpm_autoencoder_256d_v2022_05_21.keras filter=lfs diff=lfs merge=lfs -text
 model_zoo/ECG2HF/ecg_5000_hf_quintuplet_dropout_v2023_04_17.keras filter=lfs diff=lfs merge=lfs -text
+model_zoo/ECG2Age/*.keras filter=lfs diff=lfs merge=lfs -text

model_zoo/ECG2Age/.gitattributes

@@ -0,0 +1 @@
+.keras filter=lfs diff=lfs merge=lfs -text

model_zoo/ECG2Age/ECG2Age.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5c36946e19ed04c2781214cbd7e7df58fb08189bdb0a664ab888ecc8c8bc271
+size 687875020

model_zoo/ECG2Age/README.md

@@ -0,0 +1,101 @@
+# ECG2Age — Age Prediction from 12‑Lead ECG (Keras)
+
+A lightweight Keras/TensorFlow model that predicts chronological age from a standard 12‑lead ECG.
+
+## Table of Contents
+
+* [Model card](#model-card)
+
+  * [Inputs](#inputs)
+  * [Outputs](#outputs)
+  * [Performance](#performance)
+* [Getting started](#getting-started)
+
+  * [Environment](#environment)
+  * [Repository layout](#repository-layout)
+* [Model files](#model-files)
+* [Load & run inference](#load--run-inference)
+
+---
+
+## Model card
+
+### Inputs
+
+* **Modality**: 12‑lead resting ECG
+* **Expected shape**: `(batch, time, leads)` = `(B, 5000, 12)` by default
+
+  * Sampling rate: **500 Hz** (10 seconds ⇒ 5000 samples)
+  * If your data is `(B, 12, 5000)`, set `--channels_first` or transpose before feeding the model.
+* **Dtype / range**: `float32` normalized per‑lead (e.g., z‑score or min‑max).
+  Provide your normalization in `data/processing.py`.
+
+> Update the shape/sampling rate above if your dataset differs.
+
+### Outputs
+
+* **Target**: chronological age in **years** (scalar per record)
+* **Prediction head**: single linear unit with optional activation clipping (e.g., `ReLU` at 0 years)
+* **Loss**: MAE or Huber (configurable)
+
+### Performance
+
+Please benchmark on a **held‑out test set** (subject‑wise split). Example table with placeholders:
+
+| Metric      |    Test          |
+| ----------- | ---------------: |
+| R²          |     0.3056011687 |
+| Pearson r   |     0.5528120555 |
+
+Add a calibration plot and error vs age plot if possible (see `notebooks/metrics.ipynb`).
+
+---
+
+## Getting started
+
+### Environment
+
+```bash
+# Python ≥3.10 recommended
+python -m venv .venv
+source .venv/bin/activate
+pip install -U pip
+pip install -r requirements.txt
+# or
+pip install tensorflow==2.16.* keras==3.* numpy scipy scikit-learn matplotlib h5py
+```
+
+### Repository layout
+
+```
+.
+├── model_zoo/ECG2Age
+│   ├── ECG2Age.keras        # Keras SavedModel (via Keras 3)
+    ├── encoder_median.keras # Encoder part of ECG2Age model 
+    ├── decoder_instance_age.keras  # Decoder part of ECG2Age model
+    ├── merger.keras          #Encoder decoder merger intermediate model
+
+```
+
+---
+
+## Load & run inference
+
+### Python API
+
+```python
+import numpy as np
+import tensorflow as tf
+from keras import ops
+
+# Load model (Keras 3 format)
+model = tf.keras.models.load_model("model_zoo/ECG2Age/ECG2Age.keras", compile=False)
+
+# Example input: batch of 8 ECGs, 5000 samples, 12 leads
+x = np.random.randn(8, 5000, 12).astype("float32")
+# Optional: apply same normalization used during training
+# x = normalize_batch(x)
+
+pred_age = model.predict(x, verbose=0)  # shape (8, 1)
+print(pred_age.squeeze())
+```

model_zoo/ECG2Age/decoder_instance_age.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78a908c22c0775f85ce7e95b28c7a0956c192e21a794a79e310b99ebf5c216e8
+size 14568

model_zoo/ECG2Age/encoder_median.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02baca1ab7f14a6a38bce79ccbf6b8c41834a20234dc77e8c32d6e2d1ce009e9
+size 229079200

model_zoo/ECG2Age/merger.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aea1db12aeb1a73f1b7d8b2a00a8a3364541ea3f5152eb2abdc4459c8282b371
+size 229353750