Quickstart

Install heartKIT

We provide several installation methods including pip, uv, and Docker. Install heartKIT via pip/uv for the latest stable release or by cloning the GitHub repo for the most up-to-date. Additionally, a VSCode Dev Container is available and defined in ./.devcontainer to run in an isolated Docker environment.

Git clone
PyPI install

Clone the repository if you are interested in contributing to the development or wish to experiment with the latest source code. After cloning, navigate into the directory and install the package. In this mode, uv is recommended.

# Clone the repository
git clone https://github.com/AmbiqAI/heartkit.git

# Navigate to the cloned directory
cd heartkit

# Install the package in editable mode for development
uv sync

When using editable mode via uv, be sure to activate the python environment: source .venv/bin/activate.
On Windows using Powershell, use .venv\Scripts\activate.

Install the heartKIT package using pip or uv. Visit the Python Package Index (PyPI) for more details on the package: https://pypi.org/project/heartkit/

# Install with pip
pip install heartkit

Or, if you prefer to use uv, you can install the package with the following command:

# Install with uv
uv add heartkit

Alternatively, you can install the latest development version directly from the GitHub repository. Make sure to have the Git command-line tool installed on your system. The @main command installs the main branch and may be modified to another branch, i.e. @canary.

pip install git+https://github.com/AmbiqAI/heartkit.git@main

Or, using uv:

uv add git+https://github.com/AmbiqAI/heartkit.git@main

Requirements

Check the project’s pyproject.toml file for a list of up-to-date Python dependencies. Note that the installation methods above install all required dependencies. The following are optional dependencies only needed when running demo command using Ambiq’s evaluation board (EVB) backend:

Once installed, heartKIT can be used as either a CLI-based tool or as a Python package to perform advanced experimentation.

Use heartKIT with CLI

The heartKIT command line interface (CLI) allows for simple single-line commands to download datasets, train models, evaluate performance, and export models. The CLI requires no customization or Python code. You can simply run all the built-in tasks from the terminal with the heartkit command. Check out the CLI Guide to learn more about available options.

Heartkit commands use the following syntax:

heartkit --mode [MODE] --task [TASK] --config [CONFIG]

Or using short flags:

heartkit -m [MODE] -t [TASK] -c [CONFIG]

Where:

MODE is one of download, train, evaluate, export, or demo
TASK is one of segmentation, rhythm, beat, or denoise
CONFIG is configuration as JSON content or file path

Download datasets specified in the configuration file.

heartkit -m download -c ./download-datasets.json

Train a rhythm model using the supplied configuration file.

heartkit -m train -t rhythm -c ./configuration.json

Evaluate the trained rhythm model using the supplied configuration file.

heartkit -m evaluate -t rhythm  -c ./configuration.json

Run demo on trained rhythm model using the supplied configuration file.

heartkit -m demo -t rhythm -c ./configuration.json

Use heartKIT with Python

The heartKIT Python package allows for more fine-grained control and customization. You can use the package to train, evaluate, and deploy models for a variety of tasks. You can create custom datasets, models, and tasks and register them with corresponding factories and use them like built-in tasks.

For example, you can create a custom task, train it, evaluate its performance on a validation set, and even export a quantized TensorFlow Lite model for deployment. Check out the Python Guide to learn more about using heartKIT as a Python package.

import heartkit as hk

params = hk.HKTaskParams(...)

task = hk.TaskFactory.get("rhythm")

task.download(params)  # Download dataset(s)

task.train(params)  # Train the model

task.evaluate(params)  # Evaluate the model

task.export(params)  # Export to TFLite

Configuration parameters

quickstart-13.py82 lines · PYTHONDownload

Show full exampleCollapse example

hk.HKTaskParams(
    name="arr-2-eff-sm",
    project="hk-rhythm-2",
    job_dir="./results/arr-2-eff-sm",
    verbose=2,
    datasets=[hk.NamedParams(
        name="ptbxl",
        params=dict(
            path="./datasets/ptbxl"
        )
    )],
    num_classes=2,
    class_map={
        "0": 0,
        "7": 1,
        "8": 1
    },
    class_names=[
        "NORMAL", "AFIB/AFL"
    ],
    class_weights="balanced",
    sampling_rate=100,
    frame_size=512,
    samples_per_patient=[10, 10],
    val_samples_per_patient=[5, 5],
    test_samples_per_patient=[5, 5],
    val_patients=0.20,
    val_size=20000,
    test_size=20000,
    batch_size=256,
    buffer_size=20000,
    epochs=100,
    steps_per_epoch=50,
    val_metric="loss",
    lr_rate=1e-3,
    lr_cycles=1,
    threshold=0.75,
    val_metric_threshold=0.98,
    tflm_var_name="g_rhythm_model",
    tflm_file="rhythm_model_buffer.h",
    backend="pc",
    demo_size=896,
    display_report=True,
    quantization=hk.QuantizationParams(
        qat=False,
        format="INT8",
        io_type="int8",
        conversion="CONCRETE",
        debug=False
    ),
    preprocesses=[
        hk.NamedParams(
            name="layer_norm",
            params=dict(
                epsilon=0.01,
                name="znorm"
            )
        )
    ],
    augmentations=[
    ],
    model_file="model.keras",
    use_logits=False,
    architecture=hk.NamedParams(
        name="efficientnetv2",
        params=dict(
            input_filters=16,
            input_kernel_size=[1, 9],
            input_strides=[1, 2],
            blocks=[
                {"filters": 24, "depth": 2, "kernel_size": [1, 9], "strides": [1, 2], "ex_ratio": 1,  "se_ratio": 2},
                {"filters": 32, "depth": 2, "kernel_size": [1, 9], "strides": [1, 2], "ex_ratio": 1,  "se_ratio": 2},
                {"filters": 40, "depth": 2, "kernel_size": [1, 9], "strides": [1, 2], "ex_ratio": 1,  "se_ratio": 2},
                {"filters": 48, "depth": 1, "kernel_size": [1, 9], "strides": [1, 2], "ex_ratio": 1,  "se_ratio": 2}
            ],
            output_filters=0,
            include_top=True,
            use_logits=True
        )
    }
)


hk.HKTaskParams(
    name="arr-2-eff-sm",
    project="hk-rhythm-2",
    job_dir="./results/arr-2-eff-sm",
    verbose=2,
    datasets=[hk.NamedParams(
        name="ptbxl",

… 74 more lines