jmstate 0.2.0

Joint modeling with automatic differentiation

📦 jmstate

jmstate is a Python package for multi-state nonlinear joint modeling.
It leverages PyTorch for automatic differentiation and vectorized computation, making it efficient and scalable. The package provides a flexible framework where you can use neural networks as regression and link functions, while still offering simpler built-in options like parametric baseline hazards.

With jmstate, you can model longitudinal data jointly with multi-state transitions (e.g. health progression), capture nonlinear effects, and perform inference in complex real-world settings.

---

✨ Features

• Multi-State Joint Modeling
  Supports subjects moving through multiple states with transition intensities that depend on longitudinal trajectories and covariates.

• Nonlinear Flexibility
  Use neural networks (or any PyTorch model) as regression or link functions.

• Built-in Tools
  Includes default baseline hazards, regression, link functions, and analysis utilities.

• Automatic Differentiation & GPU Support
  Powered by PyTorch for efficient gradient computation and vectorization.

• Analysis & Visualization
  Tools for state occupancy probabilities, hazard estimation, and residual diagnostics.

---

🚀 Installation

pip install jmstate

---

📖 Learn More

For tutorials, API reference, visit the official site:
👉 jmstate Documentation