Skip to main content

Leaspy is a software package for the statistical analysis of longitudinal data.

Project description

pipeline status documentation status code coverage PyPI - license PyPI - version PyPI - downloads PyPI - versions

Leaspy - LEArning Spatiotemporal Patterns in Python

Leaspy is a software package for the statistical analysis of longitudinal data, particularly medical data that comes in a form of repeated observations of patients at different time-points.

Get started Leaspy


  1. Leaspy requires Python >= 3.7, < 3.10

  2. Create a dedicated environment (optional):

    Using conda

    conda create --name leaspy python=3.7
    conda activate leaspy

    Or using pyenv

    pyenv virtualenv leaspy
    pyenv local leaspy
  3. Install leaspy: pip install leaspy

It will automatically install all needed dependencies.


Available online at

Examples & Tutorials

The example/start/ folder contains a starting point if you want to launch your first scripts and notebook with the Leaspy package. You can find additional walkthroughs in:

  • this series of online tutorials from 2020
  • this Medium post of 2019 (warning: the plotter and the individual parameters described there have been deprecated since then)


Leaspy is a software package for the statistical analysis of longitudinal data, particularly medical data that comes in a form of repeated observations of patients at different time-points. Considering these series of short-term data, the software aims at :

  • recombining them to reconstruct the long-term spatio-temporal trajectory of evolution
  • positioning each patient observations relatively to the group-average timeline, in term of both temporal differences (time shift and acceleration factor) and spatial differences (different sequences of events, spatial pattern of progression, ...)
  • quantifying impact of cofactors (gender, genetic mutation, environmental factors, ...) on the evolution of the signal
  • imputing missing values
  • predicting future observations
  • simulating virtual patients to un-bias the initial cohort or mimics its characteristics

The software package can be used with scalar multivariate data whose progression can be modelled by a logistic shape, an exponential decay or a linear progression. The simplest type of data handled by the software are scalar data: they correspond to one (univariate) or multiple (multivariate) measurement(s) per patient observation. This includes, for instance, clinical scores, cognitive assessments, physiological measurements (e.g. blood markers, radioactive markers) but also imaging-derived data that are rescaled, for instance, between 0 and 1 to describe a logistic progression.

Main features

  • fit : determine the population parameters that describe the disease progression at the population level
  • personalize : determine the individual parameters that characterize the individual scenario of biomarker progression
  • estimate : evaluate the biomarker values of a patient at any age, either for missing value imputation or future prediction
  • simulate : generate synthetic data from the model

Further information

More detailed explanations about the models themselves and about the estimation procedure can be found in the following articles :

  • Mathematical framework: A Bayesian mixed-effects model to learn trajectories of changes from repeated manifold-valued observations. Jean-Baptiste Schiratti, Stéphanie Allassonnière, Olivier Colliot, and Stanley Durrleman. The Journal of Machine Learning Research, 18:1–33, December 2017. Open Access.
  • Application to imaging data: Statistical learning of spatiotemporal patterns from longitudinal manifold-valued networks. I. Koval, J.-B. Schiratti, A. Routier, M. Bacci, O. Colliot, S. Allassonnière and S. Durrleman. MICCAI, September 2017. Open Access
  • Application to imaging data: Spatiotemporal Propagation of the Cortical Atrophy: Population and Individual Patterns. Igor Koval, Jean-Baptiste Schiratti, Alexandre Routier, Michael Bacci, Olivier Colliot, Stéphanie Allassonnière, and Stanley Durrleman. Front Neurol. 2018 May 4;9:235. Open Access
  • Application to data with missing values: Learning disease progression models with longitudinal data and missing values. R. Couronne, M. Vidailhet, JC. Corvol, S. Lehéricy, S. Durrleman. ISBI, April 2019. Open Access
  • Intensive application for Alzheimer's Disease progression: AD Course Map charts Alzheimer's disease progression, I. Koval, A. Bone, M. Louis, S. Bottani, A. Marcoux, J. Samper-Gonzalez, N. Burgos, B. Charlier, A. Bertrand, S. Epelbaum, O. Colliot, S. Allassonniere & S. Durrleman, Scientific Reports, 2021. 11(1):1-16 Open Access
  • website related to the application of the model for Alzheimer's disease
  • Disease Course Mapping webpage by Igor Koval


The package is distributed under the BSD 3-Clause license.


The development of this software has been supported by the European Union H2020 program (project EuroPOND, grant number 666992, project HBP SGA1 grant number 720270), by the European Research Council (to Stanley Durrleman project LEASP, grant number 678304) and by the ICM Big Brain Theory Program (project DYNAMO).



Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

leaspy-1.3.1.tar.gz (556.3 kB view hashes)

Uploaded source

Built Distribution

leaspy-1.3.1-py3-none-any.whl (597.0 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page