SaddlePoint Mosaics

Bayesian latent class analysis of heterogeneous epidemiological cohorts, with multiple risks and informative censoring, and of data from clinical trials, with possibly inhomogeneous treatment associations. ​​Clinical outcomes may take the form of time-to-event variables (e.g. overall or disease-free survival), or ordinal class variables (e.g. treatment response). ​

 

A comprehensive analysis report includes: 

​​

  • Detailed statistical cohort analysis of covariates and outcomes

  • The Bayes-optimal latent substructure of the cohort: the most probable number and sizes of sub-classes, and their covariate associations and base hazard rates, for each active risk

  • Risk-specific and covariate-conditioned survival curves, decontaminated for the effects of informative censoring 

  • Comparisons with outcomes of standard regression methods 

  • Associations between covariates and sub-class membership

  • Retrospective sub-class membership probabilities for all samples in the cohort (for biomarker discovery)

  • Prospective outcome prediction for unseen data 

Downloads:

brief presentation of SaddlePoint Mosaics

functionality fact sheet of version 1.1.0

Software​

SaddlePoint Discriminant 

​​

Most tools for discriminant analysis use Maximum Likelihood (ML) or Maximum a Posteriori (MAP) approximations, and therefore require the number of samples to be much larger than the number of covariates. They fail for very high dimensional data, such as imaging or genomic ones. SaddlePoint Discriminant is based on exact analytical evaluation of Bayesian parameter integrals, as opposed to point estimates, and hence has no such limitations. 

 

Analysis outcomes include: 

  • LOOCV-based estimates of training and validation errors

  • Estimates of training and validation errors derived from separated training and validation sets

  • Full classification confusion tables, for training and validation sets

  • Fast probabilistic class membership predictions for unseen data

  • Statistical description of the cohort, in terms of class-conditioned covariate characteristics and Bayesian hyper-parameters

SaddlePoint Signature

A multi-core implementation of a pipeline  designed primarily for predictive multivariate medical data analytics in the regime of high-dimensional covariates and/or undersampling. 

 

 

 

 

 

 

 


 

Its main aims are:

  • Identifying the optimal covariate selection for predictive regression (without overfitting)

  • Ranking the covariates in the optimal set

  • Quantifying the outcome prediction performance of the optimal inferred model on unseen data via cross-validations

  • Constructing optimal predictive multivariate models

  • Generating formulas for optimised personal risk scores and treatment response scores


The pipeline implements different outcome data types, including time-to-event outcomes, ordinal class outcomes, and arbitrary real-valued ordinal outcomes, and includes functionality for data visualisation and for the generation of controlled synthetic data. It uses nested multiple iterative multivariate regressions, with bootstrapping and Bayesian probabilistic protocols and adaptive parameter priors. 

Downloads:

brief presentation of SaddlePoint Signature

functionality fact sheet of version 2.8.5

SaddlePoint Graphman

 

Generation of rigorously unbiased null models for directed and non-directed cellular signalling networks is vital for the correct interpretation of experimental observations, and the reliable identification of clinically significant modules. Many researchers apply ad-hoc randomisation algorithms, which have been shown to induce mobility-related biases that invalidate tests. 

 

 

Analysis outcomes of the seed network are compiled into a comprehensive report, including:

 

  • Node-specific quantifiers, e.g. degree distribution and degree-degree correlations. 

  • Distance related quantifiers, e.g. path length statistics. 

  • Modularity measures and closed path (loop) statistics. 

  • Eigenvalue spectra of adjacency and Laplacian matrices.

Hierarchically constrained null models are generated, obtained by MCMC randomisation with canonical move acceptance probabilities that target tailored maximum entropy network ensembles.

© 2018 by Saddle Point Science Ltd. Registered company, number 10221950. Address: 71 Oaks Avenue, Worcester Park, KT4 8XE, UK.