Features
Advanced statistical capabilities for complex biomedical data
Saddle Point Science combines advanced statistical methodology with automated analytical workflows to help researchers extract reliable insights from challenging biomedical datasets.
From high-dimensional prediction and patient subgroup discovery to federated analysis, SPS provides tools designed around the challenges of modern biomedical research.
Six core capabilities
-
Analyse: Apply advanced statistical modelling to high-dimensional biomedical data.
-
Predict: Generate personalised risk and treatment response predictions.
-
Discover: Identify hidden patient subgroups and differences between populations.
-
Stratify: Characterise patient populations and assess potential responder groups.
-
Federated collaborations: Learn from distributed datasets without requiring patient-level data to be pooled.
-
Validate: Assess robustness and reliability through integrated validation approaches.
What we can do and why it matters
Analyse complex biomedical datasets while accounting for the statistical challenges associated with high dimensionality and limited sample sizes.
1. Overfitting & Model Reliability
Features:
-
Probabilistic variable selection
-
Bias correction
-
Statistical regularisation
-
Analytical parameter optimisation
-
Correction of nuisance parameters
-
Informative covariate missingness
-
No data sacrifice for hyperparameter optimisation
-
Overfitting assessment
2. Patient-Level & Subgroup Analysis
Saddle Point Science can help identify and characterise differences between individual patients and groups within a cohort.
Features:
-
Personalized outcome risk prediction
-
Personalised treatment response prediction
-
Latent subgroup detection and characterization
-
Cohort stratification with personalized patient subgroup probabilities
-
Biomarker exploration
3. Distributed Data Analysis
Learn from multiple datasets without pooling patient-level data
Features:
-
Local statistical analysis
-
Federated statistical inference
-
Distributed datasets
-
Multi-centre analysis
-
Statistical information sharing instead of patient-level data
-
Combined inference across centres
4. Validation & Quality Control
Built-in checks for more reliable results
Features:
-
External validation
-
Evaluate predictions using additional datasets where available.
-
-
Simulated data
-
Test model behaviour under controlled conditions.
-
-
Shadow analysis
-
Use outcome-randomised data to investigate potential false-positive findings.
-
-
Uncertainty assessment
-
Quantify uncertainty around predictions and subgroup membership.
-
-
Consistency checks
-
Check analytical outputs and model behaviour.
-
-
Reproducible analysis
-
Use structured workflows to support repeatable analyses.
-
5. Automated Analysis
From data to results with fewer manual steps
Features:
-
Automated analytical pipelines
-
Automated model selection
-
Automated parameter estimation
-
Automated validation procedures
-
Automated reporting
-
Reproducible workflows
-
Structured outputs
-
Reduced manual analysis
Research outcomes supported
What can you analyse?
Supported data / analysis types
Designed for a broad range of biomedical datasets
Data characteristics
-
High-dimensional
-
Limited sample size
-
Heterogeneous populations
-
Missing information
-
Distributed datasets
Outcome types
-
Time-to-event
-
Ordinal classes
-
Continuous / real-valued
-
Multiple risks
Features by product
Saddle Point solutions
Built on advanced mathematical and statistical approaches, including:
Do you have a challenging biomedical dataset?
Our team can help you determine whether our statistical solutions are suitable for your research question and dataset, and whether we may be able to help you extend your analyses to extract more information from your datasets.