Advanced Statistics & Signal Processing
Statistical modeling, signal analysis, and reproducible workflows, tailored to deliver publication-ready results across medical, engineering, and research domains.
Dipl.-Ing. Dr.techn. Thomas Suppan · TU Graz · Award of Excellence (Austria)
Services and Expertise
Statistics
Statistical modeling, inference, and uncertainty analysis.
Signal Processing
Analysis and modeling of technical and physiological signals.
Machine Learning
Data-driven modeling with an emphasis on transparent validation.
Scientific Support
Methodological support for publications and research.
View all services →
Methodological Approach
TS-Analytics is currently in preparation and focuses on transparent and well-founded methods for the analysis of complex medical and technical data, with a strong emphasis on reproducibility and scientific validity.
Reproducible Workflows
Analyses are designed around reproducible workflows that support traceability, validation, and reliable reuse across studies and projects.
Statistical Modeling
Methods are selected based on the data structure and study context, with explicit modeling of uncertainty and underlying assumptions.
Domain-Aware Signal Analysis
Signal processing approaches are intended to be adapted to the physical origin of the data, supporting meaningful interpretation beyond purely numerical results.
Publication-Ready Results
Results are prepared in line with scientific standards, including clear documentation, transparent reporting, and figures suitable for publications.
Selected Publications & Projects
Selected scientific publications and selected projects illustrating methodological work across statistics, signal processing, and applied data analysis.
Medical & Biostatistics
Engineering & Measurement Science
Feasibility and diagnostic agreement of Apple Watch and KardiaMobile electrocardiograms compared with standard 12-lead ECG in children, S. Kurath-Koller et al., Scientific Reports, 2025
Non-parametric Bland–Altman agreement analysis was applied to quantify how closely mobile ECG measurements matched the reference 12-lead ECG without relying on normality assumptions. This robust approach provided clinically interpretable agreement metrics and highlighted parameter-specific biases and variability, supporting evidence-based validation of wearable diagnostics in pediatric populations.
Bayesian estimation of discrete system states from high-dimensional data
Discrete system states are estimated from high-dimensional measurements using a probabilistic modeling framework. Dimensionality reduction via principal component analysis is combined with Gaussian mixture model–based likelihoods and sequential Bayesian inference using a Hidden Markov Model.
Brain oxygenation monitoring during neonatal stabilization and resuscitation and its potential for improving preterm infant outcomes: a systematic review and meta-analysis with Bayesian analysis, M. Bruckner et al., European Journal of Pediatrics, 2025
A Bayesian meta-analysis was conducted, pooling the included trials in a hierarchical model to estimate the treatment effect. Results were reported as posterior effect distributions and probabilities of clinically meaningful benefit, providing direct, decision-oriented evidence rather than relying on p-values alone.
Thermal Drifts of Capacitive Flow Meters: Analysis of Effects and Model-Based Compensation, T. Suppan et al., IEEE Transactions on Instrumetation and Measurement, vol. 70, 2021
Thermal-drift mechanisms in capacitive flow sensing are analyzed and translated into a physics-informed measurement model. Building on this, a Bayesian joint-estimation framework is developed to infer both primary flow parameters and drift-related nuisance terms, separating the signal of interest from temperature effects.
Nonlinear Trend Filtering for Noisy Signals with Sharp Transients
A nonlinear trend estimation method is developed for noisy signals exhibiting both sharp transients and slowly varying segments. The trend is obtained by minimizing an ℓ₁-regularized second-order difference using a majorization–minimization approach, resulting in an efficient iterative scheme. The method enables robust separation of underlying signal structure from noise while preserving abrupt changes that are typically smoothed out by classical filtering techniques.
Medical & Biostatistics
Feasibility and diagnostic agreement of Apple Watch and KardiaMobile electrocardiograms compared with standard 12-lead ECG in children, S. Kurath-Koller et al., Scientific Reports, 2025
Non-parametric Bland–Altman agreement analysis was applied to quantify how closely mobile ECG measurements matched the reference 12-lead ECG without relying on normality assumptions. This robust approach provided clinically interpretable agreement metrics and highlighted parameter-specific biases and variability, supporting evidence-based validation of wearable diagnostics in pediatric populations.
Brain oxygenation monitoring during neonatal stabilization and resuscitation and its potential for improving preterm infant outcomes: a systematic review and meta-analysis with Bayesian analysis, M. Bruckner et al., European Journal of Pediatrics, 2025
A Bayesian meta-analysis was conducted, pooling the included trials in a hierarchical model to estimate the treatment effect. Results were reported as posterior effect distributions and probabilities of clinically meaningful benefit, providing direct, decision-oriented evidence rather than relying on p-values alone.
Engineering & Measurement Science
Bayesian estimation of discrete system states from high-dimensional data
Discrete system states are estimated from high-dimensional measurements using a probabilistic modeling framework. Dimensionality reduction via principal component analysis is combined with Gaussian mixture model–based likelihoods and sequential Bayesian inference using a Hidden Markov Model.
Thermal Drifts of Capacitive Flow Meters: Analysis of Effects and Model-Based Compensation, T. Suppan et al., IEEE Transactions on Instrumetation and Measurement, vol. 70, 2021
Thermal-drift mechanisms in capacitive flow sensing are analyzed and translated into a physics-informed measurement model. Building on this, a Bayesian joint-estimation framework is developed to infer both primary flow parameters and drift-related nuisance terms, separating the signal of interest from temperature effects.
Nonlinear Trend Filtering for Noisy Signals with Sharp Transients
A nonlinear trend estimation method is developed for noisy signals exhibiting both sharp transients and slowly varying segments. The trend is obtained by minimizing an ℓ₁-regularized second-order difference using a majorization–minimization approach, resulting in an efficient iterative scheme. The method enables robust separation of underlying signal structure from noise while preserving abrupt changes that are typically smoothed out by classical filtering techniques.
View Publications & Projects →
About TS-Analytics
TS-Analytics is currently in preparation and focuses on methodological expertise in statistics, signal processing, and data analysis for medical, engineering, and research-driven applications. The work builds on a strong academic background in electrical engineering and applied statistics, with experience in peer-reviewed research and applied projects. A particular emphasis is placed on well-founded methodologies, reproducible workflows, and clear reporting.
Founded by Dipl.-Ing. Dr. techn. Thomas Suppan
More about TS-Analytics →
Get in Touch
If you are working on a technically or scientifically challenging project and require methodological expertise, feel free to get in touch to discuss future collaborations.
Free initial call · NDA possible · Remote collaboration across AT/DE/EU