Scientific interests
Scientific interests
Embedded Files
Functional data analysis and kernel methods.
Distributional data analysis in wereable data analysis.
Uncertainty quantification: Conformal prediction.
Statistical analysis in metric spaces
Causal inference in digital medicine
Survival analysis with machine learning models
Missing and survey data
Empirical process theory and U-statistics
Relevant applications in digital and personalized medicine, diabetes, and physical activity
Multilevel models and reliability analysis
Relevant Publications
Relevant Publications
*A 6-minute sub-maximal run test to predict VO2 max*
*A 6-minute sub-maximal run test to predict VO2 max*
*Authors:* Matabuena, Marcos; Vidal, Juan C.; Hayes, Philip R.; Huelin Trillo, Fernando.
*Authors:* Matabuena, Marcos; Vidal, Juan C.; Hayes, Philip R.; Huelin Trillo, Fernando.
*Source:* Journal of Sports Sciences, 36(22): 2531–2536, 2018 (Taylor & Francis).
*Source:* Journal of Sports Sciences, 36(22): 2531–2536, 2018 (Taylor & Francis).
*Relevance:* First indirect methodology to reliably predict maximum oxygen consumption using a functional data model.
*Relevance:* First indirect methodology to reliably predict maximum oxygen consumption using a functional data model.
*Glucodensities: A new representation of glucose profiles using distributional data analysis*
*Glucodensities: A new representation of glucose profiles using distributional data analysis*
*Authors:* Matabuena, Marcos; Petersen, Alexander; Vidal, Juan C.; Gude, Francisco.
*Authors:* Matabuena, Marcos; Petersen, Alexander; Vidal, Juan C.; Gude, Francisco.
*Source:* Statistical Methods in Medical Research, 30(6): 1445–1464, 2021 (SAGE Publications).
*Source:* Statistical Methods in Medical Research, 30(6): 1445–1464, 2021 (SAGE Publications).
*Relevance:* Introduces distributional data analysis for wearable data and digital health derived from continuous glucose monitoring (CGM), highlighting advantages over traditional CGM metrics.
*Relevance:* Introduces distributional data analysis for wearable data and digital health derived from continuous glucose monitoring (CGM), highlighting advantages over traditional CGM metrics.
*Kernel biclustering algorithm in Hilbert spaces*
*Kernel biclustering algorithm in Hilbert spaces*
*Authors:* Matabuena, Marcos; Vidal, Juan C.; Padilla, Oscar Hernan Madrid; Sejdinovic, Dino.
*Authors:* Matabuena, Marcos; Vidal, Juan C.; Padilla, Oscar Hernan Madrid; Sejdinovic, Dino.
*Source:* Advanced Data Analysis & Classification, 2025.
*Source:* Advanced Data Analysis & Classification, 2025.
*Relevance:* First method of biclustering to create patient phenotypes using kernel methods for functional data analysis.
*Relevance:* First method of biclustering to create patient phenotypes using kernel methods for functional data analysis.
*Distributional data analysis of accelerometer data from the NHANES database using nonparametric survey regression models*
*Distributional data analysis of accelerometer data from the NHANES database using nonparametric survey regression models*
*Authors:* Matabuena, Marcos; Petersen, Alexander.
*Authors:* Matabuena, Marcos; Petersen, Alexander.
*Source:* Journal of the Royal Statistical Society: Series C (Applied Statistics), 72(2): 294–313, 2023 (Oxford University Press).
*Source:* Journal of the Royal Statistical Society: Series C (Applied Statistics), 72(2): 294–313, 2023 (Oxford University Press).
*Relevance:* Presents distributional representation models for physical activity data, validated with NHANES accelerometer data.
*Relevance:* Presents distributional representation models for physical activity data, validated with NHANES accelerometer data.
*Hypothesis testing for matched pairs with missing data by maximum mean discrepancy: An application to continuous glucose monitoring*
*Hypothesis testing for matched pairs with missing data by maximum mean discrepancy: An application to continuous glucose monitoring*
*Authors:* Matabuena, Marcos; Félix, Paulo; Ditzhaus, Marc; Vidal, Juan; Gude, Francisco.
*Authors:* Matabuena, Marcos; Félix, Paulo; Ditzhaus, Marc; Vidal, Juan; Gude, Francisco.
*Source:* The American Statistician, 77(4): 357–369, 2023 (Taylor & Francis).
*Source:* The American Statistician, 77(4): 357–369, 2023 (Taylor & Francis).
*Relevance:* First hypothesis test for paired missing data from random objects in metric spaces, applied to digital clinical trials and wearable data summaries.
*Relevance:* First hypothesis test for paired missing data from random objects in metric spaces, applied to digital clinical trials and wearable data summaries.
*Causal survival embeddings: Non-parametric counterfactual inference under censoring*
*Causal survival embeddings: Non-parametric counterfactual inference under censoring*
*Authors:* García-Meixide, Carlos; Matabuena, Marcos.
*Authors:* García-Meixide, Carlos; Matabuena, Marcos.
*Source:* Statistical Methods & Medical Research , 2025.
*Source:* Statistical Methods & Medical Research , 2025.
*Relevance:* First framework for counterfactual inference of marginal survival curves using kernel mean embeddings.
*Relevance:* First framework for counterfactual inference of marginal survival curves using kernel mean embeddings.
Functional proportional hazards mixture cure model with applications in cancer mortality in NHANES and post-ICU recovery*
Functional proportional hazards mixture cure model with applications in cancer mortality in NHANES and post-ICU recovery*
*Authors:* Ghosal, Rahul; Matabuena, Marcos; Zhang, Jiajia.
*Authors:* Ghosal, Rahul; Matabuena, Marcos; Zhang, Jiajia.
*Source:* Statistical Methods in Medical Research, 32(11): 2254–2269, 2023 (SAGE Publications).
*Source:* Statistical Methods in Medical Research, 32(11): 2254–2269, 2023 (SAGE Publications).
*Relevance:* First cure survival model accommodating functional covariates, enabling wearable data usage. Particularly relevant for degenerative disorders (e.g., ALS) where a latent cure fraction is possible.
*Relevance:* First cure survival model accommodating functional covariates, enabling wearable data usage. Particularly relevant for degenerative disorders (e.g., ALS) where a latent cure fraction is possible.
Neural interval-censored survival regression with feature selection*
Neural interval-censored survival regression with feature selection*
*Authors:* García-Meixide, Carlos; Matabuena, Marcos; Abraham, Louis; Kosorok, Michael R.
*Authors:* García-Meixide, Carlos; Matabuena, Marcos; Abraham, Louis; Kosorok, Michael R.
*Source:* Statistical Analysis and Data Mining: The ASA Data Science Journal, 17(4): e11704, 2024 (Wiley Online Library).
*Source:* Statistical Analysis and Data Mining: The ASA Data Science Journal, 17(4): e11704, 2024 (Wiley Online Library).
*Relevance:* First paper leveraging neural networks to predict interval-censored outcomes in survival analysis
*Relevance:* First paper leveraging neural networks to predict interval-censored outcomes in survival analysis
Multivariate scalar on multidimensional distribution regression with application to modeling the association between physical activity and cognitive functions*
*Authors:* Ghosal, Rahul; Matabuena, Marcos.
*Source:* Biometrical Journal, 66(7): e202400042, 2024 (Wiley Online Library).
*Relevance:* First paper integrating multidimensional functional distributional representations as predictors in compositional data, combining semi-parametric flexibility with interpretability.
Denoising data with measurement error using a reproducing kernel-based diffusion model*
*Authors:* Yi, Mingyang; Matabuena, Marcos; Wang, Ruoyu.
*Source:* arXiv preprint arXiv:2501.00212, 2024.
*Relevance:* Proposes an elegant solution to measurement error using diffusion models, surpassing slow Gaussian deconvolution methods, with utility in quantifying medical device errors in digital clinical trials
Conformal uncertainty quantification using kernel depth measures in separable Hilbert spaces*
*Authors:* Matabuena, Marcos; Ghosal, Rahul; Mozharovskyi, Pavlo; Padilla, Oscar Hernan Madrid; Onnela, Jukka-Pekka.
*Source:* arXiv preprint arXiv:2405.13970, 2024.
*Relevance:* Introduces a faster predictive-inference methodology for functional data in a separable Hilbert space, leveraging kernel depth measures for robust uncertainty quantification.
Multilevel functional distributional models with application to continuous glucose monitoring in diabetes clinical trials*
*Authors:* Matabuena, Marcos; Crainiceanu, Ciprian M.
*Source:* arXiv preprint arXiv:2403.10514, 2024.
*Relevance:* First multilevel approach for comparing interventions in digital clinical trials conducted in free-living conditions, using distributional data analysis.
Uncertainty quantification for intervals*
*Authors:* García-Meixide, Carlos; Kosorok, Michael R.; Matabuena, Marcos.
*Source:* arXiv preprint arXiv:2408.16381, 2024.
*Relevance:* Provides a general methodology for prediction regions of interval data via conformal prediction + bootstrap, introducing a new class of functions for interval data.
Uncertainty quantification in metric spaces*
*Authors:* Lugosi, Gábor; Matabuena, Marcos.
*Source:* arXiv preprint arXiv:2405.05110, 2024.
*Relevance:* Proposes a general framework using conformal prediction and kNN for heteroscedastic regression in metric spaces, essential for personalized medicine.
*Conformal prediction in dynamic biological systems*
*Authors:* Portela, Alberto; Banga, Julio R.; Matabuena, Marcos.
*Source:* arXiv preprint arXiv:2409.02644, 2024.
*Relevance:* Demonstrates conformal prediction for dynamic biological systems using jackknife methods, outperforming traditional Bayesian approaches in uncertainty quantification.
Some researchers with whom I have had the privilege of working
IMS prize
Discovering, constructing, and nurturing scientific collaborators holds immense significance. My exceptional collaborators have played a pivotal role in shaping my thoughts, advancing my professional journey, and influencing the problems I choose to tackle. Collaborating with distinguished scientists worldwide not only brings immense satisfaction but also opens extraordinary avenues to make a profound difference in the lives of millions.
Page updated
Google Sites
Report abuse