CARG HUB: AI for bioreactors

This research is all about improving how we monitor bioprocesses in the biopharmaceutical industry, making it faster and less expensive. The project focuses on developing advanced tools called nonlinear Kalman estimators (NKEs) to tackle these issues. Specifically, it aims to solve three big challenges: estimating important process parameters in complex biological systems, automatically tuning these estimation tools for different production conditions, and making the estimators more adaptable and accurate for different types of processes.

Publications

2026-03-27

Manuscripts

Model-Based Characterization of the Metabolism of Recombinant Adeno-Associated Virus (rAAV) Production via Human Embryonic Kidney (HEK293) Cells

Published in Bioengineering

by S. Khodadadi Karimvand, M. Cuperlovic-Culf M, A.A. Kamen, M. Bolic

This paper presents a kinetic-metabolic model of AAV6 production in HEK293 cells, fitted to bioreactor metabolite data and analyzed with Sobol sensitivity methods to identify the parameters most affecting virus production dynamics.

Download Paper

2025-09-30

Manuscripts

Batch Bayesian Auto-Tuning for Nonlinear Kalman Estimators

Published in Nature Scientific Reports

by C. Iglesias Jr, M. Bolic

BAT is a batch Bayesian auto-tuning method for nonlinear Kalman estimators that jointly optimizes all key filter components using all available measurements, improving tuning accuracy, consistency, and practical reliability without relying on ground-truth states or manual adjustment.

Download Paper

2024-12-17

Conferences

Two Students: Enabling Uncertainty Quantification in Federated Learning Clients

Published in NeurIPS 2024 Workshop BDU

by C. Iglesias Jr, et al.

This paper is about novel Bayesian approaches for federated learning.

Download Paper

2024-01-01

Manuscripts

How Not to Make the Joint Extended Kalman Filter Fail with Unstructured Mechanistic Models

Published in Sensors

by Iglesias Jr, CF, Bolic, M.

Addresses the problems in joint estimation of states and parameters with an extended Kalman filter.

Download Paper

2023-01-01

Manuscripts

rAAV Manufacturing: The Challenges of Soft Sensing during Upstream Processing

Published in Bioengineering

by Iglesias, CF, Ristovski, M, Bolic, M, Cuperlovic-Culf, M

This is a survey paper about machine learning models for rAAV manufacturing.

Download Paper

2022-01-01

Manuscripts

Monitoring the Recombinant Adeno-Associated Virus Production using Extended Kalman Filter

Published in Processes

by Iglesias Jr, CF, Xu, X, Mehta, V, Akassou, M, Venereo-Sanchez, A, Belacel, N, Bolic, M

This paper monitors the rAAV production using the online viable cell density measurements and estimating the other state variables measured at a low frequency.

Download Paper

2021-01-01

Conferences

Handling Massive Proportion of Missing Labels in Multivariate Long-Term Time Series Forecasting

Published in Journal of Physics: Conference Series

by Iglesias, J. C., Mehta, V., Venereo-Sanchez, A., Xu, X., Robitaille, J., Voyer, R., others

This paper handles situation where there are sparse labels for training machine learning forecast models for time series.

Download Paper

Researchers

Miodrag Bolic
Principal Investigator

Alumni

Cristovão Freitas Iglesias Junior (Alumni)
PhD

Academic Collaborators

Researcher Name	University/Institute	Position
M. Cuperlovic-Culf	NRC, Canada	Senior Research Officer

Industrial Partners

Company Name	Location
National Research Council Canada (NRC)	Ottawa, Canada

CARG HUB: AI for bioreactors

Publications

Model-Based Characterization of the Metabolism of Recombinant Adeno-Associated Virus (rAAV) Production via Human Embryonic Kidney (HEK293) Cells

Batch Bayesian Auto-Tuning for Nonlinear Kalman Estimators

Two Students: Enabling Uncertainty Quantification in Federated Learning Clients

How Not to Make the Joint Extended Kalman Filter Fail with Unstructured Mechanistic Models

rAAV Manufacturing: The Challenges of Soft Sensing during Upstream Processing

Monitoring the Recombinant Adeno-Associated Virus Production using Extended Kalman Filter

Handling Massive Proportion of Missing Labels in Multivariate Long-Term Time Series Forecasting

Researchers

Miodrag Bolic

Alumni

Cristovão Freitas Iglesias Junior (Alumni)

Academic Collaborators

Industrial Partners