Robust Learning Algorithms for Bioengineering Systems

• Author / Creator

  Nadadoor Srinivasan, Venkat R.

• Biological engineering is a domain of study that involves applying known engineering principles to biological systems. Qualitative studies in the field of biology have undergone tremendous advancements in the last two decades but quantitation is still in its early stages due to various complexities involved in its design, control, and operation. The current state of research in the field of bioengineering involves mostly elementary quantitation of biological systems without a strong grasp into the fundamentals of engineering. Advanced learning algorithms can help overcome some of the problems generally associated with biological systems including model complexity, noisy measurements, and data scarcity. In the current study, bioengineering problems are viewed from process system engineering perspective with a focus on three aspects: modeling, monitoring, and fault detection. The three representative bioengineering problems chosen to cover the three aforementioned aspects are:
  Modeling a gene network: Accurate inference of gene network can provide information that can lead to new ideas for treating complex diseases. A novel algorithm for building gene networks from microarray datasets using a first principles differential equations model is proposed. The proposed algorithm was able to obtain a good estimate of the gene connectivity matrix for an experimental dataset on a nine gene network in Eschericia coli.
  Monitoring a microalgal bioreactor system: Monitoring of process conditions in algal cultures helps in maximizing oil productivity. A support vector regression based algorithm is proposed for monitoring the culture conditions of an algal bioreactor system. The multivariate sensor built using an experimental dataset gave good predictions for the concentrations of biomass, glucose and percentage oil content.
  Detection of transplant rejection: Early detection of graft rejection is mandatory to effectively treat and prevent cardiac dysfunction. An algorithm based on hypothesis testing is proposed for detecting biomarkers useful for detection of rejection. The chosen biomarkers are validated on publicly available microarray datasets. For these datasets, the biomarkers obtained based on the proposed method were able to achieve a good separation between the successful and failed transplant classes.

  The methodologies and strategies proposed in this thesis have helped in the modeling, monitoring, and fault detection of bioengineering systems.

• Subjects / Keywords

  • Gene Network
  • Statistical Methods
  • Monitoring
  • Fault Detection
  • Modelling
  • Transplant Rejection
  • Bioreactor System

• Graduation date

  Spring 2012

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/R3W88Q

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.