Matching Items (4)
Filtering by
- All Subjects: Biomedical Engineering
Description
Heart transplantation is the final treatment option for end-stage heart failure. In the United States, 70 pediatric patients die annually on the waitlist while 800 well-functioning organs get discarded. Concern for potential size-mismatch is one source of allograft waste and high waitlist mortality. Clinicians use the donor-recipient body weight (DRBW) ratio, a standalone metric, to evaluate allograft size-match. However, this body weight metric is far removed from cardiac anatomy and neglects an individual’s anatomical variations. This thesis body of work developed a novel virtual heart transplant fit assessment tool and investigated the tool’s clinical utility to help clinicians safely expand patient donor pools.
The tool allowed surgeons to take an allograft reconstruction and fuse it to a patient’s CT or MR medical image for virtual fit assessment. The allograft is either a reconstruction of the donor’s actual heart (from CT or MR images) or an analogue from a health heart library. The analogue allograft geometry is identified from gross donor parameters using a regression model build herein. The need for the regression model is donor images may not exist or they may not become available within the time-window clinicians have to make a provisional acceptance of an offer.
The tool’s assessment suggested > 20% of upper DRBW listings could have been increased at Phoenix Children’s Hospital (PCH). Upper DRBW listings in the UNOS national database was statistically smaller than at PCH (p-values: < 0.001). Delayed sternal closure and surgeon perceived complication variables had an association (p-value: 0.000016) with 9 of the 11 cases that surgeons had perceived fit-related complications had delayed closures (p-value: 0.034809).
A tool to assess allograft size-match has been developed. Findings warrant future preclinical and clinical prospective studies to further assess the tool’s clinical utility.
The tool allowed surgeons to take an allograft reconstruction and fuse it to a patient’s CT or MR medical image for virtual fit assessment. The allograft is either a reconstruction of the donor’s actual heart (from CT or MR images) or an analogue from a health heart library. The analogue allograft geometry is identified from gross donor parameters using a regression model build herein. The need for the regression model is donor images may not exist or they may not become available within the time-window clinicians have to make a provisional acceptance of an offer.
The tool’s assessment suggested > 20% of upper DRBW listings could have been increased at Phoenix Children’s Hospital (PCH). Upper DRBW listings in the UNOS national database was statistically smaller than at PCH (p-values: < 0.001). Delayed sternal closure and surgeon perceived complication variables had an association (p-value: 0.000016) with 9 of the 11 cases that surgeons had perceived fit-related complications had delayed closures (p-value: 0.034809).
A tool to assess allograft size-match has been developed. Findings warrant future preclinical and clinical prospective studies to further assess the tool’s clinical utility.
ContributorsPlasencia, Jonathan (Author) / Frakes, David H (Thesis advisor) / Kodibagkar, Vikram (Thesis advisor) / Sadleir, Rosalind (Committee member) / Kamarianakis, Yiannis (Committee member) / Zangwill, Steven (Committee member) / Pophal, Stephen (Committee member) / Arizona State University (Publisher)
Created2018
Description
Over the past 20 years, the fields of synthetic biology and synthetic biosystems engineering have grown into mature disciplines, leading to significant breakthroughs in cancer research, diagnostics, cell-based medicines, biochemical production, etc. Application of mathematical modelling to biological and biochemical systems have not only given great insight into how these systems function, but also have lent enough predictive power to aid in the forward-engineering of synthetic constructs. However, progress has been impeded by several modes of context-dependence unique to biological and biochemical systems that are not seen in traditional engineering disciplines, resulting in the need for lengthy design-build-test cycles before functional prototypes are generated.In this work, two of these universal modes of context dependence – resource competition and growth feedback –their effects on synthetic gene circuits and potential control mechanisms, are studied and characterized. Results demonstrate that a novel competitive control architecture can be utilized to mitigate the effects of winner-take-all resource competition (a form of context dependence where distinct gene modules influence each other by competing over a shared pool of transcriptional/translational resources) in synthetic gene circuits and restore circuits to their intended function. Application of the fluctuation-dissipation theorem and rigorous stochastic simulations demonstrate that realistic resource constraints present in cells at the transcriptional and translational levels influence noise in gene circuits in a nonmonotonic fashion, either increasing or decreasing noise depending on the transcriptional/translational capacity.
Growth feedback on the other hand links circuit function to cellular growth rate via increased protein dilution rate during exponential growth phase. This in turn can result in the collapse of bistable gene circuits as the accelerated dilution rate forces switches in a high stable state to fall to a low stable state. Mathematical modelling and experimental data demonstrate that application of repressive links can insulate sensitive parts of gene circuits against growth-fluctuations and can in turn increase the robustness of multistable circuits in growth contexts.
The results presented in this work aid in the accumulation of understanding of biological and biochemical context dependence, and corresponding control strategies and design principles engineers can utilize to mitigate these effects.
ContributorsStone, Austin (Author) / Tian, Xiao-jun (Thesis advisor) / Wang, Xiao (Committee member) / Smith, Barbara (Committee member) / Kuang, Yang (Committee member) / Cheng, Albert (Committee member) / Arizona State University (Publisher)
Created2023
Description
Semi-supervised learning (SSL) is sub-field of statistical machine learning that is useful for problems that involve having only a few labeled instances with predictor (X) and target (Y) information, and abundance of unlabeled instances that only have predictor (X) information. SSL harnesses the target information available in the limited labeled data, as well as the information in the abundant unlabeled data to build strong predictive models. However, not all the included information is useful. For example, some features may correspond to noise and including them will hurt the predictive model performance. Additionally, some instances may not be as relevant to model building and their inclusion will increase training time and potentially hurt the model performance. The objective of this research is to develop novel SSL models to balance data inclusivity and usability. My dissertation research focuses on applications of SSL in healthcare, driven by problems in brain cancer radiomics, migraine imaging, and Parkinson’s Disease telemonitoring.
The first topic introduces an integration of machine learning (ML) and a mechanistic model (PI) to develop an SSL model applied to predicting cell density of glioblastoma brain cancer using multi-parametric medical images. The proposed ML-PI hybrid model integrates imaging information from unbiopsied regions of the brain as well as underlying biological knowledge from the mechanistic model to predict spatial tumor density in the brain.
The second topic develops a multi-modality imaging-based diagnostic decision support system (MMI-DDS). MMI-DDS consists of modality-wise principal components analysis to incorporate imaging features at different aggregation levels (e.g., voxel-wise, connectivity-based, etc.), a constrained particle swarm optimization (cPSO) feature selection algorithm, and a clinical utility engine that utilizes inverse operators on chosen principal components for white-box classification models.
The final topic develops a new SSL regression model with integrated feature and instance selection called s2SSL (with “s2” referring to selection in two different ways: feature and instance). s2SSL integrates cPSO feature selection and graph-based instance selection to simultaneously choose the optimal features and instances and build accurate models for continuous prediction. s2SSL was applied to smartphone-based telemonitoring of Parkinson’s Disease patients.
The first topic introduces an integration of machine learning (ML) and a mechanistic model (PI) to develop an SSL model applied to predicting cell density of glioblastoma brain cancer using multi-parametric medical images. The proposed ML-PI hybrid model integrates imaging information from unbiopsied regions of the brain as well as underlying biological knowledge from the mechanistic model to predict spatial tumor density in the brain.
The second topic develops a multi-modality imaging-based diagnostic decision support system (MMI-DDS). MMI-DDS consists of modality-wise principal components analysis to incorporate imaging features at different aggregation levels (e.g., voxel-wise, connectivity-based, etc.), a constrained particle swarm optimization (cPSO) feature selection algorithm, and a clinical utility engine that utilizes inverse operators on chosen principal components for white-box classification models.
The final topic develops a new SSL regression model with integrated feature and instance selection called s2SSL (with “s2” referring to selection in two different ways: feature and instance). s2SSL integrates cPSO feature selection and graph-based instance selection to simultaneously choose the optimal features and instances and build accurate models for continuous prediction. s2SSL was applied to smartphone-based telemonitoring of Parkinson’s Disease patients.
ContributorsGaw, Nathan (Author) / Li, Jing (Thesis advisor) / Wu, Teresa (Committee member) / Yan, Hao (Committee member) / Hu, Leland (Committee member) / Arizona State University (Publisher)
Created2019
Description
Ideas from coding theory are employed to theoretically demonstrate the engineering of mutation-tolerant genes, genes that can sustain up to some arbitrarily chosen number of mutations and still express the originally intended protein. Attention is restricted to tolerating substitution mutations. Future advances in genomic engineering will make possible the ability to synthesize entire genomes from scratch. This presents an opportunity to embed desirable capabilities like mutation-tolerance, which will be useful in preventing cell deaths in organisms intended for research or industrial applications in highly mutagenic environments. In the extreme case, mutation-tolerant genes (mutols) can make organisms resistant to retroviral infections.
An algebraic representation of the nucleotide bases is developed. This algebraic representation makes it possible to convert nucleotide sequences into algebraic sequences, apply mathematical ideas and convert results back into nucleotide terms. Using the algebra developed, a mapping is found from the naturally-occurring codons to an alternative set of codons which makes genes constructed from them mutation-tolerant, provided no more than one substitution mutation occurs per codon. The ideas discussed naturally extend to finding codons that can tolerate t arbitrarily chosen number of mutations per codon. Finally, random substitution events are simulated in both a wild-type green fluorescent protein (GFP) gene and its mutol variant and the amino acid sequence expressed from each post-mutation is compared with the amino acid sequence pre-mutation.
This work assumes the existence of synthetic protein-assembling entities that function like tRNAs but can read k nucleotides at a time, with k greater than or equal to 5. The realization of this assumption is presented as a challenge to the research community.
An algebraic representation of the nucleotide bases is developed. This algebraic representation makes it possible to convert nucleotide sequences into algebraic sequences, apply mathematical ideas and convert results back into nucleotide terms. Using the algebra developed, a mapping is found from the naturally-occurring codons to an alternative set of codons which makes genes constructed from them mutation-tolerant, provided no more than one substitution mutation occurs per codon. The ideas discussed naturally extend to finding codons that can tolerate t arbitrarily chosen number of mutations per codon. Finally, random substitution events are simulated in both a wild-type green fluorescent protein (GFP) gene and its mutol variant and the amino acid sequence expressed from each post-mutation is compared with the amino acid sequence pre-mutation.
This work assumes the existence of synthetic protein-assembling entities that function like tRNAs but can read k nucleotides at a time, with k greater than or equal to 5. The realization of this assumption is presented as a challenge to the research community.
ContributorsAmpofo, Prince Kwame (Author) / Tian, Xiaojun (Thesis advisor) / Kiani, Samira (Committee member) / Kuang, Yang (Committee member) / Arizona State University (Publisher)
Created2019