Model-based clustering is a sub-field of statistical modeling and machine learning. The mixture models use the probability to describe the degree of the data point belonging to the cluster, and the probability is updated iteratively during the clustering. While mixture models have demonstrated the superior performance in handling noisy data in many fields, there exist some challenges for high dimensional dataset. It is noted that among a large number of features, some may not indeed contribute to delineate the cluster profiles. The inclusion of these “noisy” features will confuse the model to identify the real structure of the clusters and cost more computational time. Recognizing the issue, in this dissertation, I propose a new feature selection algorithm for continuous dataset first and then extend to mixed datatype. Finally, I conduct uncertainty quantification for the feature selection results as the third topic.

The first topic is an embedded feature selection algorithm termed Expectation-Selection-Maximization (ESM) model that can automatically select features while optimizing the parameters for Gaussian Mixture Model. I introduce a relevancy index (RI) revealing the contribution of the feature in the clustering process to assist feature selection. I demonstrate the efficacy of the ESM by studying two synthetic datasets, four benchmark datasets, and an Alzheimer’s Disease dataset.

The second topic focuses on extending the application of ESM algorithm to handle mixed datatypes. The Gaussian mixture model is generalized to Generalized Model of Mixture (GMoM), which can not only handle continuous features, but also binary and nominal features.

The last topic is about Uncertainty Quantification (UQ) of the feature selection. A new algorithm termed ESOM is proposed, which takes the variance information into consideration while conducting feature selection. Also, a set of outliers are generated in the feature selection process to infer the uncertainty in the input data. Finally, the selected features and detected outlier instances are evaluated by visualization comparison.

Deep learning is a sub-field of machine learning in which models are developed to imitate the workings of the human brain in processing data and creating patterns for decision making. This dissertation is focused on developing deep learning models for medical imaging analysis of different modalities for different tasks including detection, segmentation and classification. Imaging modalities including digital mammography (DM), magnetic resonance imaging (MRI), positron emission tomography (PET) and computed tomography (CT) are studied in the dissertation for various medical applications. The first phase of the research is to develop a novel shallow-deep convolutional neural network (SD-CNN) model for improved breast cancer diagnosis. This model takes one type of medical image as input and synthesizes different modalities for additional feature sources; both original image and synthetic image are used for feature generation. This proposed architecture is validated in the application of breast cancer diagnosis and proved to be outperforming the competing models. Motivated by the success from the first phase, the second phase focuses on improving medical imaging synthesis performance with advanced deep learning architecture. A new architecture named deep residual inception encoder-decoder network (RIED-Net) is proposed. RIED-Net has the advantages of preserving pixel-level information and cross-modality feature transferring. The applicability of RIED-Net is validated in breast cancer diagnosis and Alzheimer’s disease (AD) staging. Recognizing medical imaging research often has multiples inter-related tasks, namely, detection, segmentation and classification, my third phase of the research is to develop a multi-task deep learning model. Specifically, a feature transfer enabled multi-task deep learning model (FT-MTL-Net) is proposed to transfer high-resolution features from segmentation task to low-resolution feature-based classification task. The application of FT-MTL-Net on breast cancer detection, segmentation and classification using DM images is studied. As a continuing effort on exploring the transfer learning in deep models for medical application, the last phase is to develop a deep learning model for both feature transfer and knowledge from pre-training age prediction task to new domain of Mild cognitive impairment (MCI) to AD conversion prediction task. It is validated in the application of predicting MCI patients’ conversion to AD with 3D MRI images.

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.

With the development of computer and sensing technology, rich datasets have become available in many fields such as health care, manufacturing, transportation, just to name a few. Also, data come from multiple heterogeneous sources or modalities. This is a common phenomenon in health care systems. While multi-modality data fusion is a promising research area, there are several special challenges in health care applications. (1) The integration of biological and statistical model is a big challenge; (2) It is commonplace that data from various modalities is not available for every patient due to cost, accessibility, and other reasons. This results in a special missing data structure in which different modalities may be missed in “blocks”. Therefore, how to train a predictive model using such a dataset poses a significant challenge to statistical learning. (3) It is well known that different modality data may contain different aspects of information about the response. The current studies cannot afford to solve this problem. My dissertation includes new statistical learning model development to address each of the aforementioned challenges as well as application case studies using real health care datasets, included in three chapters (Chapter 2, 3, and 4), respectively. Collectively, it is expected that my dissertation could provide a new sets of statistical learning models, algorithms, and theory contributed to multi-modality heterogeneous data fusion driven by the unique challenges in this area. Also, application of these new methods to important medical problems using real-world datasets is expected to provide solutions to these problems, and therefore contributing to the application domains.

Rapid advancements in Artificial Intelligence (AI), Machine Learning, and Deep Learning technologies are widening the playing field for automated decision assistants in healthcare. The field of radiology offers a unique platform for this technology due to its repetitive work structure, ability to leverage large data sets, and high position for clinical and social impact. Several technologies in cancer screening, such as Computer Aided Detection (CAD), have broken the barrier of research into reality through successful outcomes with patient data (Morton, Whaley, Brandt, & Amrami, 2006; Patel et al, 2018). Technologies, such as the IBM Medical Sieve, are growing excitement with the potential for increased impact through the addition of medical record information ("Medical Sieve Radiology Grand Challenge", 2018). As the capabilities of automation increase and become a part of expert-decision-making jobs, however, the careful consideration of its integration into human systems is often overlooked. This paper aims to identify how healthcare professionals and system engineers implementing and interacting with automated decision-making aids in Radiology should take bureaucratic, legal, professional, and political accountability concerns into consideration. This Accountability Framework is modeled after Romzek and Dubnick’s (1987) public administration framework and expanded on through an analysis of literature on accountability definitions and examples in military, healthcare, and research sectors. A cohesive understanding of this framework and the human concerns it raises helps drive the questions that, if fully addressed, create the potential for a successful integration and adoption of AI in radiology and ultimately the care environment.

Transfer learning is a sub-field of statistical modeling and machine learning. It refers to methods that integrate the knowledge of other domains (called source domains) and the data of the target domain in a mathematically rigorous and intelligent way, to develop a better model for the target domain than a model using the data of the target domain alone. While transfer learning is a promising approach in various application domains, my dissertation research focuses on the particular application in health care, including telemonitoring of Parkinson’s Disease (PD) and radiomics for glioblastoma.

The first topic is a Mixed Effects Transfer Learning (METL) model that can flexibly incorporate mixed effects and a general-form covariance matrix to better account for similarity and heterogeneity across subjects. I further develop computationally efficient procedures to handle unknown parameters and large covariance structures. Domain relations, such as domain similarity and domain covariance structure, are automatically quantified in the estimation steps. I demonstrate METL in an application of smartphone-based telemonitoring of PD.

The second topic focuses on an MRI-based transfer learning algorithm for non-invasive surgical guidance of glioblastoma patients. Limited biopsy samples per patient create a challenge to build a patient-specific model for glioblastoma. A transfer learning framework helps to leverage other patient’s knowledge for building a better predictive model. When modeling a target patient, not every patient’s information is helpful. Deciding the subset of other patients from which to transfer information to the modeling of the target patient is an important task to build an accurate predictive model. I define the subset of “transferrable” patients as those who have a positive rCBV-cell density correlation, because a positive correlation is confirmed by imaging theory and the its respective literature.

The last topic is a Privacy-Preserving Positive Transfer Learning (P3TL) model. Although negative transfer has been recognized as an important issue by the transfer learning research community, there is a lack of theoretical studies in evaluating the risk of negative transfer for a transfer learning method and identifying what causes the negative transfer. My work addresses this issue. Driven by the theoretical insights, I extend Bayesian Parameter Transfer (BPT) to a new method, i.e., P3TL. The unique features of P3TL include intelligent selection of patients to transfer in order to avoid negative transfer and maintain patient privacy. These features make P3TL an excellent model for telemonitoring of PD using an At-Home Testing Device.

With trends of globalization on rise, predominant of the trades happen by sea, and experts have predicted an increase in trade volumes over the next few years. With increasing trade volumes, container ships’ upsizing is being carried out to meet the demand. But the problem with container ships’ upsizing is that the sea port terminals must be equipped adequately to improve the turnaround time otherwise the container ships’ upsizing would not yield the anticipated benefits. This thesis focus on a special type of a double automated crane set-up, with a finite interoperational buffer capacity. The buffer is placed in between the cranes, and the idea behind this research is to analyze the performance of the crane operations when this technology is adopted. This thesis proposes the approximation of this complex system, thereby addressing the computational time issue and allowing to efficiently analyze the performance of the system. The approach to model this system has been carried out in two phases. The first phase consists of the development of discrete event simulation model to make the system evolve over time. The challenges of this model are its high processing time which consists of performing large number of experimental runs, thus laying the foundation for the development of the analytical model of the system, and with respect to analytical modeling, a continuous time markov process approach has been adopted. Further, to improve the efficiency of the analytical model, a state aggregation approach is proposed. Thus, this thesis would give an insight on the outcomes of the two approaches and the behavior of the error space, and the performance of the models for the varying buffer capacities would reflect the scope of improvement in these kinds of operational set up.

Mathematical modeling and decision-making within the healthcare industry have given means to quantitatively evaluate the impact of decisions into diagnosis, screening, and treatment of diseases. In this work, we look into a specific, yet very important disease, the Alzheimer. In the United States, Alzheimer’s Disease (AD) is the 6th leading cause of death. Diagnosis of AD cannot be confidently confirmed until after death. This has prompted the importance of early diagnosis of AD, based upon symptoms of cognitive decline. A symptom of early cognitive decline and indicator of AD is Mild Cognitive Impairment (MCI). In addition to this qualitative test, Biomarker tests have been proposed in the medical field including p-Tau, FDG-PET, and hippocampal. These tests can be administered to patients as early detectors of AD thus improving patients’ life quality and potentially reducing the costs of the health structure. Preliminary work has been conducted in the development of a Sequential Tree Based Classifier (STC), which helps medical providers predict if a patient will contract AD or not, by sequentially testing these biomarker tests. The STC model, however, has its limitations and the need for a more complex, robust model is needed. In fact, STC assumes a general linear model as the status of the patient based upon the tests results. We take a simulation perspective and try to define a more complex model that represents the patient evolution in time.

Specifically, this thesis focuses on the formulation of a Markov Chain model that is complex and robust. This Markov Chain model emulates the evolution of MCI patients based upon doctor visits and the sequential administration of biomarker tests. Data provided to create this Markov Chain model were collected by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. The data lacked detailed information of the sequential administration of the biomarker tests and therefore, different analytical approaches were tried and conducted in order to calibrate the model. The resulting Markov Chain model provided the capability to conduct experiments regarding different parameters of the Markov Chain and yielded different results of patients that contracted AD and those that did not, leading to important insights into effect of thresholds and sequence on patient prediction capability as well as health costs reduction.



The data in this thesis was provided from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.edu). ADNI investigators did not contribute to any analysis or writing of this thesis. A list of the ADNI investigators can be found at: http://adni.loni.usc.edu/about/governance/principal-investigators/ .

Technology advancements in diagnostic imaging, smart sensing, and health information systems have resulted in a data-rich environment in health care, which offers a great opportunity for Precision Medicine. The objective of my research is to develop data fusion and system informatics approaches for quality and performance improvement of health care. In my dissertation, I focus on three emerging problems in health care and develop novel statistical models and machine learning algorithms to tackle these problems from diagnosis to care to system-level decision-making.

The first topic is diagnosis/subtyping of migraine to customize effective treatment to different subtypes of patients. Existing clinical definitions of subtypes use somewhat arbitrary boundaries primarily based on patient self-reported symptoms, which are subjective and error-prone. My research develops a novel Multimodality Factor Mixture Model that discovers subtypes of migraine from multimodality imaging MRI data, which provides complementary accurate measurements of the disease. Patients in the different subtypes show significantly different clinical characteristics of the disease. Treatment tailored and optimized for patients of the same subtype paves the road toward Precision Medicine.

The second topic focuses on coordinated patient care. Care coordination between nurses and with other health care team members is important for providing high-quality and efficient care to patients. The recently developed Nurse Care Coordination Instrument (NCCI) is the first of its kind that enables large-scale quantitative data to be collected. My research develops a novel Multi-response Multi-level Model (M3) that enables transfer learning in NCCI data fusion. M3 identifies key factors that contribute to improving care coordination, and facilitates the design and optimization of nurses’ training, workload assignment, and practice environment, which leads to improved patient outcomes.

The last topic is about system-level decision-making for Alzheimer’s disease early detection at the early stage of Mild Cognitive Impairment (MCI), by predicting each MCI patient’s risk of converting to AD using imaging and proteomic biomarkers. My research proposes a systems engineering approach that integrates the multi-perspectives, including prediction accuracy, biomarker cost/availability, patient heterogeneity and diagnostic efficiency, and allows for system-wide optimized decision regarding the biomarker testing process for prediction of MCI conversion.