Preeclampsia (PE) is a major cause of perinatal and maternal morbidity and mortality. In the United Kingdom, the National Institute for Clinical Excellence (NICE) has issued guidelines on routine antenatal care recommending that at the booking visit a woman’s level of risk for PE should be determined and the subsequent intensity of antenatal care should be based on this risk assessment. This method relies on a risk scoring system derived from maternal characteristics and medical history; the performance of screening by this method is poor with detection of less than 50% of cases of preterm-PE and term-PE. The objective of this thesis is to develop a method for the estimation of the patient-specific risk for PE by combining the a priori risk based on maternal characteristics and medical history with the results of biophysical and biochemical markers obtained at 11-13 weeks’ gestation. Such early identification of high-risk pregnancies could lead to the use of pharmacological interventions, such as low-dose aspirin, which could prevent the development of the disease. The data for the thesis were derived from two types of studies: First, prospective screening in 65,771 singleton pregnancies, which provided data for maternal factors and serum pregnancy associated plasma protein-A (PAPP-A). In an unselected sequential process we also measured uterine artery pulsatility index (PI) in 45,885 of these pregnancies, mean arterial pressure (MAP) in 35,215 cases and placental growth factor (PLGF) in 14,252 cases. Second, cases-control studies for evaluating the ten most promising biochemical markers identified from search of the literature; for these studies we used stored serum or plasma samples obtained during screening and measured PLGF, Activin-A, Inhibin-A, placental protein-13 (PP13), P-selectin, Pentraxin-3 (PTX-3), soluble Endoglin (sEng), Plasminogen activator inhibitor-2 (PAI-2), Angiopoietin-2 (Ang-2) and soluble fms-like tyrosine kinase-1 (s-Flt-1). A competing risk model was developed which is based on Bayes theorem and combines the prior risk from maternal factors with the distribution of biomarkers to derive patient-specific risk for PE at different stages in pregnancy. The prior risk was derived by multiple regression analysis of maternal factors in the screening study. The distribution of biophysical and biochemical markers was derived from both the screening study and the case-control studies. The prior risk increased with advancing maternal age, increasing weight, was higher in women of Afro-Caribbean and South-Asian racial origin, those with a previous pregnancy with PE, conception by in vitro fertilization and medical history of chronic hypertension, type 1 diabetes mellitus and systemic lupus erythematosus (SLE) or antiphospholipid syndrome (APS). The estimated detection rate (DR) of PE requiring delivery at < 34, < 37 weeks’ gestation and all PE, at false positive rate (FPR) of 10%, in screening by maternal factors were 51, 43 and 40%, respectively. The addition of biochemical markers to maternal factors, including maternal serum PLGF and PAPPA, improved the performance of screening with respective DRs of 74, 56 and 41%. Similarly, addition of biophysical markers to maternal factors, including uterine artery PI and MAP, improved the performance of screening with respective DRs of 90, 72 and 57%. The combination of maternal factors with all the above biophysical and biochemical markers improved the respective DRs to 96, 77 and 54%. The findings of these studies demonstrate that a combination of maternal factors, biophysical and biochemical markers can effectively identify women at high-risk of developing PE.

Preeclampsia is a hypertensive disorder of pregnancy that is diagnosed after the 20th week of gestation. It is defined by the American College of Obstetrics and Gynecology as de novo hypertension of at least 140/90 in a pregnant woman. Proteinuria with the hypertension is sufficient but not required for the diagnosis, especially if a woman displays severe symptoms such as headache, blurry vision, right upper quadrant pain, and low platelet count. Despite significant research, preeclampsia continues to kill 76,000 mothers and 500,000 babies per year worldwide. It causes short and long term consequences such as future metabolic and cardiovascular events for the mother and the child born during a pregnancy affected by preeclampsia. A delay in diagnosis and delayed access to appropriate care is a core cause of the preeclampsia related morbidity and severe mortality worldwide. Despite being in the medical literature since the time of the ancient Greeks, there is currently no significant predictive, preventative, therapeutic, and curative agent for preeclampsia except for an often preterm delivery of the fetus. The complex pathogenesis of preeclampsia has challenged the ability to effectively predict preeclampsia to decrease the delay in this diagnosis. Consequently, an early intervention or triage to higher level obstetric care is hindered. The lack of an early biomarker for preeclampsia also represents a major barrier to treat preeclampsia before major clinical symptoms emerge and the cycle of future cardiovascular risk for mom and baby begins. Novel, very early pregnancy predictive tests for preeclampsia may provide significant clinical utility. Furthermore, a biomarker that is linked with an early pathogenic mechanism in the first trimester development of preeclampsia would reveal a new avenue of early, first trimester intervention to treat and prevent this devastating disease. This work details the search for such a biomarker linked to an early initiator of the molecular pathogenesis of preeclampsia. These microRNA data highlight very important dysregulated mechanisms including immunologic, cell growth, and angiogenic mechanisms. T cells and the role of indoleamine 2,3 dioxygenase (IDO) is important in the early, maternal immune tolerance to the placenta and pregnancy. As poor placentation is a core cause of preeclampsia, a decreased immune tolerance to it is hypothesized to lead to preeclampsia. Furthermore, low IDO activity has been observed in the placentas of preeclamptic pregnancies which may make it a viable biomarker. These IDO-knock out mouse data, demonstrate that chronic IDO deficiency is sufficient to cause some of the core phenotypes of preeclampsia including renal dysfunction, vascular endothelial dysfunction, fetal growth restriction, and a slight increase in systolic blood pressure. This model does not completely phenocopy human preeclampsia. An investigation of early markers that are linked to vascular, immune, and renal abnormalities highlights the vasopressin pathway as a potential biomarker and early initiator of the pathogenesis of preeclampsia. These data demonstrate that copeptin, as a stable marker of vasopressin secretion, is robustly predictive of the development of late pregnancy human preeclampsia, as early as the 6th week of gestation. Furthermore, a mouse model with chronic infusion of vasopressin throughout mouse gestation phenocopies all the essential aspects of human preeclampsia: pregnancy specific hypertension, proteinuria, pathognomonic glomerular endotheliosis, fetal growth restriction, and increased fetal death. Further research must be done to elucidate the immunologic, vascular, and fetal programming phenotypes of this model. This work posits the possibility that the vasopressin pathway may provide new predictive, preventative, therapeutic, and potentially curative modalities for preeclampsia.

Student attrition is one of the long-standing problems facing higher education institutions despite the extensive research that has been undertaken to address it. To increase students’ success and retention rates, there is a need for early alert systems that facilitate the identification of at-risk students so that remedial measures may be taken in time to reduce the risk. However, incorporating ML predictive models into early warning systems face two main challenges: improving the accuracy of timely predictions and the generalizability of predictive models across on-campus and online courses. The goal of this study was to develop and evaluate predictive models that can be applied to on-campus and online courses to predict at-risk students based on data collected from different stages of a course: start of the course, 4th week, 8th week, and 12th week. In this research, several supervised machine learning algorithms were trained and evaluated on their performance. This study compared the performance of single classifiers (Logistic Regression, Decision Trees, Naïve Bayes, and Artificial Neural Networks) and ensemble classifiers (using bagging and boosting techniques). Their performance was evaluated in term of sensitivity, specificity, and Area Under Curve (AUC). A total of four experiments were conducted based on data collected from different stages of the course. In the first experiment, the classification algorithms were trained and evaluated based on data collected before the beginning of the semester. In the second experiment, the classification algorithms were trained and evaluated based on week-four data. Similarly, in the third and fourth experiments, the classification algorithms were trained and evaluated based on week-eight and week-12 data. The results demonstrated that ensemble classifiers were able to achieve the highest classification performance in all experiments. Additionally, the results of the generalizability analysis showed that the predictive models were able to attain a similar performance when used to classify on-campus and online students. Moreover, the Extreme Gradient Boosting (XGBoost) classifier was found to be the best performing classifier suited for the at-risk students’ classification problem and was able to achieve an AUC of ≈ 0.89, a sensitivity of ≈ 0.81, and specificity of ≈ 0.81 using data available at the start of a course. Finally, the XGBoost classifier was able to improve by 1% for each subsequent four weeks dataset reaching an AUC of ≈ 0.92, a sensitivity of ≈ 0.84, and specificity of ≈ 0.84 by week 12. While the additional learning management system's (LMS) data helped in improving the prediction accuracy consistently as the course progresses, the improvement was marginal. Such findings suggest that the predictive models can be used to identify at-risk students even in courses that do not make significant use of LMS. The results of this research demonstrated the usefulness and effectiveness of ML techniques for early identification of at-risk students. Interestingly, it was found that fairly reliable predictions can be made at the start of the semester, which is significant in that help can be provided to at-risk students even before the course starts. Finally, it is hoped that the results of this study advance the understanding of the appropriateness and effectiveness of ML techniques when used for early identification of at-risk students.

A form of laboratory tests are performed to determine battery degradation due to charging and discharging of batteries (cycling). This is done as part of quality assurance in battery production since a certain amount of degradation corresponds to the end of the battery lifetime. Currently, this requires a significant amount of cycling. Thus, if it’s possible to decrease the number of cycles required, the time and costs for battery degradation testing can be reduced. The aim of this thesis is therefore to create a model for prediction of battery lifetime while using early cycle data. Further, to assist planning regarding scale of cycle testing this study aims to examine the impact of implementing such a prediction model in production. To examine which data driven model that should be used to predict the battery lifetime at the company, extensive feature engineering is performed where measurements from specific cycles are used, inspired by the previous work of Severson et al. (2019) and Fei et al. (2021). Two models are then examined: Linear Regression with Elastic net and Support Vector Regression. To investigate the extent to which an implementation of such a model can affect battery testing capacity, two scenarios are compared. The first scenario is that of the current cycle testing at the company and the second scenario involves implementing a prediction model. The comparison then examines the time required for battery testing and the number of machines to cycle the batteries (cyclers). Based on the results obtained, the data driven model that should be implemented is a Support Vector Regression model with features relating to different battery cycling phases or measurements, such as charge process, temperature and capacity. It can also be shown that if a battery lifetime prediction model is implemented, it can reduce the time and number of cyclers required for testing with approximately 93 %, compared to traditional testing.

No description available.