No description available.

No description available.

No description available.

Objective: This thesis focuses on design and evaluation of a disease prediction system that be able to detect hidden and upcoming diseases of an individual. Unlike previous works that has typically relied on precise medical examinations to extract symptoms and risk factors for computing probability of occurrence of a disease, the proposed disease prediction system is based on similar patterns of disease comorbidity in population and the individual to evaluate the risk of a disease. Methods: We combine three machine learning algorithms to construct the prediction system: an item based recommendation system, a Bayesian graphical model and a rule based recommender. We also propose multiple similarity measures for the recommendation system, each useful in a particular condition. We finally show how best values of parameters of the system can be derived from optimization of cost function and ROC curve. Results: A permutation test is designed to evaluate accuracy of the prediction system accurately. Results showed considerable advantage of the proposed system in compare to an item based recommendation system and improvements of prediction if system is trained for each specific gender and race. Conclusion: The proposed system has been shown to be a competent method in accurately identifying potential diseases in patients with multiple diseases, just based on their disease records. The procedure also contains novel soft computing and machine learning ideas that can be used in prediction problems. The proposed system has the possibility of using more complex datasets that include timeline of diseases, disease networks and social network. This makes it an even more capable platform for disease prediction. Hence, this thesis contributes to improvement of the disease prediction field. / Graduate / 0800 / 0766 / 0984 / mehrdadmansouri@yahoo.com

abstract: Proteins are essential for most biological processes that constitute life. The function of a protein is encoded within its 3D folded structure, which is determined by its sequence of amino acids. A variation of a single nucleotide in the DNA during transcription (nSNV) can alter the amino acid sequence (i.e., a mutation in the protein sequence), which can adversely impact protein function and sometimes cause disease. These mutations are the most prevalent form of variations in humans, and each individual genome harbors tens of thousands of nSNVs that can be benign (neutral) or lead to disease. The primary way to assess the impact of nSNVs on function is through evolutionary approaches based on positional amino acid conservation. These approaches are largely inadequate in the regime where positions evolve at a fast rate. We developed a method called dynamic flexibility index (DFI) that measures site-specific conformational dynamics of a protein, which is paramount in exploring mechanisms of the impact of nSNVs on function. In this thesis, we demonstrate that DFI can distinguish the disease-associated and neutral nSNVs, particularly for fast evolving positions where evolutionary approaches lack predictive power. We also describe an additional dynamics-based metric, dynamic coupling index (DCI), which measures the dynamic allosteric residue coupling of distal sites on the protein with the functionally critical (i.e., active) sites. Through DCI, we analyzed 200 disease mutations of a specific enzyme called GCase, and a proteome-wide analysis of 75 human enzymes containing 323 neutral and 362 disease mutations. In both cases we observed that sites with high dynamic allosteric residue coupling with the functional sites (i.e., DARC spots) have an increased susceptibility to harboring disease nSNVs. Overall, our comprehensive proteome-wide analysis suggests that incorporating these novel position-specific conformational dynamics based metrics into genomics can complement current approaches to increase the accuracy of diagnosing disease nSNVs. Furthermore, they provide mechanistic insights about disease development. Lastly, we introduce a new, purely sequence-based model that can estimate the dynamics profile of a protein by only utilizing coevolution information, eliminating the requirement of the 3D structure for determining dynamics. / Dissertation/Thesis / Doctoral Dissertation Physics 2016

Deep Learning (DL) has been broadly applied to solve big data problems in biomedical fields, which is most successful in image processing. Recently, many DL methods have been applied to analyze genomic studies. However, genomic data usually has too small a sample size to fit a complex network. They do not have common structural patterns like images to utilize pre-trained networks or take advantage of convolution layers. The concern of overusing DL methods motivates us to evaluate DL methods' performance versus popular non-deep Machine Learning (ML) methods for analyzing genomic data with a wide range of sample sizes. In this paper, we conduct a benchmark study using the UK Biobank data and its many random subsets with different sample sizes. The original UK Biobank data has about 500k participants. Each patient has comprehensive patient characteristics, disease histories, and genomic information, i.e., the genotypes of millions of Single-Nucleotide Polymorphism (SNPs). We are interested in predicting the risk of three lung diseases: asthma, COPD, and lung cancer. There are 205,238 participants have recorded disease outcomes for these three diseases. Five prediction models are investigated in this benchmark study, including three non-deep machine learning methods (Elastic Net, XGBoost, and SVM) and two deep learning methods (DNN and LSTM). Besides the most popular performance metrics, such as the F1-score, we promote the hit curve, a visual tool to describe the performance of predicting rare events. We discovered that DL methods frequently fail to outperform non-deep ML in analyzing genomic data, even in large datasets with over 200k samples. The experiment results suggest not overusing DL methods in genomic studies, even with biobank-level sample sizes. The performance differences between DL and non-deep ML decrease as the sample size of data increases. This suggests when the sample size of data is significant, further increasing sample sizes leads to more performance gain in DL methods. Hence, DL methods could be better if we analyze genomic data bigger than this study. / Graduate

Strawberry production in California is limited by plant diseases such as Macrophomina root rot (caused by Macrophomina phaseolina) and Botrytis fruit rot (BFR) (caused by Botrytis cinerea). Current disease management strategies are compromised due to fumigant regulations or ineffective disease management practices. This thesis investigated methods to potentially improve the management of these two diseases. Host plant resistance evaluations for Macrophomina root rot were conducted for the 2020-2021 and 2021-2022 growing seasons. Fifty-one strawberry genotypes were screened in two field experiments where plants were inoculated artificially with Macrophomina phaseolina in both seasons. A wide range of plant resistance to Macrophomina root rot was observed. The three most resistant genotypes based on final plant mortality were ‘17C721P606’, ‘Yunuen’, and ‘Xareni’ in 2020-2021; ‘UCD Mojo’, ‘Mariposa’, and ‘Dayana’ in 2021-2022. A summary of similar experiments done in the previous four years showed ‘Osceola’ as highly resistant. Disease severity varied among years for specific genotypes as well as the average final mortality for all genotypes in the experiments. Strong positive associations were found for soil temperature during the first month after planting (R2= 0.79, P2= 0.79, P A survey of BFR levels in commercial strawberry fields with and without fungicide applications was conducted in Santa Maria, CA in 2021 and 2022. Weather stations were installed at each field to collect leaf wetness duration and temperature data and calculate the BFR risk factor based on the Strawberry Advisory System (StAS) developed at the University of Florida. There were no statistically significant differences between fungicide and no-fungicide treatments for both in-field and postharvest BFR incidence in 2021 and in-field BFR incidence in 2022, while no-fungicide treatment showed higher postharvest BFR incidence in 2022. BFR levels were low in both years. In 2021, average in-field BFR incidence for fungicide and no-fungicide treatments were 2.6 ± 0.3% and 2.5 ± 0.4%, respectively. Average postharvest BFR incidence for fungicide and no-fungicide treatments were 1.8 ± 0.2% and 2.0 ± 0.3%, respectively. In 2022, average in-field BFR incidence for fungicide and no-fungicide treatments were 3.0 ± 0.4% and 3.7 ± 0.4%, respectively. Average postharvest BFR incidence for fungicide and no-fungicide treatments were 0.6 ± 0.1% and 1.5 ± 0.2%, respectively. Risk factor from StAS was significantly associated with BFR incidence in 2021, but not in 2022. Screening new strawberry genotypes against Macrophomina root rot should be ongoing as part of a standard process for determining the susceptibility of currently grown and potentially new cultivars. Additional research under more diverse weather conditions is necessary to verify the impacts of reducing fungicide use in BFR management and to validate the use of StAS in making fungicide use decisions in California fields.

At present, the global population is increasing, while soil and fresh water resources for crop production are declining. It is important to adopt sustainable practices to optimise the use of limited natural resources without compromising the environment, and to enhance continuous production in the long term. The rapid growth of UK strawberry industry has been achieved through the precision use of varieties, nutrients and polythene tunnels. This intensive production has caused significant environmental impacts especially Greenhouse Gas (GHG) emissions from the production. Strawberry powdery mildew (Podosphaera aphanis) is a major fungal disease affecting strawberry production worldwide particularly in polythene tunnels. The disease can result in yield losses of up to 70% of the crop. A ruleQbased system was used in the field trials to predict high risk days of P. aphanis development, taking into account the optimal environmental conditions conducive to conidial germination and disease development. The results (Chapter 3) showed that the use of this prediction system achieved satisfactory control of P. aphanis in commercial strawberry production, with reduced fungicide applications compared with commercial spray programme. The results were consistent in two consecutive years and on different varieties. In addition, it was suggested that the use of the prediction system may also lead to lower GHG emissions associated with fewer fungicide applications, thereby benefit strawberry growers both environmentally and economically. Results from 2014 & 2015 silicon fertigation trials showed that the use of a silicon nutrient via the fertigation system reduced the strawberry susceptibility to P. aphanis and twoQspotted spider mites (Tetranychus urticae Koch) in two consecutive years on different varieties (Chapter 4). In both years, crops received the silicon nutrient only without fungicides had both lower rate of epidemic (r) and lower value of Area Under the Disease Progress Curve (AUDPC) (r = 0.0036, AUDPC = 475 in 2014; r = 0.001, AUDPC = 267 in 2015) compared with the untreated control (r = 0.0042, AUDPC = 662 in 2014; r = 0.0011, AUDPC = 281 in 2015). Silicon also delayed the epidemic buildQup in the silicon nutrient only treatment for approximately two weeks compared with the untreated control. Crops from the silicon nutrient plus fungicides treatment had lower susceptibility (r = 0.0012 in 2014; r = 0.0004 in 2015) than those from the fungicides only treatment (r = 0.0017 in 2014; r = 0.0005 in 2015) suggesting that the silicon nutrient may also enhance fungicides performance in reducing the epidemic buildQup when used together. Moreover, the presence of T. urticae on strawberry leaves was significantly lower (P < 0.001) in plants treated with the silicon nutrient than those without. In addition, initial results suggested that silicon may play a positive role in raising °Brix of strawberry leaf petiole, improving pollen viability, and influencing the length of flower receptacle and stamens. Maltmas Farm has a wide range of semiQnatural habitats that provide food and nesting resources for wild pollinators. Hoverflies, bumblebees and solitary bees were found to be the main wild pollinators that pollinate commercial strawberries at Maltmas Farm (Chapter 5). The number of pollinators in tunnels or open fields significantly correlated with the abundance of strawberry flowers (P < 0.05). Pollinator presence also differed between groups throughout the day and over the seasons. Hoverflies appeared early in the day and were abundant in summer months; bumblebees and solitary bees were present most of the day and throughout the season, whereas honeybees were only active in sunny days. Temperatures, relative humidity and cloud coverage also affected pollinator presence. In addition, pollinator activity was not significantly (P > 0.05) affected by the application of the silicon nutrient via the fertigation system. The integrated use of the prediction system (to reduce fungicide applications and subsequent GHG emissions), the silicon nutrient (to reduce crop susceptibility to P. aphanis and T. urticae), and sustainable farmland management (to encourage the presence of wild pollinators) could help strawberry growers to achieve a more sustainable production.

Millions of people in the United States alone suffer from undiagnosed or late-diagnosed chronic diseases such as Chronic Kidney Disease and Type II Diabetes. Catching these diseases earlier facilitates preventive healthcare interventions, which in turn can lead to tremendous cost savings and improved health outcomes. We develop algorithms for predicting disease occurrence by drawing from ideas and techniques in the field of machine learning. We explore standard classification methods such as logistic regression and random forest, as well as more sophisticated sequence models, including recurrent neural networks. We focus especially on the use of medical code data for disease prediction, and explore different ways for representing such data in our prediction algorithms.

This thesis contains a review of previous work in the fields of encoding sequential healthcare data and predicting graft- versus- host disease, a medical condition, based on patient history using machine learning. A new encoding of such data is proposed for machine learning purposes. The proposed encoding, called bag of binned weighted events, is a combination of two strategies proposed in previous work, called bag of binned events and bag of weighted events. An empirical experiment is designed to evaluate the predictive performance of the proposed encoding over various binning windows to that of the previous encodings, based on the area under the receiver operating characteristic curve (AUC) metric. The experiment is carried out on real- world healthcare data obtained from Swedish registries, using the random forest and the logistic regression algorithms. After filtering the data, solving quality issues and tuning hyperparameters of the models, final results are obtained. These results indicate that the proposed encoding strategy performs on par, or slightly better than the bag of weighted events, and outperforms the bag of binned events in most cases. However, differences in metrics show small differences. It is also observed that the proposed encoding usually performs better with longer binning windows which may be attributed to data noise. Future work is proposed in the form of repeating the experiment with different datasets and models, as well as changing the binning window length of the baseline algorithms. / Denna avhandling innehåller en recension av tidigare arbete inom områden av kodning av sekventiell sjukvårdsdata och förutsägelse av transplantat- mot- värdsjukdom, ett medicinskt tillstånd, baserat på patienthistoria med maskininlärning. En ny kodning av sådan data föreslås i maskininlärningssyfte. Den föreslagna kodningen, kallad bag of binned weighted events, är en kombination av två strategier som föreslagits i tidigare arbete, kallad bag of binned events och bag of weighted events. Ett empiriskt experiment är utformat för att utvärdera den föreslagna prestandan för den föreslagna kodningen över olika binningfönster jämfört med tidigare kodningar, baserat på AUC- måttet. Experimentet utförs på verkliga sjukvårdsdata som erhållits från svenska register, med random forest och logistic regression. Efter filtrering av data, lösning av kvalitetsproblem och justering av hyperparametrar för modellerna, erhålls slutliga resultat. Dessa resultat indikerar att den föreslagna kodningsstrategin presterar i nivå med, eller något bättre än bag of weighted events, och överträffar i de flesta fall bag of binned events. Skillnader i mått är dock små. Det observeras också att den föreslagna kodningen vanligtvis fungerar bättre med längre binningfönster som kan tillskrivas dataljud. Framtida arbete föreslås i form av att upprepa experimentet med olika datamängder och modeller, samt att ändra binningfönstrets längd för basalgoritmerna.