|
|
Hokkaido University Collection of Scholarly and Academic Papers >
Theses >
博士 （獣医学） >
The Analysis of Infectious Diseases via Machine Learning
|
Files in This Item:
|
|
|
Related Items in HUSCAP:
|
| Title: | The Analysis of Infectious Diseases via Machine Learning |
| Other Titles: | 機械学習による感染症の解析 |
| Authors: | TESSMER, HEIDI LYNN Browse this author |
| Issue Date: | 29-Jun-2018 |
| Publisher: | Hokkaido University |
| Abstract: | This thesis introduces two projects applying machine learning methods to the realm of bioinformatics. In Chapter 1, we look at a regression problem involving the parameter values associated with the SEIR pidemiological model while in Chapter 2 we explore viral host classification. Chapter 1 - To estimate and predict the transmission dynamics of respiratory viruses, the estimation of the basic reproduction number, R0, is essential. Recently, approximate Bayesian computation methods have been used as likelihood free methods to estimate epidemiological model parameters, particularly R0. In this paper, we explore various machine learning approaches, the multi-layer perceptron, convolutional neural network, and long-short term memory, to learn and estimate the parameters. Further, we compare the accuracy of the estimates and time requirements for machine learning and the approximate Bayesian computation methods on both simulated and real-world epidemiological data from outbreaks of influenza A(H1N1)pdm09, mumps, and measles. We find that the machine learning approaches can be verified and tested faster than the approximate Bayesian computation method, but that the approximate Bayesian computation method is more robust across different datasets. Chapter 2 - Infectious diseases which transfer between species are particularly difficult to manage. Knowing the natural host for an infectious agent makes it easier to prevent interspecies transmissions. However, with new and re-emerging disease, it can be difficult to know what the reservoir host is. In the second half of this thesis, we conducted a principal component analysis using data from the fruit bat and wild duck, along with a selection of single-stranded RNA viruses found in each animal. Historically, the virus-host relationship has often been examined using two components, that is, the G+C content of the genomes and the rate ratio of CpG in the genome. However, numerous data discrepancies exist which cannot be explained with mathematical models built from this technique. In this study, we found several alternative components that could be used to infer the host animal species of RNA viruses. Using these alternative components, we may be able to build a mathematical model that more closely simulates the virus-host genetic relationship. With this information, we may be able to identify genetic signatures in viruses which can uniquely identify the natural host species. In future, this information could help identify the animal source of a new outbreak. |
| Conffering University: | 北海道大学 |
| Degree Report Number: | 甲第13258号 |
| Degree Level: | 博士 |
| Degree Discipline: | 獣医学 |
| Examination Committee Members: | (主査) 教授 鈴木 定彦, 教授 高田 礼人, 准教授 瀧川 一学 （情報科学研究科）, 准教授 小柳 香奈子 （情報科学研究科）, 助教 大森 亮介 |
| Degree Affiliation: | 獣医学研究科（獣医学専攻） |
| Type: | theses (doctoral) |
| URI: | http://hdl.handle.net/2115/71247 |
| Appears in Collections: | 学位論文 (Theses) > 博士 （獣医学）
課程博士 (Doctorate by way of Advanced Course) > 獣医学院(Graduate School of Veterinary Medicine)
|
|
