Date of Award

2023

Degree Type

Dissertation

Degree Name

Doctor of Philosophy in Civil and Environmental Engineering

Department

Civil and Environmental Engineering

First Advisor

Ali Shafqat Akanda

Abstract

Over 2 billion people globally are living in water-stressed regions. With population increase, poor management of natural resources, and a changing climate, water scarcity has become a widespread phenomenon in today’s world. As a result, waterborne diseases remain a major global health problem due to a lack of safe water supply and sanitation facilities. A large part of the global population is still at risk of having waterborne and water-related diarrheal diseases despite many scientific advancements. Meanwhile, Climate Change is exacerbating waterborne disease patterns and trends, as seen by the large increase in diarrheal cases globally. It is thus imperative to identify the links between hydroclimatic variables and diarrheal diseases (i.e., Acute Watery Diarrhea (AWD), cholera) to better understand the trigger and transmission process of these diseases and empower intervention protocols.

In a post-disaster setting, triggers, and transmission of diarrheal diseases -- cholera could be very fast due to extreme hydroclimatic variables and destruction of existing water supply and sanitation system; whereas, in an endemic setting, it is more related to hydroclimatic processes and seasonal outbreak patterns. The influence of hydroclimatic variables also changes with different spatial scales as different regions may have different hydroclimatic patterns. Therefore, in forecasting diarrheal disease risk, it is important to identify hydroclimatic variables that are most influential diarrheal outbreaks in any region or spatial scale. It is challenging, time consuming and costly to get access to hydroclimatic data from weather stations, whereas freely accessible Remote Sensing data with high spatiotemporal resolution and global coverage can be a solution to this. Another challenge is to disseminate this information to people living in remote areas. An early warning system that disseminates information through a smartphone application could be beneficial for people to prepare for any upcoming surge of diarrheal or cholera cases.

This dissertation encompasses identifying impactful remotely sensed hydroclimatic variables to forecast waterborne diarrheal disease risk in post-disaster and endemic scenarios in underdeveloped and developing countries using Geospatial Modeling and Machine Learning approaches, where the impact of spatial scale change in forecasting and an advanced early warning disease forecasting dissemination tool are also included. Machine Learning models are able to ingest large earth observations and environmental datasets and identify the importance of variables that lead to cholera risk. In the weighted sum geospatial modeling approach, each variable is then classified to multiple cholera risk layers based on ranked risk, and correlation values. The geospatial modeling needs researcher’s contribution in selecting variables, assigning classes and weightage to get the forecasted cholera risk layers, whereas the machine learning approach does not require human contribution. This combined machine-human approach can potentially be used in different locations with similar hydroclimatic backgrounds and similar imperative interventions to develop cholera early warning systems. Forecasting diarrheal disease risk and disseminating through innovative use of smartphone applications -- an effective early warning system can reduce the disease burden by creating awareness among people in advance in addition to providing additional time to manage to supply required interventions - vaccine, oral saline, medicine, and improved water and sanitation facilities.

Download

Available for download on Thursday, September 05, 2024

Share

COinS