Date of Award

3-28-2016

Document Type

Honors Thesis

Department

Engineering & Computer Science

First Advisor

Rodney L. Summerscales

Abstract

Inexpensive paper-based biosensors can be valuable screening tools to test for various illnesses, but it is often challenging to design them to produce a visual change that can easily be identified by untrained users. This research examines one method of compensating for the lack of distinct visual cues by developing and testing a mobile application that uses a machine learning algorithm (k-Nearest Neighbors) to analyze a picture of a sensor and determine whether it shows a positive or negative result. The machine learning algorithm was trained on a set of labeled sensor images and k-fold cross-validation was used to analyze its classification accuracy.

Share

COinS