Improving Bacterial Ear Infection Diagnostics with an Image Classification Algorithm

Author

Wanying Li

Date of Award

2017

First Advisor

Michael Bergman

Second Advisor

Jackson Liscombe

Abstract

Acute otitis media (AOM), or ear infection, is the most common childhood infections. It accounts for 60% of antibiotic prescriptions in children, making it the main reason for the use of antibiotics in children [1]. About 60% of cases of ear infection are caused by bacteria, 40% are viral or a combination of both [2]. Only severe bacterial ear infection requires antibiotics. Visual examination through an otoscope is the primary but inaccurate method for the diagnosis of bacterial ear infection. The inaccuracy in diagnosis leads to two main consequences: 1) children may experience the buildup of ear pain and discomfort, which will result in prolong treatment time; 2) based on a survey that asked pediatricians whether or not they would prescribe antibiotics based on different patient profiles, 45% of pediatricians prophylactically administer antibiotics. It is estimated that 50% of antibiotic prescriptions for ear infections are unnecessary and potentially lead to a weaker immune system, expose the child to subsequent episodes of ear infection, and further increase the likelihood of antibiotic resistance [3, 4]. The goal is to come up with a rapid, effective, and non-invasive solution that will improve the accuracy of diagnosis of bacterial ear infection in order to decrease the unnecessary use of antibiotics and potentially the treatment time.

Recommended Citation

Li, Wanying, "Improving Bacterial Ear Infection Diagnostics with an Image Classification Algorithm" (2017). Senior Theses. 1159.
https://digitalcommons.bard.edu/sr-theses/1159