Log In

Forgot Password?
Create New Account

Loading... please wait

Abstract Details

Machine Learning Analysis of Anosmia and Ageusia after COVID-19 Vaccination: A Vaccine Adverse Event Reporting System (VAERS) Study
General Neurology
P10 - Poster Session 10 (8:00 AM-9:00 AM)
2-002

A/A was reported after COVID-19 vaccination in adults.

To investigate whether there is an association between Anosmia and Ageusia (A/A) and COVID-19 vaccination in adults.

The reporting rate of A/A cases after COVID-19 vaccination was compared to the rate of cases after other vaccinations in three periods: the vaccine period (December 2020-July 2021); the pre-COVID-19 vaccine period (April 2020-November 2020) and the pre-COVID-19 period (January 2019-August 2019). Self-controlled case series analysis and case-centered analysis was used. The risk period of probable association was defined as six weeks after vaccination. Machine Learning was performed using the random forest method with ROC curve validation.

1852 and 5 patients with A/A were reported after COVID-19 vaccination and other vaccinations during the vaccine period, respectively. The reporting rate of A/A after COVID-19 vaccination, was significantly higher compared to A/A after other vaccinations (9.5 vs 0.021 per 1 million p<0.0001), however, the rate was still within the range of incidence of A/A in the general population. Only 20 and 14 cases of A/A were reported during the pre-vaccine and pre-COVID-19 period, respectively. Using self-controlled and case centered analyses, there was a significant difference in the reporting rate of A/A between risk period and control period (89% vs 4-6% p < 0.00001). The reporting rate of A/A was not different between manufacturers. Preliminary data from Machine Learning/Confusion Matrix showed 94.97% accuracy to classify prediction of hospitalization using a binary category.

There is no significant increase in the reporting rate of A/A after COVID-19 vaccination compared to the incidence in the general population. Although the reporting rate of A/A after COVID-19 vaccination was significantly higher during the risk period it was in the expected incidence range. Machine learning techniques found that patient age and time between vaccinations and reporting of A/A are important predictive factors of hospitalization.

Authors/Disclosures
Kranthi K. Mandava
PRESENTER
Mr. Mandava has nothing to disclose.
Mustafa Jaffry Mr. Jaffry has nothing to disclose.
No disclosure on file
Kazim Jaffry Mr. Jaffry has nothing to disclose.
No disclosure on file
Nizar Souayah, MD, FAAN (NJMS) Dr. Souayah has received publishing royalties from a publication relating to health care.