Machine learning algorithm identifies undiagnosed episodes of pertussis

2023, 06 July

| 2 min read|
Grégoire Dugast
Thumbnail for the article: Estimating the pertussis burden in adolescents and adults in the US - QH 2023
Experts at Quinten Health have developed a supervised machine learning algorithm to identify undiagnosed episodes of pertussis in adolescents and adults reporting acute respiratory illness. As pertussis is often misidentified, misdiagnosed or misreported, it is complex to assess its incidence rate in the above-mentioned populations. The aim was therefore to better estimate the incidence in these populations, and to assess the impact of chronic comorbidities on the probability of contracting pertussis or a more severe form of the disease.


Considered a vaccine-preventable childhood disease, pertussis remains a persistent problem at all ages. Immunity conferred by vaccination or previous infection wears off over time. If booster vaccinations are not given regularly, adolescents and adults may be vulnerable to infection. They could thus contribute to the reappearance of the disease and the emergence of periodic epidemics.

In a database of electronic medical records containing more than 80 million patients across the United States, the inclusion of pertussis episodes identified by the algorithm developed by Quinten Health multiplied the estimated incidence of this pathology by 110 on average. This result highlights the extent of potential underestimation and/or lack of knowledge of the disease in the population evaluated. The team also showed that asthma, immunodeficiency, chronic obstructive pulmonary disease (COPD), obesity, Crohn’s disease and diabetes (types 1 and 2) are comorbidities that increase the risk of contracting pertussis.

The strength of this study lies in its use of a large, geographically-diverse population sample and comprehensive, real-world data representative of actual clinical practice. Its results will help to better inform public health vaccination and booster programs.

The paper “Estimating the pertussis burden in adolescents and adults in the United States between 2007 and 2019” was written by Denis Macina, Sachin Mathur, Maryia Dvareskaya, Sam Ekhtiari, Paul Hayat, Martin Montmerle and Chathuri Daluwatte. It was published in Human Vaccines & Immunotherapeutics in May 2023.


latest articles