Smarter prescription of azithromycin using machine learning, with Prof. Wim Janssens

Prophylactic use of azithromycin is common in the management of COPD patients with frequent acute exacerbations. But its application is conservative due to concerns around adverse events, interactions with other drugs, and bacterial resistance. Given these concerns, it is important to identify patients who are likely to be responders or non-responders.

According to the NICE guidance in the UK, you should consider azithromycin if the patient:

• Doesn't smoke.
• Has already accounted for non-drug treatments, inhaled therapies and vaccines.
• And has one or more of the following kinds of exacerbations:
  • Frequent exacerbations, more than four times a year, with high sputum production
  • Prolonged exacerbations with high sputum production.
  • Exacerbations causing hospitalisation.

In this episode, Thorax social media editor Dr. Kate Diomede speaks to Prof. Wim Janssens¹, author of a study proposing five parameters to predict the individual treatment effect of azithromycin. These parameters were developed using a machine learning model trained using the MACRO and COLUMBUS datasets. This led to the creation of an online calculator, serving as an aid in decision making.

 

Read the paper: Identifying azithromycin responders with an individual treatment effect model in COPD

 

1. Laboratory of Respiratory Diseases and Thoracic Surgery (BREATHE), KU Leuven, Leuven, Belgium

 

Please engage in the conversation through the social media channels (Twitter - @ThoraxBMJ; Facebook - Thorax.BMJ) and subscribe on your preferred platform, to get the latest episodes directly on your device each month. We would love to hear your thoughts on the podcast, you can leave a review on Apple Podcasts or Spotify.