Analysis Group Collaborators Use LASSO-Based Machine Learning Model to Identify Predictors of Biologic Therapy Initiation Among Patients with Severe Asthma
September 29, 2026
Biologic therapies can improve symptom control and quality of life among patients with severe asthma, and clinical guidelines recommend their use. Despite this, the number of eligible patients initiating these therapies remains low. To investigate this gap, Analysis Group Principal François Laliberté, Vice President Guillaume Germain, and Manager Sean MacKnight collaborated with John J. Oppenheimer from Rutgers New Jersey Medical School and Tom Corbridge, Alexandra Stach-Klysh, and Arijita Deb from GSK to identify and quantify factors associated with biologic initiation among patients with severe asthma.
Using a health care claims database that has census-level representation across US patient populations and insurers, the researchers used LASSO regression – a machine learning model – to identify predictors for initiating biologics across a broad range of demographically and clinically distinct patients.
In a publication detailing their research in Therapeutic Advances in Respiratory Diseases, the authors report that biologic initiation was less likely among younger patients, those whose care was managed by a primary care physician, and those who were prescribed high-dose inhaled corticosteroids or antibiotics. They also found that biologic initiation was more likely, among other factors, in patients with signs of more severe respiratory disease (e.g., upper respiratory tract infection, pneumonia, eosinophilic disorders) and in those who were prescribed systemic corticosteroids, leukotriene modifiers, or single-inhaler triple therapy.
The authors conclude that “while asthma guidelines dictate when biologics should be prescribed, many eligible patients are not receiving them in a timely manner.”