Development of an Asthma Exacerbation Risk Prediction Model for Conversational Use by Adults in England.

Area of Science:

• Health Informatics
• Clinical Epidemiology
• Public Health

Background:

• Asthma exacerbations pose significant risks to patients and increase healthcare costs.
• Accurate risk assessment and behavioral interventions are crucial for asthma management.
• Personalized risk prediction models can enhance patient care and reduce healthcare burdens.

Purpose of the Study:

• To develop a personalized risk prediction model for asthma exacerbations.
• To utilize routine healthcare data for risk assessment.
• To integrate risk modeling into automated conversational systems for patient support.

Main Methods:

• Utilized pseudonymized primary care electronic health records from the Clinical Practice Research Datalink (CPRD) Aurum database.
• Employed logistic regression incorporating variables such as age, gender, ethnicity, socioeconomic status, and clinical asthma history.
• Validated the model across three temporal cohorts totaling over 1.2 million patients.

Main Results:

• The full model achieved an Area Under the ROC Curve (AUC) of 0.72, with a restricted model at 0.71.
• A simplified model using easily self-reported variables demonstrated comparable predictive performance.
• Identified key predictors of exacerbation including older age, female gender, medication history, comorbidities, and socioeconomic deprivation.

Conclusions:

• A moderately accurate model for predicting asthma exacerbations within 3 months was developed using routine electronic healthcare record data.
• A simplified model, suitable for self-reporting via platforms like WhatsApp, showed no substantial difference in predictive performance.
• This approach facilitates personalized risk assessment and potential interventions for asthma exacerbations.