Related Experiment Video
Updated: Sep 2, 2025
Asthma Detection Research Based on Voice Signal Processing and Machine Learning
Published on: July 22, 2025
The external validation of the asthma prediction tool in the French ELFE cohort
Sadia Khan1, El Hassane Ouaalaya1, Jonviea D Chamberlain2,3
1Bordeaux Population Health Research Center, EPICENE Team, INSERM, UMR 1219, Bordeaux University, Bordeaux, France.
Insights
A noninvasive asthma prediction tool (APT) effectively identified children at high risk for asthma in a French cohort. This tool shows promise for primary care use, aiding early asthma detection in children.
Area of Science:
- Pediatric Pulmonology
- Epidemiology
- Biostatistics
Background:
- Current asthma prediction scores for children often require invasive procedures, limiting their use in primary care.
- The Leicestershire Respiratory Cohort (LRC) developed a noninvasive Asthma Prediction Tool (APT) with promising initial results.
- External validation of the APT in a large French general population cohort (ELFE) was conducted.
Purpose of the Study:
- To externally validate the noninvasive Asthma Prediction Tool (APT) in the French general population.
- To assess the performance and generalizability of the APT for identifying children at risk of developing asthma.
- To evaluate the APT's utility in a primary care setting.
Main Methods:
- The APT was applied to children in the Étude Longitudinale Française depuis l'Enfance (ELFE) cohort at Age 1.
- Asthma was defined at Age 5 by parental report of "asthma ever" or "wheezing in the past 12 months".
- Logistic regression, receiver operating characteristic (ROC) curves (AUC), Brier scores, and Hosmer-Lemeshow tests were used for performance and calibration assessment.
Main Results:
- The APT categorized 84.9% of 10,689 children as low-risk, 13.1% as medium-risk, and 2% as high-risk for asthma at Age 5.
- Children in the high-risk category were 13 times more likely to develop asthma (OR=13.8, 95% CI=10.2-18.8).
- The APT demonstrated comparable performance in the ELFE cohort (AUC=0.68, Brier score=0.14) to the LRC cohort (AUC=0.74, Brier score=0.16), indicating robust discriminative ability and calibration.
Conclusions:
- The external validation confirms the APT's effectiveness in identifying children at risk for asthma in a general French population.
- The noninvasive nature and robust performance of the APT support its potential application by general practitioners for early asthma detection.
- The study provides further evidence for the APT's comprehensive measure of effectiveness, encouraging its use in primary care settings.
Background:
Existing predictive scores for early identification of children at high risk of developing asthma include invasive procedures, and hence have limited utility in a primary care setting. The Leicestershire respiratory cohort (LRC) has developed a noninvasive asthma prediction tool (APT) for children with promising results. We aimed to perform its external validation in the French general population Étude Longitudinale Française depuis l'Enfance (ELFE) cohort.
Methods:
Predictive scores were determined at Age 1 and the primary outcome of asthma was defined as parental reporting of "asthma ever or "wheezing in the past 12 months" at Age 5. Logistic regression was used to calculate the odds ratio (OR) and performance measures, and discriminative performance was reported using the receiver operating curve and area under curve (AUC). Calibration was assessed using Hosmer-Lemeshow goodness-of-fit test and visualized with a calibration plot. Overall performance was determined using Brier scores.
Results:
Of the 10,689 children analyzed: 84.9% were at low, 13.1% medium, and 2% at high risk of developing asthma at Age 5. Children in the medium-risk category were three times more likely to develop asthma (OR = 3.3, 95% confidence interval [CI] = 2.97-3.78) whereas 13 times more likely in the high-risk category (OR = 13.8, 95% CI = 10.2-18.8). The tool's AUC was comparable: LRC 0.74 versus ELFE 0.68; as were the Brier scores LRC 0.16 versus ELFE 0.14. The tool's performance was robust to changes in inclusion criteria and outcome definitions.
Conclusions And Relevance:
Results of the present study and previous validation studies performed in high-risk populations provide a comprehensive measure of the effectiveness of the APT, providing encouragement for its application by general practitioners.
Related Concept Videos
01:30Asthma-IV: Diagnostic and Management
Clinical Assessment for Asthma:
This is the first step in diagnosing and managing asthma. It includes:
01:24Asthma-III: Symptoms and Complications
Classification of Asthma
01:26Asthma-II: Pathophysiology and Classification
Additionally, environmental and genetic factors play crucial roles in determining an individual's susceptibility to asthma and the severity of their condition.
Critical processes in asthma pathophysiology include:
01:29Asthma-I: Introduction
01:20Asthma: Pathogenesis and Management
Asthma is classified as allergic and non-allergic. Allergens such as dust mites, pollen, and pet dander trigger allergic asthma, while factors like cold air, intense emotions, or exercise can induce non-allergic asthma.
01:19Antiasthma Drugs: Leukotriene Modifiers
Leukotriene modifiers work through two distinct mechanisms: