Introduction Mobile health is increasingly empowering patients to monitor their disease. Commercially-available digital peak flow meters are used by asthma patients to self-monitor their condition and claim to predict deteriorations in peak flow measurements. Thus, evaluation of their ability to provide accurate peak flow predictions is essential.

Aim The aim of this study was to assess the accuracy of a digital peak flow meter in predicting next-day peak flow measurements.

Methods The digital peak flow meter connects to the patient‘s smartphone, allowing the patient to upload their peak flow measurements to a mobile application (figure 1). Peak flow measurements are divided into three zones: ‘Green’ (>80% of the patient’s best peak flow measurement), ‘Yellow’ (60–80%) and ‘Red’ (<60%). The mobile application uses deep learning neural networks to analyse the patient’s peak flow measurements over the last 2 weeks, to predict which peak flow zone the patient will be in tomorrow. Peak flow measurements and predictions for a random 2-month period were analysed.

Results Between February-April 2022, 23,485 peak flow zone predictions were made. The average patient age was 26.1 years and 51.5% (13,646/26,485) of the peak flow measurements were performed by females. Patients originated from 24 countries. 92.0% (21,655/23,485) of next-day peak flow zone predictions were correct. The algorithm’s average forecasted probability of predicting the correct peak flow zone was 94.0 ± 8.6%. The average peak flow measurement was 93.3 ± 17.6% of the patient’s personal best.

• Download figure
• Open in new tab
• Download powerpoint

Abstract S118 Figure 1

Screenshot from the mobile application’s clinician dashboard, showing the patient data which can be uploaded to the mobile application and reviewed by clinicans

Conclusions Digital peak flow monitoring using machine learning algorithms can predict next-day peak flow measurements with high accuracy, thus alerting patients to impending deteriorations in their asthma control. Further work is needed to assess whether patients tailor their asthma self-management based on these predictions. Patient perspective on digital peak flow monitoring should also be explored to determine whether digital peak flow monitoring can be an acceptable alternative to traditional peak flow monitoring.

Please refer to page A212 for declarations of interest related to this abstract.