Mathematical Analysis of Signals and Clinical Parameters Provided by Non-invasive Home Ventilation Devices

SAGE-NIV: Surveillance and Artificial Intelligence Guidance for Exacerbations in COPD Patients With Home Non-Invasive Ventilation

This study will look at people with COPD who use a home breathing machine called non-invasive ventilation (NIV). NIV machines collect information about your breathing, such as air flow, pressure, and mask leaks.

Researchers want to use a computer program, called artificial intelligence (AI), to study this information. The goal is to find early signs that your breathing may be getting worse.

People with COPD who already use NIV at home may join this study. The study does not change your treatment. It only uses the breathing data already recorded by your NIV machine.

The computer program will look for patterns in the data. These patterns may help doctors:

Notice early warning signs of a COPD flare-up Find problems with how you and the machine work together Improve the way NIV is monitored at home The main goal is to create a tool that helps patients and doctors manage home NIV more easily and more safely.

This study proposes the development of an artificial intelligence (AI) system to monitor and analyse detailed non-invasive mechanical ventilation (NIV) data in COPD patients, with the aim of predicting clinical exacerbations and improving home management.

Analysis of data from home NIV devices allows assessment of patient compliance, detection of leaks and asynchronies, and monitoring of upper airway events. However, the potential of these data to improve ventilation management in COPD patients has been limited, in part due to the lack of tools to process and interpret the detailed records. Transforming these data into an open format opens up the possibility of applying artificial intelligence to analyse large amounts of information and develop predictive models.

The multi-centre, observational, longitudinal study design will include COPD patients on NIV therapy who meet adherence criteria. Detailed leak, pressure and flow time data, previously decrypted and converted into a data format readable by analysis software, will be analysed. The identified metrics will be evaluated by machine learning algorithms using techniques such as random forest and neural networks.

Expected outcomes include the development of an automated predictive model to enable early detection of exacerbations and improved patient-ventilator synchronisation, moving towards more efficient and personalised telemonitoring in home NIV management.

Inclusion Criteria:

• Age between 40 and 80 years.
• COPD diagnosed by pulmonary function tests.
• Home NIV therapy with good adherence (minimum daily compliance > 5 hours) for at least 6 months.
• Users of the ResMed LUMIS 150 ventilator. This is due to the presence of the decoding tool and a larger storage capacity (more than 100 nights) in the removable device of the ventilator.
• Acute exacerbation requiring hospital admission or home care.

Exclusion Criteria:

• Lack of informed consent.
• Previous clinical instability defined by the need for antibiotics and/or systemic corticosteroids in the two months prior to the inclusion exacerbation, excluding the 48 hours prior to admission, as this was considered part of the inclusion clinical picture.