Personalized Digital Health and Artificial Intelligence in Childhood Asthma

Asthma in Children

Digital stethoscope;, digital health, machine learning, Asthma

Asthma is a chronic inflammatory disease of the airways that causes recurrent episodes of wheezing, breathing difficulties and coughing. The prevalence of asthma is 8% in school-aged children and 30% in preschoolers, making asthma the first chronic disease in children. Symptoms are due to diffuse but variable airway obstruction, reversible spontaneously or after inhalation of beta2 agonists (β-2a) such as salbutamol. Exacerbations of asthma are frequent and difficult to assess by parents and the patient himself. It is estimated that approximately 2.5% of children with asthma are hospitalized annually. The global burden caused by asthma can thus be reduced by improving early detection of bronchial obstruction, prescribing immediate treatment with the appropriate background therapy, and reliably and objectively assess response to treatment. The natural history of asthma symptoms in children shows a great intra and inter-individual variability. The difficulty of assessing the severity of an attack by the parents or the child himself, when he is old enough to control his chronic disease, is a key element in the management of asthma and allows the treatment to be adapted quickly, sometimes avoiding hospitalization. Healthcare professionals can assess the severity of the episode using the Pediatric Respiratory Assesment Measure (PRAM) score, which has the advantage of being adaptable at any age. The Global Alliance against Chronic Respiratory Diseases (GARD) integrates in its diagnostic strategy for chronic respiratory diseases, the lung function test, which allows the quantification of respiratory function in the context of diagnosis and long-term follow-up. Although spirometry are non-invasive tests, they still require a high level of patient cooperation, which remains problematic before the age of 7 years. The digital stethsocope integrates a capacity for recording auscultations and data transmission to high-performance software. This has made it possible to extend auscultation beyond what was audible to the human ear alone (over 20-20,000 Hertz).Auscultatory sounds analysis, particularly those most often associated with obstructive syndrome could be simple, reproducible and a reliable method of assessing the severity and response to treatment in children's asthma. Major advances in signal processing and unsupervised learning in artificial intelligence research provide the potential for high-performance analysis of physiological measures.

Aim: Develop an artificial intelligence based algorithm for unsupervised diagnostic and classification of childhood asthma exacerbation. Methodology: A Longitudinal prospective monocentric observational study will be performed in the Pediatric Emergency Division (PED) and the Pediatric Respiratory Unit (PRU) of the Geneva University Hospitals (HUG) during 24 months. This clinical study will include patients aged from 2 to16 years with acute asthma exacerbations. The intervention consists in recording auscultation of asthmatic patients at rest, during acute exacerbation and after treatment by bronchodilatators (β-2 agonists) inhalation in the PED, with a Digital Stethoscope (DS). Auscultation will be recorded during hospitalization every day, at home 7 days after the acute episode, combining intdoor and outdoor measures, and evaluating the exposome. A last record will be done at 6 to 8 weeks after the acute episode, with a lung function test if the patient is up to 7 years. A validation and training audio database will be constituted for the development of Artificial Intelligence (AI) algorithms, allowing analysis of respiratory rate, inspiratory/expiratory time ratio, PRAM score, wheezing variation of intensity and unsupervised diagnosis. Expected results: Creation of a performant AI algorithm for unsupervised acute asthma exacerbation diagnosis, with \> 70 % of Sensitivity and \> 70% of Specificity compared to the expert. Response to treatment will improve patient empowerment and personalized medicine in childhood asthma management.

No related papers found yet.

None listed

Switzerland