A recent study published in the BMC Medical Informatics and Decision Making journal entitled “Predicting asthma control deterioration in children” reveals that researchers from the University of Utah have developed a model that can anticipate deterioration of asthma in children.

Asthma is a chronic inflammatory disease characterized by narrowed and swollen airways due to mucus accumulation in the lungs. Patients with asthma may experience difficulties in breathing, coughing, frequent respiratory infections, and chest tightness. It is believed that asthma is caused by a combination of genetic and environmental factors like allergens, pollutants, dust, and cigarette smoking, among others.

From a therapeutic standpoint, there is no cure for asthma, but several treatment options are available to manage the disease. For instance, mild cases could primarily be handled by avoiding triggers like cigarette smoke, air pollution, and food containing sulfites. For severe cases, medications like salbutamol and oral corticosteroids are generally utilized to treat symptoms or prevent further exacerbation (asthma attacks). When lifestyle changes and medications fail, treatment based on oxygen therapy, magnesium sulfate administrated intravenously, and alternative medicines are considered.

Asthma is recognized as a major public health problem affecting approximately 7% of the U.S. population. The disease could involve individuals of any age, but is more common in children than adults. In the U.S., around 7.1 million children live with asthma, which costs about $9.3 billion in healthcare expenses. In addition, asthma control in children is substandard. This results in frequent admission to a hospital following an emergency room visit. From a diagnostic viewpoint, though a number of predictive models are available, there are no precise models that can anticipate asthma deterioration before occurrence. This new study performed at the University of Utah reported development of the first machine learning models able to predict signs of asthma deterioration in children one week prior to an exacerbation.

In this research project, the researchers utilized weekly a method based on a self-monitoring tool for asthma called the Asthma Symptom Tracker. The approach was validated on a total of 210 children recruited during hospitalization. They were assessed for a period of more than two years in a total of 2,912 weekly evaluations. During this period, data were gathered then combined with other criteria like patient attributes and environmental variables such as carbon monoxide, sulfur dioxide, temperature, relative humidity, precipitation, and tree pollen count, among others. The results suggested that the developed model was 71.8% accurate, 73.8% sensitive, and 71.4% specific in predicting occurrence of deterioration of asthma in children one week before incidence.

In summary, these findings highlight the promising usefulness of the developed model in predicting asthma attacks in children one week before the occurrence, as the overall success rate is around 75%. However, the researchers identified key potential parameters that could improve the model. In the future, if the model is integrated into an electronic device, it may allow for a real-time self-monitoring system to record early signs of asthma deterioration. This, in turn, would reduce exacerbations, improve the well-being of patients, and decrease healthcare expenses.