Artificial intelligence latest application! Improve lung diagnostic accuracy

A new study shows that artificial intelligence helps to better understand the results of lung function tests in the diagnosis of long-term lung disease.

The European Respiratory Association's International Congress demonstrated on September 4, 2016 shows this achievement, which is the first exploration of the potential application of artificial intelligence in improving the diagnostic accuracy of lung diseases.

Current tests require a series of methods, including a lung function test to measure the volume (volume) and velocity (flow) of air during breathing, followed by a plethysmographic test to measure static lung volume and airway resistance, and finally a diffusion test. Measure the volume of oxygen and other gases that shuttle through the alveoli. The analysis of these test results is mostly based on expert opinion and international guidelines, trying to detect images in the results.

In this new study, the researchers collected data from 968 experimenters who were all involved in a complete lung function test for the first time. All participants received the first clinical diagnosis based on pulmonary function tests and all other necessary additional tests (such as CT scan, electrocardiogram, etc.). The final diagnosis also requires medical experts to jointly verify.

The researchers then explored whether the concept of "machine learning" could be used to analyze complete lung function tests. Using machine learning algorithms, you can learn and execute predictive data analysis.

The team developed an algorithm that included conventional lung function parameters and clinical variables such as smoking history, body mass index, and age. Based on clinical and pulmonary function data, this algorithm makes the most probable diagnosis recommendations.

Wim Janssens of the University of Leuven in Belgium is a senior member of the study and said: â€œWe have demonstrated in this new study that artificial intelligence can provide us with more accurate diagnoses. The significance of the development of this algorithm is that we can The standard, objective, non-prejudicial approach mimics the complex reasoning processes used by clinicians for diagnosis."

At present, clinicians must also rely on the number of parameters to analyze the results. With artificial intelligence, the machine can observe a collection of all images at once, which helps to produce a more accurate diagnosis. This has already occurred in other health areas and it automatically interprets the ECG results commonly used in clinical practice as a decision support system.

The first author of the study, Marko Topalovic, from the University of Leuven, Belgium, said: "The advantage of this method is that it can more accurately and automatically explain the results of lung function tests to better detect the disease. This can not only help without experience. The clinician has a lot of benefits for the entire medical field because it saves the time to complete the final diagnosis and reduces the additional tests that the clinician is currently using to confirm the diagnosis."

The next step in the research team's plan is to test the algorithm in different populations and to improve the system's decision-making power and lung function verification by continuously updating clinically diagnosed lung function data.

Via EurekAlert

Lei Feng Network Press: This article by Lei Feng network (search "Lei Feng network" public number concern) exclusive compilation, without permission prohibited reprint!