An Algorithm that Simplifies the Detection and Diagnosis of Breathing Difficulties

Achievement date: 
2016
Outcome/accomplishment: 

A novel algorithm that measures the harmonic behavior of a patient’s breathing provides accurate, robust, and efficient detection and diagnosis of wheezing in the lungs. The detection method needs to process only 1% of breathing data to render 98% accuracy, using the formula developed by researchers at the Engineering Research Center (ERC) for Advanced Self-Powered Systems of Integrated Sensors and Technologies (ASSIST), an NSF-funded center based at North Carolina State University.

Impact/benefits: 

The algorithm’s efficiency promises quicker and more accurate indication of breathing difficulties in patients, giving doctors better insight into their lung condition. The speed and efficiency of the computations enhances the effectiveness of wearable technologies because of its lower processing requirements and power consumption.

Explanation/Background: 

Earlier efforts to detect wheezing had focused on the timing and frequency of breathing and wavelet transforms. That approach led to measures of many elements in the patterns detected and employed machine learning to assess a patient’s breathing health. The method was complicated by the breathing and many varied characteristics that don’t lend themselves to efficient spectrum analysis. The result was massive data that proved to be computationally demanding to process.

The new approach focuses on the harmonic behavior of wheeze signals, which happens to be the most important characteristic for faster and more accurate analysis and detection. The framework was validated using experimental results obtained on publicly available datasets of breathing sounds recorded using a microphone in a stethoscope. Since the new algorithm method allows for large subsampling factors, its computational cost is remarkably low. The algorithm can assess the onset time, pitch, and magnitude (or volume) of wheezing sounds to provide diagnostic information about the condition of the lungs.

 

The number of tones present in a recorded wheeze signal indicates how many airway occlusions have occurred. In addition, wheezes with higher frequencies are associated with obstruction of the small airways while low-pitched wheezes with lower frequencies are related to diseases of larger airways.