CardioPulmonary Exercise Framework (CPEF)


More than four years ago we expanded our research interests towards information extraction and analysis of Cardio-Pulmonary Exercise Tests (CPET). The goal was to try and approach the analysis results and accuracy of an official medical gas analyzer through pure audio analysis.

Through extensive testing and fine-tuning using over 120 real-life exercise test sessions and in collaboration with field experts, doctors and physicians we aimed for automatic ventilatory threshold extraction.

Using DSP we are extracting the Breath Intensity, which is a very accurate (94%) match to the Ventilatory Equivalent (VE) and the Respiratory Rate. Various descriptors stemmed from that process and were ultimately combined to the final detection framework.

Our custom framework is successfully detecting the ventilatory thresholds for an heterogenous set of patients/sessions. Samples have the following characteristics.

  • Range from an age span of 18-65
  • Include all possible physical conditions (from professional athletes to people with very low physical condition)
  • Time spans of sessions analyzed range from 9 mins to 25 mins of exercise testing
  • Overall VT and RCT detection rate was over 92% (spot-on) and when deviation we achieved very small detection error rates

Our exercise test analysis framework will soon be available for iOS as well as a stand-alone OSX desktop version with full visual analysis, graphs, plotting and details for sports and medical labs and physicians. Feel free to contact us for more details and customized solutions.