Enhancing RespiHeart Monitoring System Through Advanced AI and Machine Learning Techniques

Purpose and goal:

The project has aimed to improve filtering, processing and analysis of the RespiHeart signal, by applying AI support and machine learning. This has been achieved through enhanced measurement precision in real time regarding two main parameters: breathing rate and heart rate. An objective of the project was also to evaluate the possibilities of reading out systolic blood pressure through the RespiHeart signal. Hypothesis testing has been able to verify this possibility.

Expected results and effects:

Time series algorithms and machine learning techniques have been developed and tested in aim of refining the model inputs and improving the current analysis and calculation model, as well as real-time presentation. In addition to improved measurement precision and presentation of real-time data, the project had an overall objective to achieve a deeper understanding of the measurement signal and its development possibilities.

The prerequisites for increasing the accuracy of the RespiHeart system as well as for generating in-depth clinical insights have been strengthened. Approach and implementation:

The project has included three phases during October 2023 to March 2024: (1) Data preparation and collection, (2) Model development and training, (3) Documentation and reporting. The starting point for the project has been data from voluntary subjects, consisting of RespiHeart data and data from reference methods. In ongoing collaboration between RespiHeart and RISE, the data has been structured, filtered, analyzed and tested against developed AI models. Knowledge transfer has been ensured through ongoing coordination meetings and documentation.