Dr. Hau-Tieng Wu, Duke University

Title: Can applied harmonic analysis turn fetus into a radar of maternal stress?

Compared with the snapshot health information, long-term physiological signal provides health information from the other dimension. To extract correct features from long-term physiological signal and convert them into clinically useful information, we encounter new challenges. In this talk, we focus on one challenge toward this goal, called the single-channel blind source separation problem. We often have limited physiological channels with complicated statistical features, like time-varying amplitude, frequency, and non-sinusoidal pattern, and the signal quality is often impaired by non-stationary noise. I will discuss recent progress in dealing with this challenge by combining diffusion geometry and time-frequency analysis. I will then discuss its application to a clinical trial to detect maternal stress based on the fetal electrocardiogram (ECG) analysis from a single channel maternal abdominal ECG signal. Its application to other fields will be mentioned if time permits.