The rising popularity of wearable health devices for personal as well as clinical use, makes them the perfect measurement tool, for a deep understanding of the human condition. Personal wearable devices feel intrinsic and unobtrusive to the user and through longitudinal, passive and ubiquitous sensing, can reveal new insights into human behaviour in an unprecedented manner. Clinical devices can quantify objective and ecological measurements for a better understanding of the biology and pathophysiology of complex conditions.
Synonymous with the device data, is big data. Health devices generate massive amounts of enigmatic big data and require advanced computational strategies and signal processing analysis. Pairing the predictive power of machine learning with advanced signal processing techniques can aid translation of sensor data into medical knowledge discovery.
I've divided my work into two subsections, computational sleep science and neurophysics. The first focuses on personal wearables and how they can be leveraged to improve quality of life via improved sleep quality. The second focuses on clinical EEGs. These machines measure the voltage of neurons firing in the brain. The analysis of these signals can lead to better neurological screening and treatment for conditions that are notoriously complex.
My research develops signal processing methods for personal wearables and clinical devices to between understand the human condition and improve quality of care & quality of life, for all.
In recent years, the public’s increased knowledge of the importance of sleep is fostering the pervasive inclusion and adoption of consumer devices for measuring activity and sleep. This wealth of data inspired my dissertation topic of computational sleep science. The importance of a good night’s sleep is paramount to quality of life, and its insufficience can lead to exacerbation of a variety of health problems, particularly given the clinical bottleneck of seeing a sleep specialist. I developed a human activity recognition system, a deep learning architecture, and a recommendation system to help wearable device users be pro-active about their sleep health.
Deep learning models have achieved state-of-the-art results in a wide variety of tasks in computer vision, natural language processing and speech recognition. The fact that deep learning models automatically learn abstract feature representations from raw features, while also optimizing on the target prediction tasks, makes them an attractive solution for analyzing wearable device data.
The current clinical process for evaluating neurological or neurodevelopment disorders from an electroencephalography (EEG) recording, relies on a clinician’s manual inspection. The electrodynamic signal of the brain is exceptionally complex, and EEG recordings can be minutes, hours, even days long. In contrast, pathologic activity may occur in just a fraction of a second. Our approach views the brain as a dynamical system and computes several nonlinear measures to model the electrodynamics and identify digital biomarkers.