A new paper from Rose Faghih, Ph.D., Assistant Professor of Electrical and Computer Engineering and the Director of the Computational Medicine Laboratory, and her doctoral student Rafiul Amin describes how they have developed a novel inference engine to obtain brain information from raw electrodermal activity (EDA) recordings, eradicating previous challenges from earlier methods.
IEEE Transactions on Biomedical Engineering published “Identification of Sympathetic Nervous System Activation From Skin Conductance: A Sparse Decomposition Approach With Physiological Priors” in its May 2021 issue.
“We utilize physiological knowledge about the system for reliable identification of the solution,” Amin said. “We address some of the vital challenges for rigorous analysis; by implementing our algorithm on data collected from 109 healthy participants, we establish that our approach can infer brain information with high reliability. Our novel inference engine will eventually help clinicians and researchers with accurate quantification of brain information for health tracking and other clinical/non-clinical applications.”
Amin said that they compared their proposed method with previous approaches, and so far, their method is outperforming the others, especially when it comes to capturing brain activity while suppressing the noise.
“Inference of brain activation related to emotional status helps us track mental health and potentially prevent severe consequences such as suicide,” Amin said. “Regular tracking of brain activity using our approach can also lead to the early detection of diseases like diabetic neuropathy. The transmission of brain activation, including the ones related to EDA, is performed by small nerves to the different regions of the body. EDA can be measured and monitored regularly at neuropathy-prone skin regions of the body to track the received brain activation. If a skin location has neuropathy – if the small nerves are damaged – that region will not receive this brain activation. Our approach can accurately detect the amount of brain activation received in different skin locations for a given stimulus. Thus, our method has the great potential to help to detect diabetic neuropathy.”
Faghih pointed out that there were several other applications for their research, such as tracking pain, cognitive stress tracking and wakefulness, among other things.
“For example, a baby patient experiencing severe pain after a surgical procedure cannot express the level of pain,” she said. “Doctors can eventually utilize EDA recordings and inferred brain activation to evaluate how much pain the baby patient is experiencing to provide necessary intervention.”
Outside of medical applications, Faghih noted that day-to-day tracking of brain activation and arousal levels with their EDA-based approach could enable designing interventions for optimizing productivity.