Theodoros Zanos

Published by Daniel at September 1, 2022

Theodoros P. Zanos is a Greek computer scientist and engineer born in Drama, Greece, in July 1980. He attended Aristotle University in Thessaloniki, Greece, where he earned a degree in electrical and computer engineering in 2004. He then moved to the United States to pursue a career in Biomedical Engineering. He graduated with a Master of Science degree in 2006 from the University of Southern California in Los Angeles. He continued in the institution to get a Ph.D. degree in Biomedical Engineering. Dr. Zanos was extremely interested in Artificial intelligence, with his Ph.D. thesis focusing on developing machine learning and system identification approaches for Multi-Input and Output hippocampal neural circuits to use in a neuroprosthesis platform.(1,2,3)

Extensive Research Background

During his time at the University of Southern California, he was a Biomedical Engineering Society member. He also assumed the role of Research Assistant at the National Science Foundation (NSF), funded the Biomimetic Microelectronic Systems Engineering Resource Center and the NIH/National Institute of Biomedical Imaging and Bioengineering (NIH/NIBIB)-sponsored Biomedical Simulation Resource. Zanos held the position of Research Collaborator in a neuroscience project related to field potentials and neural spike timing at the University of Washington. His main research interests were nonlinear system identification and modeling neural systems, neural prostheses, and applications to neural information processing.(1,2,3)

A Knowledge-avid Mind

Zanos’ curiosity to better understand neural data analysis and its uses in the neuroscience field did not stop after his Ph.D. thesis. After graduating, he longed to understand its application in neural information processing and its interfaces with the nervous system to treat diseases. In 2009, he was recruited as a postdoctoral fellow by Dr. Christopher Pack to work at the Montreal Neurological Institute (MNI), McGill, in Montreal, Canada. His research work focused on combining high-channel-count primate electrophysiology with machine-learning-based neural data analysis methods to relate neural activity to behavior and cognition.

In 2014 he became Research Associate in this institution. After two years, he accepted the Assistant Professor position at the Feinstein Institute for Medical Research and the same position at the Hofstra North-Shore-LIJ School of Medicine at Hofstra University.(1,2,3)

Leveraging an Autonomous Asset

Theodoros Zanos has devoted much of his career to studying neural functioning and neurophysiology, particularly in autonomic nervous system functioning. His previous work on hippocampal circuits and extensive research on local field potentials, cytokine-specific neural pathways, and saccade neurophysiology between 2010 and 2019 gave him insight into developing a method for quantifying autonomic responses. The technique uses non-invasive sensors to record continuous six-lead electrocardiography, respiration rate, blood pressure, pupil diameter, and electrodermal activity monitoring in healthy patients with BMIs < 30 Kg/m2 that underwent a series of tests that measure sympathetic and parasympathetic activity. Non-invasive methods for recording autonomic responses are sought to be applied to early diagnosis, monitoring, and assessment of diseases associated with autonomic dysfunction, such as diabetes mellitus and Parkinson’s disease. Recent publications have revealed transcutaneous auricular vagus nerve stimulation as a short-term adjunctive therapy for fatigue and pain experienced in patients with systemic lupus erythematosus. Favorable responses maintained through the period of study, reported by physician’s assessments and patient’s reports, in a double-blind trial comparing this method with placebo make this therapy a promising candidate for managing these significant concerns in quality of life and holds the promise of future research on the autonomic responses as therapeutic tools.(4,5)

Deep Learning for Deep Sleep

Another exciting area of use for deep learning is to change paradigms with large-volume and evidence-based data analysis. Zanos recently published an article in Nature about a deep-learning predictive model to estimate vital signs’ nocturnal stability in hospitalized patients. This tool’s objective is to avoid unnecessary disruptions in sleep, as this is the most common complaint in hospitalized patients and has been associated with adverse outcomes such as hypertension, cognitive impairment, length of stay, and mortality. The tool requires a small set of regular vital assessments to predict overnight stability for any given patient night, with a low risk of misclassification as stable (0.02%), saving around 50% of overnight monitoring. This model also counts with an adjustable threshold to meet the sternest safety criteria.(6)

The tool requires a small set of regular vital assessments to predict overnight stability for any given patient night, with a low risk of misclassification as stable (0.02%), saving around 50% of overnight monitoring. This model also counts with an adjustable threshold to meet the sternest safety criteria.(6)

Involved, Innovative, Implemented

The pandemic introduced a unique set of challenges, including the threat that this new microorganism has imposed and the overwhelming number of cases that have saturated the healthcare system. During the pandemic, Dr. Zanos’ work has been remarkable, tackling these two aspects by either participating in the report of patients’ characteristics and early outcomes for the disease or using machine learning and its ability to perform practical tools in clinical decision making. Along the same lines, he contributed significantly to the medical community by demonstrating the non-clinical utility of an interpretable mortality predictor developed earlier in the pandemic. As of June 2020, no mortality predictor had been implemented nor generated by U.S. institutions or with U.S. data. Dr. Zanos working for Northwell health, developed and validated a publicly available calculator (NOCOS). This assessment tool uses six parameters readily available in most institutions: serum blood urea nitrogen, age, absolute neutrophil count, red cell distribution width, oxygen saturation, and serum sodium as the most optimal predictors for 7-day-survival in patients hospitalized for COVID-19.(7,8,9)

As of June 2020, no mortality predictor had been implemented nor generated by U.S. institutions or with U.S. data. Dr. Zanos working for Northwell health, developed and validated a publicly available calculator (NOCOS). This assessment tool uses six parameters readily available in most institutions: serum blood urea nitrogen, age, absolute neutrophil count, red cell distribution width, oxygen saturation, and serum sodium as the most optimal predictors for 7-day-survival in patients hospitalized for COVID-19.(7,8,9)

Acknowledged by many

Dr. Zanos continues to prosper in his academic career. He has authored more than 30 peer-reviewed publications in journals like Neuron, PNAS, JAMA, Nature Machine Intelligence, and Journal of Neuroscience. He has received many awards, including the Senior Thesis Award from his alma mater, The Aristotle University of Thessaloniki, in 2004, the Center of Excellence in Commercialization and Research Award in 2010, the Excellence in Research Award in 2018, and the Jean Timmins Award in 2012.(2)

Nowadays, along with his seat as the head of the Neural and Data Science Lab, he continues to work as an Assistant Professor at Feinstein Institute, Hofstra University. Zanos is also an Adjunct Associate Professor at the New York Institute of Technology Department of Electrical Engineering.(2,3)

References