Pairing sensors with deep learning to improve heart failure treatment

Posted: November 10, 2020

Congestive heart failure affects nearly six million Americans, with 670,000 diagnosed annually. It remains one of the leading causes of hospital admission, readmission and death in the U.S., and is one of the costliest diseases to treat.

Emre Ertin small
Ertin

Toward efforts to improve treatments, the National Science Foundation recently awarded The Ohio State University a three-year, $750,000 grant to team leader and Electrical and Computer Engineering (ECE) Professor Emre Ertin.

His team also includes co-investigators Ping Zhang, who is an assistant professor in Ohio State’s Departments of Biomedical Informatics (BMI) and Computer Science and Engineering (CSE), as well as John Fisher, a senior research scientist at MIT.

Ertin said Ohio State ECE postdoctoral researchers Siddharth Baskar and Nithin Sugavanam are also "key to the success of the program."

Ertin and colleagues have developed a new approach to combine sensors and deep machine learning to not only assess hospitalization risks for congestive heart failure patients, but also factor in patient data from a multitude of sources, including electronic health records, to provide a more precise medical regimen.

They will aim to design, create and validate an easy-to-use sensor patch, combining four key tools to assess real-time cardiac and lung functions: electrocardiogram, bio-radio frequency, bio-impedance, and seismocardiogram.

“In this project, we will pursue proactive approaches to health care, supported by innovations in  noninvasive multimodal sensor systems and paired with interpretable deep learning models, for assessing the risk of chronic disease progression,” Ertin said.

Ever rising health care costs and the growing population of aging adults with chronic conditions necessitates new predictive, personalized and proactive approaches to cardiovascular health. He said it’s not enough to predict the risk of decompensated heart failure through late symptoms like weight gain and labored breathing.

The new technology, Ertin said, reduces the need and high cost of surgeries typically required for implanted monitors, which can result in extended hospital stays.

He added that the joint sensor models developed in this project will provide insights into cardiovascular health previously only available through implanted sensors and catheterizations in surgery.

Noninvasive measurements from the sensor patch are then paired with data from a patient’s electronic health records and deep learning models to achieve long-term therapy targets.

sensor patch close-up
Sensor patch on mannequin

“The design of the sensor patch will explore new techniques, by integrating signals from a wide range of frequency bands, into a single flexible board operating autonomously under a power budget,” he said.

The award earned by Ohio State is provided via the Chemical, Bioengineering, Environmental and Transport Systems (CBET) division of NSF. It supports innovative research and education in the fields of chemical engineering, biotechnology, bioengineering, and environmental engineering, and in areas that involve clean and sustainable energy.

edited version of article by Ryan Horns, Department of Electrical and Computer Engineering

Categories: ResearchFaculty