Public Release: 

Penn Researchers Develop 'Smart' Intensive Care Unit System Using Advanced Computer Intelligence

University of Pennsylvania School of Medicine

New Tools Assist Monitoring Of ICU Patients, And Improve Efficiency

(Philadelphia, PA) -- Researchers at the University of Pennsylvania Medical Center have developed a "smart" intensive care unit (ICU) system that improves vital-sign monitoring of critically-ill patients. By collecting and analyzing several vital signs simultaneously, the smart ICU system could be used to assist physicians and nurses in monitoring patients' physiological parameters -- thus enhancing patient care. This innovative, medical application of artificial computer intelligence will be presented by C. William Hanson, III, MD, associate professor of anesthesia and section chief of anesthesia/critical care medicine at Penn, at the American Society of Anesthesiologists' annual meeting on Tuesday, October 20, in Orlando.

Collecting data such as heart rate, blood pressure, and blood flow measurements is critical to patient care, but it requires a great deal of time and must be analyzed by an experienced clinician. To enhance productivity, advanced computer intelligence can be used to convert a patient's vital-sign measurements into easy-to-follow visual models. "We've designed a system that takes accepted, available information and translates it into a three-dimensional graphic analysis," says Dr. Hanson. "With the smart ICU, we could potentially spot dangerous deviations from a patient's 'ideal' vital-sign range and remedy problems quickly."

The system utilizes two artificial computer intelligence tools: neural networks and fuzzy logic. The neural network works much like the human brain -- it learns how to behave as it interacts with data, and is reinforced for positive performance while being "punished" for poor performance. The smart ICU system's neural network can quickly learn the ideal vital signs for a given patient. Fuzzy logic is a mathematical representation of the way humans think and behave, and is more advanced than traditional computer logic because of its ability to manipulate "fuzzier" concepts such as "almost," "near," and "very far." These two relatively new artificial intelligence tools have successfully been used to enhance performance in a broad range of areas, including air conditioning, elevator, and subway systems. The integration of the two tools allows for the optimal "smart" system with the ability to learn and adapt to varying situations.

One medical application of artificial computer intelligence is hemodynamic analysis: the evaluation of blood pressure, heart rate, and blood flow to the heart. Hemodynamic analysis is typically performed manually by clinicians who compile information from various monitors in order to evaluate each patient's condition. In the study being presented by Dr. Hanson, hemodynamic data was collected non-invasively from 10 patients' electronic bedside instruments to measure cardiac performance. This data consisted of pulmonary artery occlusion, or blockage, pressure (PAOP), heart rate (HR), and cardiac output (CO). The artificial computer intelligence system produced three-dimensional maps, or graphs, on a computer screen which illustrated each patient's hemodynamic status over designated periods of time ranging from one hour to one week. "We've identified a new way to streamline the information analysis process, thereby improving the efficiency of patient care," explains Dr. Hanson. "The smart ICU is designed to support clinicians, not replace them; these tools can perform complicated tasks and help to recognize important trends in a patient's health status."


Editor's Notes: Dr. Hanson can be reached directly at (215) 662-3753 until October 15. From October 16 - 21, please contact Diane Giaccone to reach him at the ASA annual meeting in Orlando.

Dr. Hanson's research appears in the October supplement of Anesthesiology, a peer-review scientific journal.


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.