Prediction of Critical Pulmonary Shunts in Infants

Radoslav Ivanov, James Weimer, Allan F. Simpao, Mohamed A. Rehman, Insup Lee

Research output: Contribution to journalArticlepeer-review

6 Scopus citations


As a first step toward the development of closed-loop medical cyber-physical systems, this paper presents a monitor for blood oxygen concentration that predicts critical drops in oxygen levels caused by pulmonary shunts in infants. Although blood oxygen concentration is one of the most closely monitored vital signs in modern operating rooms, it cannot be measured noninvasively and is currently monitored by a time-delayed proxy-the hemoglobin oxygen saturation. To predict sharp drops in blood oxygen concentration, we employ available noninvasive respiratory measurements and build a parameterized physiological model of the circulation of these gases through the cardiopulmonary system. Since the model parameters (e.g., metabolic rate) are unknown and vary greatly across patients, we utilize a parameter-invariant detector designed to provide a constant false alarm rate for different patients regardless of the values of the parameters and robust to missing measurements. Finally, we evaluate the performance of the detector on real patient data collected during surgeries performed at the Children's Hospital of Philadelphia. As evaluated on 61 patients experiencing a drop in blood oxygen concentration, the detector achieves a detection rate of about 85% with a potentially life-saving early warning of 90 s on average. In addition, it achieves a false alarm rate of 0.95 false alarms per hour (about 0.5% of the tests) across 314 patients who did not experience a pulmonary shunt.

Original languageEnglish (US)
Article number7439793
Pages (from-to)1936-1952
Number of pages17
JournalIEEE Transactions on Control Systems Technology
Issue number6
StatePublished - Nov 2016
Externally publishedYes


  • Medical cyber-physical systems (MCPSs)
  • parameter-invariant detectors
  • time-series analysis

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Electrical and Electronic Engineering


Dive into the research topics of 'Prediction of Critical Pulmonary Shunts in Infants'. Together they form a unique fingerprint.

Cite this