Ambient Intelligence in Postoperative Critical Care: First Observations of a Novel Monitoring Approach

Background: Clinical decision making in acute care is a complex process, especially in demanding environments such as operating theaters or intensive care units. In principal, three sources serve as the information basis for clinical decision making: (1) the electronic health and medical record, (2) the primary measured vital values at bedside, and (3) the subjective review and judgment of the caring physician. Decision making based on this information sources thus lags reproducibility, precision and does not scale in both temporal and capacitive dimensions. Our aim is to establish a novel paradigm of real-time clinical phenotype-monitoring that addresses the aforementioned shortcomings.

Method: We developed a real-time data collection framework that is composed of structured electronic health and medical record data (diagnoses, procedures, ward transfers), continuous high-frequency vitals (ECG and pressure waveforms, medications, laboratory values, ventilator settings and measurements, etc.) and as a novel component, a multispectral live video stream of the patient's face and shoulders. The video stream is analyzed in real-time and transformed into a 3D-model of the head, which allows location invariant data analysis over time by U-V mapping. A terminal for bedside visualization has been established and serves the clinicians and nurses for monitoring during the study. A software toolkit for prospective and retrospective data querying and analysis has been developed and will be soon published as open-source software to allow widespread research in this emerging field.

We initiated a prospective cohort study for developing and evaluating our novel monitoring and are currently enrolling patients. In approximately 115 patients, study acceptance had been uniformly high. Here, we present the first observations and highlight emerging analytical opportunities and challenges, as well as the up scaling of the project toward post-cardiothoracic neurological complications and clinical phenotypes in neonates.

Conclusion: Ambient intelligence in medical care is an emerging field with substantial potential to define a completely novel type of continuous patient monitoring.

No conflict of interest has been declared by the author(s).

Publication History

Article published online:
03 February 2022