Control Algorithms and Test Beds for Automated Critical Care System
Abstract
ABSTRACT The transport of critically ill casualties requires the highest level of intensive care, including the administration of medications to resuscitate and stabilize the casualty. Today’s clinical practice is that medication dose is adjusted by human clinician. Yet, optimizing the medication dose is difficult due to a large individual variability in the patient’s response to critical care medicine. So, the standard clinical practice today is to iteratively and empirically adjust medication dose in a given patient, seeking to maximize the desired beneficial effects and minimize the deleterious side effects. To ensure quality monitoring and treatment by eliminating the challenges related to the current manual dose adjustment, this research concerns the automated critical care system (ACCS), a deployable closed-loop control system that can autonomously monitor and sustain the casualties. In theory, a well-designed ACCS may help clinicians make superior adjustments to medication doses, to avoid dangerous delays in making dose adjustments, and also to avoid dose adjustments that are far from being optimal. Current ACCS capabilities present two unique opportunities in developing and validating ACCS: (i) adaptive personalized control of medication therapy for optimal balance between performance and robustness of control systems, and (ii) high-fidelity model-based simulation that reproduces physiologic responses to critical care medications as viable platform to streamline the translation and deployment of emerging ACCS capabilities. Pursuing these opportunities necessitate new dose-response models, which are simple enough to be readily personalized using limited clinical data available in the low-resource settings, and also are physiologically transparent to reproduce complex medication responses. This research proposes “hybrid physiologic-phenomenological” dose-response model as unified basis to develop control algorithms and validation test beds for ACCS. These dose-response models are constructed by integrating physiologic principles and phenomenological observations. Thus, they can exhibit adaptive personalization capability and physical transparency, thereby providing a basis to explore the above opportunities concurrently. The technical objectives of this research are: (i) to develop hybrid dose-response models relevant to critical care medications, (ii) to develop closed-loop control algorithms and simulation-based test beds for ACCS based on the hybrid models, and (iii) to validate the control algorithms based on the simulation-based test beds (both of which are developed in this research). In the Office of Naval Research, ACCS is a technological priority in the Warfighter Protection and Applications Division (342). Therefore, it is highly relevant to enhance the Naval capability. Especially, this research may contribute to the future Naval capability by showing that automated closed-loop control systems can (i) correctly and vigilantly guide medication treatments, (ii) can operate solely based on the measurements available in low-resource settings, and (iii) provide the capability to titrate to very exacting resuscitation goals. Ultimately, the technologies developed in this research may enable a new generation of critical care that potentially improve the patient outcomes as a consequence. The PI is an expert researcher in dynamic systems and control with its applications to real-world challenges in bio-systems and healthcare, having a solid track record on developing closed-loop medication control, decision-support and physiologic monitoring algorithms and systems. Thus, the PI is highly qualified to successfully conduct this research. The Office of Naval Research Young Investigator Award will provide a significant boost to the PI’s multi-disciplinary research program on closed-loop control and decision support systems for combat casualty care.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2016
- Source ID
- N000141512018
Entities
People
- Jin-Oh Hahn
Organizations
- Office of Naval Research
- United States Navy
- University of Maryland