Predicting fluid-response, the heart of hemodynamic management: A model-based solution

Rachel Smith; Christopher G Pretty; Geoffrey M Shaw; Thomas Desaive; J Geoffrey Chase

doi:10.1016/j.compbiomed.2021.104950

Predicting fluid-response, the heart of hemodynamic management: A model-based solution

Comput Biol Med. 2021 Dec:139:104950. doi: 10.1016/j.compbiomed.2021.104950. Epub 2021 Oct 15.

Authors

Rachel Smith¹, Christopher G Pretty², Geoffrey M Shaw³, Thomas Desaive⁴, J Geoffrey Chase²

Affiliations

¹ Department of Mechanical Engineering, University of Canterbury, New Zealand. Electronic address: rachel.smith@pg.canterbury.ac.nz.
² Department of Mechanical Engineering, University of Canterbury, New Zealand.
³ Christchurch Hospital Intensive Care Unit, New Zealand.
⁴ GIGA - In Silico Medicine, University of Liège, Liège, Belgium.

PMID: 34678480
DOI: 10.1016/j.compbiomed.2021.104950

Abstract

Background: Intravenous fluid infusions are an important therapy for patients with circulatory shock. However, it is challenging to predict how patients' cardiac stroke volume (SV) will respond, and thus identify how much fluids should be delivered, if any. Model-predicted SV time-profiles of response to fluid infusions could potentially be used to guide fluid therapy.

Method: A clinically applicable model-based method predicts SV changes in response to fluid-infusions for a pig trial (N = 6). Validation/calibration SV, SV_mea, is from an aortic flow probe. Model parameters are identified in 3 ways: fitting to SV_mea from the entire infusion, SV_fl^fit, from the first 200 ml, SV_fl²⁰⁰, or from the first 100 ml, SV_fl¹⁰⁰. RMSE compares error of model-based SV time-profiles for each parameter identification method, and polar plot analysis assesses trending ability. Receiver-operating characteristic (ROC) analysis evaluates ability of model-predicted SVs, SV_fl²⁰⁰ and SV_fl¹⁰⁰, to distinguish non-responsive and responsive infusions, using area-under the curve (AUC), and balanced accuracy as a measure of performance.

Results: RMSE for SV_fl^Fit, SV_fl²⁰⁰, and SV_fl¹⁰⁰ was 1.8, 3.2, and 6.5 ml, respectively, and polar plot angular limit of agreement from was 11.6, 28.0, and 68.8°, respectively. For predicting responsive and non-responsive interventions SV_fl²⁰⁰, and SV_fl¹⁰⁰ had ROC AUC of 0.64 and 0.69, respectively, and balanced accuracy was 0.75 in both cases.

Conclusions: The model-predicted SV time-profiles matched measured SV trends well for SV_fl^Fit, SV_fl²⁰⁰, but not SV_fl¹⁰⁰. Thus, the model can fit the observed SV dynamics, and can deliver good SV prediction given a sufficient parameter identification period. This trial is limited by small numbers and provides proof-of-method, with further experimental and clinical investigation needed. Potentially, this method could deliver model-predicted SV time-profiles to guide fluid therapy decisions, or as part of a closed-loop fluid control system.

Keywords: Fluid responsiveness; Fluid resuscitation; Hemodynamic monitoring; Intensive care unit; Mathematical modelling.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Animals
Area Under Curve
Fluid Therapy*
Heart
Hemodynamics*
Humans
Stroke Volume
Swine