Machine learning algorithm pipeline for endothelial damage detection and adverse outcome prediction.

Project Information

MACLEA-ENDO

Grant agreement ID: 101062306

DOI

10.3030/101062306

Project closed

EC signature date 15 July 2022

Start date 1 November 2022

End date 31 October 2024

Funded under

• Marie Skłodowska-Curie Actions (MSCA)

Total cost

No data

EU contribution

€ 165 312,96

Coordinated by

FUNDACIO INSTITUT DE CIENCIES FOTONIQUES
Spain

Periodic Reporting for period 1 - MACLEA-ENDO (Machine learning algorithm pipeline for endothelial damage detection and adverse outcome prediction.)

Reporting period: 2022-11-01 to 2024-10-31

Summary of the context and overall objectives of the project

Non-invasive assessment of the health of microvasculature in order to optimize treatment strategies and predict patient outcomes, remains a major challenge in intensive care settings. One non-invasive method that has been proposed combines microvascular tissue oxygenation (StO2) measurements with a vascular occlusion test (VOT) applied to the patient’s arm. This procedure interrogates the local microvascular reactivity by inflating a pneumatic cuff to locally stop blood flow. While the cuff is inflated, is possible to extract a surrogate biomarker of metabolism, while the cuff deflates it is possible to extract biomarkers related to endothelial function. While this approach provides valuable insights, the resulting parameters are limited, serving primarily as indirect surrogates of tissue metabolism and endothelial function, and depend heavily on performing a VOT. Furthermore, the absence of standardized protocols for parameter extraction has restricted the method's clinical adoption in the critical care.
The MACLEA-ENDO project focused on improving these techniques for assessing microvascular health in the critical care. It has investigated the relationships between various VOT-derived parameters and explored methods to stratify patients according to their microvascular reactivity aiding in the identification of patients who deviated from typical patterns. The project also aimed to refine the prediction of endothelial dysfunction and evaluate oxygen metabolism by utilizing only measurements at rest. This aspect could potentially eliminate the need for procedures like VOT, reducing patient discomfort and further protocol standardization. By doing so, it the project could potentially impact the development of individualized care, reduce reliance on complex protocols for patient monitoring in the intensive care unit (ICU).

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

Non-invasive assessment of the microvascular health remains a major challenge in critical care. A commonly used approach relies solely on tissue oxygenation measurements combined with a VOT on the patient’s arm. However, the lack of standardization in this method has made it difficult to integrate into routine monitoring of microvascular health in the ICU.

For the first time, data from over 300 patients were utilized to characterize the microvascular and endothelial health in the diverse and heterogeneous population admitted to the ICU by means of optical data combined with VOT using a unique optical platform developed in the hosting group.
These data have been utilized to perform a comprehensive characterization of both healthy and general ICU population. The tissue oxygenation, perfusion as well as the local baseline metabolism have been characterized at rest and in the response to VOT. The patient displayed an impaired microvascular reactivity in all the parameters, in particular, to the ones related to endothelial function, with generally lower baseline metabolism and slower reoxygenation and reperfusion during the cuff deflation.
The initial analysis utilized basic statistical methods to explore the relationships between optical data and VOT-derived parameters, correlations with demographic variables and potential confounding factors.
Within this framework, the project focused on standardizing measurement protocols. Notably, measurements at rest were shown to provide valuable parameters, such as the oxygen metabolism, which correlate with VOT-derived metabolic data. Shortened VOTs and machine learning were also explored as a potential method for predicting VOT-derived endothelial parameters.

The project also explored various non-machine learning techniques to identify patterns and outliers in this heterogeneous ICU population, such as principal component analysis (PCA) using optics and VOT-derived parameters. A preliminary attempt was also made to analyse whole time-series data without extracting specific parameters, furthermore, providing quantitative analysis methods for standardization and to identify subjects whose behaviour deviated from the general ICU population.

This work represents a step forward in standardizing patient monitoring in the ICU, utilizing optics for more efficient and non-invasive assessments.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

The MACLEA-ENDO project has advanced the characterization of microvascular health by means of hybrid diffuse optics and the use of various strategies to identify outliers and standardize the monitoring of the patients in the ICU. The project has provided new insights into extract critical parameters from baseline measurements. These results have the potential to significantly impact clinical practices by improving patient care and streamlining diagnostic procedures. To ensure further uptake and success, continued research is needed to refine these strategies, demonstrate their effectiveness in clinical settings, and integrate them into routine care practices. The development of supportive regulatory frameworks and further validation studies will be critical for widespread adoption in ICUs.
The MACLEA-ENDO project was conducted in collaboration with another Horizon 2020 initiative, VASCOVID, and significantly contributed to the advancement and improvement of devices capable of measuring parameters that could reduce or even replace the need for protocols involving VOT or similar manoeuvres. This collaboration has played a vital role in the development of more versatile devices, which are now being multiplied with the plan of being distributed across Europe for further research into these critical aspects.
These efforts are the foundation for continued exploration and refinement of the analysis and characterization of microcirculation in intensive care settings. Building on these advancements, I plan to further investigate and enhance these methods, ensuring their integration into clinical practice and contributing to improved patient outcomes in ICUs across Europe and beyond.

maclea-endo.png