Towards Assistive Diagnoses in m-Health: A Gray-box Neural Model for Cerebral Autoregulation Index

TL;DR

This paper introduces a gray-box neural model to estimate the cerebral autoregulation index (ARI) from blood pressure data, enabling non-invasive, assistive diagnoses in m-Health applications for cerebrovascular health monitoring.

Contribution

It presents a novel gray-box neural approach to approximate ARI using blood pressure and cerebral blood flow data, improving non-invasive cerebral autoregulation assessment.

Findings

The model shows good performance compared to existing phenomenological models.

Preliminary results indicate effective relation modeling between blood pressure and cerebral blood flow.

The approach offers potential for non-invasive, mobile health diagnostics.

Abstract

The cerebral autoregulation system (CAS), is a mechanism which aims to regulate pressure variations occurring in the cerebral circulatory system. At present, there only exist invasive methods and, in turn, they are not used to prevent cerebrovascular accidents. Nowadays, the emergent concept of m-Health allows to use mobile devices to assist the cerebral autoregulation index (ARI). For this, it is necessary to find novel models which allow to approximate the ARI by using the blood pressure value. This work proposes a gray-box neural model to find a relation between the arterial blood pressure (ABP) and the cerebral blood flow velocity (CBFV) in order to obtain the ARI. Preliminary results show good performance by using a phenomenological model in comparison to the Aaslid-Tiecks model.