← Back to Projects

Continuous Blood Pressure Monitoring using Artificial Neural Network

April 30, 2014

blood pressure monitoringneural networks

Project Details

• Topic: Investigation and Development of a Novel Continuous Blood Pressure (BP) Monitoring System Based on Artificial Neural Network (ANN).
• A final year thesis project in partial fulfilment of requirements for B.E. (Mechanical) at National University of Singapore (NUS).
• The project involved use of knowledge from medical, computer and programming disciplines.

Motivation for Project

• Healthcare has become an important issue in aging countries such as Singapore.
• There is a need to have such solutions where people can monitor their health themselves.
• Continuous vital signs monitoring devices are needed that are economical, portable, simple and reliable.
• Among these vital signs, continuous monitoring of BP is very important, especially for elderly people with chronic cardiac conditions.

Objectives of Project

The main objectives of this project are:

• To develop a Peaks Detection Algorithm (PDA) for Pulse Transit TIme (PTT) calculation.
• To develop an ANN model to obtain beat-to-beat BP values. This involves training and testing the ANN, debugging the model and incorporating the model into a vital signs monitoring device.

Tasks

The above-mentioned objectives were achieved via the following tasks:

• An integrated code consisting of following was generated:
  • User Prompt
  • Peaks Detection Algorithm (PDA)
  • PTT Calculation
  • ANN Model using Levenberg-Marquardt Algorithm (LMA)
• An ECGnPPG Unit (EPU) was developed to determine the Electrocardiogram (ECG) signals and Photoplethysmography (PPG) signals for calculating the PTT.
• A preliminary study was conducted on 10 subjects to validate the objectives.
• MatLab software was used as a tool for this project.

The following images illustrate some of above tasks:

Methodology (A modified methodology was adopted due to unavailability of ccNexfin Finapres device)

ANN Training Methodology

EPU

User Prompt (Input data from subjects was gathered using this prompt in MatLab)

PTT Calculation

ANN Model (Only 1 layer was used and LMA was used to train ANN)