Home

Project Title: Machine Learning-Based ECG Data Processing for Heart Equipment Efficiency

https://miro.com/app/board/uXjVMg91aNI=/?share_link_id=525940594520

Introduction:

Cardiovascular diseases (CVD) represent a significant global health challenge, contributing to a substantial portion of worldwide mortality [1]. In response to this pressing issue, our project aims to harness the power of machine learning for enhancing the efficiency of heart equipment. This report outlines the scope, objectives, and the proposed workflow for our project.

Project Scope:

Our project encompasses the following key aspects:

1. Real-Power Efficiency: Our primary objective is to optimize the power efficiency of heart equipment, particularly focusing on ECG monitoring devices.

2. Software Acceleration: The project considers the relevance of software acceleration techniques to enhance the performance of ECG data processing.

3. Filtering vs. Non-Filtering Data: We will investigate the impact of data filtering on our system. While the goal is not to draw definitive conclusions, this analysis will shed light on the computational intensity associated with unfiltered data.

4. NRF 53 Series Hardware: We will utilize NRF 53 series hardware, including the Power Profiler Kit, to measure power consumption and optimize power efficiency.

Proposed Workflow:

The project will follow the following workflow:

1. Data Acquisition: We will simulate data acquisition using an external microcontroller, such as the ESP32, to gather data points from the MIT-BIH database.

2. Signal Processing: Our main microcontroller, the nRF53, will perform signal integrity detection and assess signal quality. Additionally, it will employ compression techniques to efficiently manage data.

3. Data Transmission: Processed data, represented as RR interval averages, will be transmitted over Bluetooth Low Energy (BLE). This approach will reduce current consumption compared to transmitting raw ECG signals to either a desktop application or a mobile application [Li et al., 1970].

4. Cloud Integration: The transmitted data will undergo post-processing and be sent to Azure Cloud IoT Hub, where Azure Functions and Azure Machine Learning will be employed for predicting cardiac arrests, pulse detection, and assessing various other health-related risks.

5. Results Dissemination: The final results will be communicated back to either the desktop application or the mobile application.

Conclusion:

This report serves as an overview of our project's objectives, scope, and proposed workflow. By integrating machine learning, hardware optimization, and efficient data processing, we aim to contribute to the development of more effective and power-efficient heart monitoring equipment. This project holds the potential to make a significant impact on the healthcare industry and improve patient outcomes.

For more detailed information and updates on our project, please refer to the GitHub wiki page, which will serve as a central resource for all team members to stay informed and collaborate effectively.

Developer Setup

• https://blog.golioth.io/program-mcu-from-wsl2-with-usb-support/

References:

• [1] “World Health Statistics 2013,” World Health Organisation.

• [2] C. J. Deepu, X. Zhang, C. H. Heng, and Y. Lian, “An ECG-on-chip with QRS detection & lossless compression for low power wireless sensors,” An ECG-on-chip with QRS detection & Lossless Compression for Low Power Wireless Sensors, https://scholarbank.nus.edu.sg/handle/10635/130537 (accessed Sep. 30, 2023).

• (Deepu et al., 2016)

• [3] C. J. Deepu and Y. Lian, A low complexity lossless compression scheme for wearable ECG sensors, https://ieeexplore.ieee.org/abstract/document/7251912/ (accessed Sep. 29, 2023).

• (Deepu & Lian, 2015)

• [4] C. J. Deepu, X. Y. Xu, X. D. Zou, and Y. Lian, “An ECG-on-chip for wearable cardiac monitoring devices - arxiv.org,” An ECG-on-chip for wearable cardiac monitoring devices, https://arxiv.org/pdf/1409.8020 (accessed Sep. 29, 2023).

• (Deepu et al., 2010)

• [5] D. John and Y. Lian, A joint QRS detection and data compression scheme for wearable sensors, https://ieeexplore.ieee.org/document/6863633 (accessed Sep. 27, 2023).

• (John & Lian, 2015)

• [6] Y. Zou et al., “An energy-efficient design for ECG recording and R-peak detection based ...,” An Energy-Efficient Design for ECG Recording and R-Peak Detection Based on Wavelet Transform, https://ieeexplore-ieee-org.libproxy1.nus.edu.sg/document/6949637/ (accessed Sep. 27, 2023).

• (Zou et al., 2015)

• [7] J. Li et al., “Low power optimisations for IOT wearable sensors based on evaluation of nine QRS detection algorithms,” Low Power Optimisations for IoT Wearable Sensors Based on Evaluation of Nine QRS Detection Algorithms, https://scholar.archive.org/work/iuygairgyfbg5goqukylmav3y4 (accessed Sep. 27, 2023).

• (Li et al., 1970)

• [8] A. JOHN, R. C. Panicker, B. CARDIFF1, Y. LIAN3, and D. JOHN, “Binary classifiers for data integrity detection in wearable IOT edge ...,” Binary Classifiers for Data Integrity Detection in Wearable IoT Edge Devices, https://www.semanticscholar.org/paper/Binary-Classifiers-for-Data-Integrity-Detection-in-John-Panicker/342b4642113ac8476758aa55b4cc04850b372cda (accessed Sep. 26, 2023).

• (JOHN et al., 2020)