Exploring novel algorithms for atrial fibrillation detection by driving graduate level education in medical machine learning (2022-06-13T00:00:00.000000Z)

TL;DR

Results from the class ‘Artificial Intelligence in Medicine Challenge’, which was implemented as an online project seminar at Technical University Darmstadt, Germany, and which was heavily inspired by the PhysioNet/CinC Challenge 2017 ‘AF Classification from a Short Single Lead ECG Recording’ are reported.

Abstract

During the lockdown of universities and the COVID-Pandemic most students were restricted to their homes. Novel and instigating teaching methods were required to improve the learning experience and so recent implementations of the annual PhysioNet/Computing in Cardiology (CinC) Challenges posed as a reference. For over 20 years, the challenges have proven repeatedly to be of immense educational value, besides leading to technological advances for specific problems. In this paper, we report results from the class ‘Artificial Intelligence in Medicine Challenge’, which was implemented as an online project seminar at Technical University Darmstadt, Germany, and which was heavily inspired by the PhysioNet/CinC Challenge 2017 ‘AF Classification from a Short Single Lead ECG Recording’. Atrial fibrillation is a common cardiac disease and often remains undetected. Therefore, we selected the two most promising models of the course and give an insight into the Transformer-based DualNet architecture as well as into the CNN-LSTM-based model and finally a detailed analysis for both. In particular, we show the model performance results of our internal scoring process for all submitted models and the near state-of-the-art model performance for the two named models on the official 2017 challenge test set. Several teams were able to achieve F1 scores above/close to 90% on a hidden test-set of Holter recordings. We highlight themes commonly observed among participants, and report the results from the self-assessed student evaluation. Finally, the self-assessment of the students reported a notable increase in machine learning knowledge.

References63 items

Improving Machine Learning Education During the COVID-Pandemic Using Past Computing in Cardiology Challenges

Can challenge-based learning be effective online? A case study using experiential learning theory

A 1D Convolutional Neural Network for Heartbeat Classification from Single Lead ECG

Multiclass Yeast Segmentation in Microstructured Environments with Deep Learning

HOLMES: Health OnLine Model Ensemble Serving for Deep Learning Models in Intensive Care Units

The effect of challenge-based gamification on learning: An experiment in the context of statistics education

The Effect of Data Augmentation on Classification of Atrial Fibrillation in Short Single-Lead ECG Signals Using Deep Neural Networks

PyTorch: An Imperative Style, High-Performance Deep Learning Library

On the Variance of the Adaptive Learning Rate and Beyond

SciPy 1.0: fundamental algorithms for scientific computing in Python

Combining deep neural networks and engineered features for cardiac arrhythmia detection from ECG recordings

ECGNET: Learning Where to Attend for Detection of Atrial Fibrillation with Deep Visual Attention

Ensembling convolutional and long short-term memory networks for electrocardiogram arrhythmia detection

Parallel use of a convolutional neural network and bagged tree ensemble for the classification of Holter ECG

Abductive reasoning as a basis to reproduce expert criteria in ECG atrial fibrillation identification

The Effectiveness of Data Augmentation in Image Classification using Deep Learning

Convolutional recurrent neural networks for electrocardiogram classification

Identifying normal, AF and other abnormal ECG rhythms using a cascaded binary classifier

AF classification from a short single lead ECG recording: The PhysioNet/computing in cardiology challenge 2017

The Dark Side of Gamification: An Overview of Negative Effects of Gamification in Education

Deep Residual Learning for Image Recognition

U-Net: Convolutional Networks for Biomedical Image Segmentation

Detection of Atrial Fibrillation using model-based ECG analysis

Best practices for convolutional neural networks applied to visual document analysis

The impact of the MIT-BIH Arrhythmia Database

andKoepplH 2021OSS-Net:memory efficient high resolution semantic segmentation

andTumanovA 2020HOLMES: health onlinemodel ensemble serving

A andChen LC 2020Axial-DeepLab: Stand-AloneAxial-Attention for Panoptic Segmentation Computer Vision –ECCV2020 (Springer

andAssimakopoulos V 2020The effect of challenge-based gamification on learning: an experiment

Array programming with NumPy Nature

StruckMandMaier A 2020The effect of data augmentation on classification of atrial

FradiM,MachhoutMandHossenA 2020Deep learning-based approach for atrialfibrillation detection International Conference on SmartHomes andHealth Telematics (Berlin: Springer

Françani AO, ReichC andKoepplH 2020Multiclass yeast segmentation inmicrostructured environments

LiH andXie J 2019Combining deep neural networks and engineered features for cardiac

PietquinO andCourville A 2017Modulating early visual processing by languageAdvances in Neural Information Processing Systems pp 6597–607

Attention is all you needAdvances in Neural Information Processing Systems

LiHandXieJ2017ENCASE:AnENsembleClASsi ﬁ ErforECGclassi ﬁ cationusingexpert featuresanddeep

ENCASE: An ENsemble ClASsifiEr for ECG classification using expert features and deep 2017 Computing in Cardiology (CinC)

NarkilahtiSandGabboujM2017Detectionofatrial ﬁ brillationinecghand-helddevices usingarandomforestclassi ﬁ er

A andBengio Y 2016Deep Learning 1 (Cambridge:MITPress) (https://doi.org/10.7551/mitpress

YeungDY,WongWKandWooWC2015Convolutional LSTMNetwork: amachine learning approach

Ayyappa AP andEschenbrenner B 2014Gamification of education: a review of literature International Conference onHCI in Business (Berlin: Springer

2020Data augmentation for electrocardiogram classificationwith deep neural network

Open source ECG analysis Computers in Cardiology

Long short-term memory Neural Comput

MachhoutMandHossenA2020Deeplearning-basedapproachforatrial ﬁ brillationdetection

PietquinOandCourvilleA2017Modulatingearlyvisualprocessingbylanguage

arrhythmiadetectionfromECGrecordings

ReichC,

deeplearning 2020IEEEConf

ZengQ

:anunsupervisedrepresentationlearning

LipGYH etal 2016Atrial ﬁ brillation

YuilleAandChenLC2020Axial-DeepLab

Cetin O¨and Koeppl H 2021 OSS-Net: memory efficient high resolution semantic segmentation of 3D medical data British Machine Vision Conf. British Machine Vision Association

HochreiterSandSchmidhuberJ1997Longshort-termmemory

HamariJ

Advancesin

ﬁ brillationusingmodel-basedECG analysis 200819thInt

NonakaNandSeitaJ2020Dataaugmentationforelectrocardiogramclassi ﬁ

LeonhardtKandHoogAntinkC2021Multi-labelclassi ﬁ cationofcardiacabnormalitiesformulti-leadECG recordingsbasedonauto-encoderfeaturesandaneuralnetworkclassi ﬁ er 2021ComputinginCardiology(CinC)

,RoyADandSahaDK2019Automaticidenti ﬁ cationofarrhythmiafromecgusingalexnetconvolutionalneuralnetwork

VanDerMaatenLandWeinbergerKQ2017Denselyconnectedconvolutionalnetworks