Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Deep Learning for Digital Healthcare Applications
The goal of this project is to develop a passive monitoring system to enable patients and their physicians to follow the development of their condition and to raise an alert before the sickness becomes life-threatening.
Keywords: Machine Learning, Deep Learning, GANs, Health Computing, Coughs, Cough Detection, Signal Processing, Asthma, Chronic Obstructive Pulmonary Disease (COPD), Python, Tensorflow, Pandas
Cough events associated to Asthma, COPD, and COVID-19 can be used as an indicator to assess the gravity of the sickness. The in-house developed deep learning models can objectively detect changes in a patient’s health remotely, which would make it possible to deploy interventions before the situation becomes life-threatening.
Cough events associated to Asthma, COPD, and COVID-19 can be used as an indicator to assess the gravity of the sickness. The in-house developed deep learning models can objectively detect changes in a patient’s health remotely, which would make it possible to deploy interventions before the situation becomes life-threatening.
This project aims for the student(s) to take a leading role in the enhancement and evaluation of our in-house developed cough detection Deep Learning models that have recently been published by the group. To make these models more performant, the candidate(s) might need to revise the architecture of the model, scrap additional audio files from the web and develop state-of-the-art data augmentation techniques involving GANs. Furthermore, the selected student(s) will also gain hands-on experience with handling large amounts of audio data collected on real patients to train a deep learning model to detect coughing sounds. Our group has collected more than one million sounds from patients at the moment of writing, with more than 100’000 of these being coughs. This represents one of the largest coughing sounds databases available worldwide.
This project aims for the student(s) to take a leading role in the enhancement and evaluation of our in-house developed cough detection Deep Learning models that have recently been published by the group. To make these models more performant, the candidate(s) might need to revise the architecture of the model, scrap additional audio files from the web and develop state-of-the-art data augmentation techniques involving GANs. Furthermore, the selected student(s) will also gain hands-on experience with handling large amounts of audio data collected on real patients to train a deep learning model to detect coughing sounds. Our group has collected more than one million sounds from patients at the moment of writing, with more than 100’000 of these being coughs. This represents one of the largest coughing sounds databases available worldwide.
ETH Zürich
David Cleres
WEV G 217
Weinbergstrasse 56/58
8092 Zürich
dcleres@ethz.ch
https://www.c4dhi.org/projects/digital-biomarker-copd/
ETH Zürich David Cleres WEV G 217 Weinbergstrasse 56/58 8092 Zürich