Inspiration

Whooping Cough is still a leading cause of death in many third-world countries for children because of a lack of education of the differences between a normal cough and a whooping cough. We attempt to provide a solution for this.

What it does

Cough drop takes a microphone input recording of a baby's cough and runs it though machine learning models to advise the parents whether the child has whooping cough. This signal processing pipeline was designed specifically for open source applications, with the

How I built it

Many models were compared, including a Fourier Transform spectral analysis, a machine learning model with spectral characteristics, and the final machine learning model with the MFCC coefficients. A web app with communication between the model was made with audio recording capabilities.

What's next for Cough Drop

Next steps could include making a python module of just the signal processing pipeline for further applications in medical contexts like asthma, bronchitis, Aalzheimer's Disease, and more. It would also be beneficial to make the model more robust by switching to unsupervised learning that can identify differences in signals just from the raw data without input labels.

Built With

Share this project:

Updates