Kuestion.at

Inspiration

COVID-19 infected person can show no visible symptoms such as rising body temperature, fever, and coughing. Corona test units are not widely available Some stories circulated in the media, some people died when waiting for the test result Test results also does not guarantee if the person is free from Corona virus.

What it does

Extract some distinctive features from the recorded lungs sound.

How I built it

Code Artificial Intelligence and Machine Learning using Python

Challenges I ran into

Cleaning noise to get the real lung sound is not easy Interpretation of lung sound is subjective, mostly dependent on doctors experience with his or her ear sensitivity

Accomplishments that I'm proud of

With limited sample, the ML model is able to detect lungs sounds which has high probability infected with COVID19. Creating an ML model which also can differentiate healthy lungs and lungs affected with asthma.

What I learned

GDPR is hard

What's next for Kuestion.at

Create a GDPR compliant platform so that more medical professionals to upload more lung sounds Engage with public cloud services to get more compute processing power for machine learning training It has wider application such as detecting asthma, bronchitis, emphysema, and other chronic obstructive pulmonary disease

Built With

• php
• python