Contribute to suppress the pandemic by further increasing the reliability and automation of PCR testing, and analysis. But after the corona crisis, our software can also be widely used in any labs working around the clock to identify a variety of human or animal pathogens and viruses for medical doctors, veterinarians, livestock producers, and food safety agencies.
What it does
There is a huge room for improvement in the evaluation of the PCR tests. The raw output is a time series that has much more information in it than what we generally use. Instead of just checking whether the fluorescent signal crosses a certain threshold after a certain number of cycles, machine learning will help us much better exploiting the information from this time series.
How I built it
Created a website to collect data for reinforcement learning. Created an ANN to learn from the data and establish a weight system by which our neural network reaches our goal.
Challenges I ran into
Very hard to get the needed variety of input data to assemble our training data set.
Accomplishments that I'm proud of
We are proud of our team. There are trained data scientists, biologists, web developers working in our team. There are university professors, university divisions collaborating with us. We are proud of all of us!
What I learned
Working with people with different expertise is a great experience for all of us. We learn a tremendous amount from each other.
What's next for Machine Learning PCR analysis
Collecting more and more PCR data to train our network is our priority now.