medNet

Inspiration

Consumer costs for the top 100 drugs has risen from $1,260 to $9,400 in the past 5 years.

What it does

medNet utilizes a Deep Neural Network to find repositioning opportunities for previously FDA approved drugs.

How we built it

Trained a DNN on cell differential expression data collected from the Broad Institute to predict treatment classes for a given cell line.

Challenges we ran into

Processing the data and training the neural network. Finding clear and concise ways to pitch our ideas.

Accomplishments that we're proud of

Trained a Neural Network that predicts treatment classes with high accuracy (rivals others in the field who employed different methodologies such as Aliper, et. al).

What we learned

How to translate a societal problem (the high cost of drugs) into a research question (how gene expression data can be predicative of drug treatment classes). Which then can be formed into a sustainable business model capable of changing the industry.

What's next for medNet

We will continue to employ the cutting edge models in our predictive technology. An example of this would be employing a convolutional neural network.

Built With

• python