The Problem

Drug discovery, the process of creating and testing new drugs for human use, is difficult due to the vast number of existing chemical compounds and pathways.

The Antidote

Using online chemical databases such as PubChem, ChEMBL, and PubMed, we generated a dataset of approximately 15,000 chemical compounds, including their properties and molecular structure. Our project uses machine learning models to calculate the probability that a chemical can be used to treat a given disease, based on similarities to the molecular structure and chemical properties of existing drug therapeutics. Combining our machine learning models with the Google Firebase API, we then built a web-based search engine that researchers can use to identify potential chemicals for use in drug therapeutics.

Share this project: