The Development Machine Learning Techniques for the Detection of Trace Pharmaceutical Compounds

The goal of this project is to train and test a machine learning algorithm to detect pharmaceutical contaminants in water. Surface nanodroplets will be generated for several key compounds. These, will be used as independent extraction medium with varying contaminant concentration. The large number of droplets will enable collection of a high quality training data set. Potential chemicals to be investigated include: antibiotics, pesticides and flame retardants. Utilizing surface nanodroplets allows for a sufficiently large training set, one of the main limitations in applying machine learning to chemistry. The various machine learning techniques and parameters will be attempted in order to obtain a predictive regression model for concentration based on spectroscopy data. By the end of the project, we aim to have an algorithm that can determine concentration of trace contaminants in an unknown sample of pharmaceutical wastewater. If sufficient time is available then we will investigate extending the library of supported chemicals to include other pollutant categories.

Faculty Supervisor:

Xuehua Zhang

Student:

Partner:

Kyungpook National University

Discipline:

Engineering

Sector:

Artificial Intelligence; Global Health; Pharmaceuticals

University:

University of Alberta

Program: