Document Type

Event

Start Date

16-7-2020 5:00 PM

Description

With the current advancements being made in the Machine Learning field, utilizing Artificial Intelligence and Deep Learning techniques to analyze complex information such as genome information has become possible. By creating and building up a database of such data as well as compatible medicines, we suspect it is possible to prescribe personalized medicines to treat patients with a high percentage of success while also being able to take into account various other health conditions and pre-dispositions. This possibility depends heavily on the ability of the Neural Network or various other Machine Learning structure to correctly interpret the information from a medical staff worker’s observations. We plan to utilize labelled genome data sets in order to test and train our project’s Machine Learning architecture then make improvements based off those results. We will be testing out various techniques for Machine Learning in order to properly process the complex and extremely long observation notes supplied to us in the data sets. Our primary goal is to make predictions on genetic mutations through the usage of text scrubbing. This will be done using Logistic Regression to classify all of the training data. Once all of the data is classified based on commonalities the algorithms detect, we will create Confusion matrices as well as calculate Accuracy and Precision on the testing data. These results will better allow us to make improvements on the current implementation.

Comments

I will be available for a virtual Q&A session on July 16th, 2020 5:00 PM-6:00 PM

Share

COinS
 
Jul 16th, 5:00 PM

Using Artificial Intelligence to Prescribe Medicine- CS 4732

With the current advancements being made in the Machine Learning field, utilizing Artificial Intelligence and Deep Learning techniques to analyze complex information such as genome information has become possible. By creating and building up a database of such data as well as compatible medicines, we suspect it is possible to prescribe personalized medicines to treat patients with a high percentage of success while also being able to take into account various other health conditions and pre-dispositions. This possibility depends heavily on the ability of the Neural Network or various other Machine Learning structure to correctly interpret the information from a medical staff worker’s observations. We plan to utilize labelled genome data sets in order to test and train our project’s Machine Learning architecture then make improvements based off those results. We will be testing out various techniques for Machine Learning in order to properly process the complex and extremely long observation notes supplied to us in the data sets. Our primary goal is to make predictions on genetic mutations through the usage of text scrubbing. This will be done using Logistic Regression to classify all of the training data. Once all of the data is classified based on commonalities the algorithms detect, we will create Confusion matrices as well as calculate Accuracy and Precision on the testing data. These results will better allow us to make improvements on the current implementation.