Predictive Model for Baseline Serum Estradiol Concentration of Female Laboratory Mice; A Moderate Approach to Artificial Intelligence Usage in Healthcare

Tangirala, Ramya, School of Engineering and Applied Science, University of Virginia
Wylie, Caitlin, EN-Engineering and Society, University of Virginia
Dunn, Jacob, EN-Biomed Engr Dept, University of Virginia
Haase, Megan, EN-Biomed Engr Dept, University of Virginia

Artificial intelligence and machine learning is an emerging technology in healthcare around the world with many positive applications improving efficiency, however its use brings about various concerns. Problems with ethics, privacy, and data accountability are some of the main concerns brought up by its use in different aspects of healthcare. The technical portion of this project shows a machine learning approach to identify baseline estradiol concentration and the STS research portion identifies the use of artificial intelligence in healthcare, its positive and negative applications, and different governments’ views on it. The creation of a machine learning approach to identifying concentration for estradiol can help show a positive use of artificial intelligence and machine learning in healthcare which can provide another example of why artificial intelligence should be used for clinical purposes. The technical portion of the project investigated the baseline estradiol concentration in female laboratory mice as there is not much research into its concentration in female mice even though estradiol is a female hormone. The concentrations, weights, and estrous cycle stages of the mice were recorded and were used to build a model to see if we could predict the estradiol concentration from the cycle stage. We found that weight was not correlated to the cycle stage and although the model did work, the concentration values may have bias in it due to previous error in identifying the estrous stage so the model was not able to be used. We also conducted image analysis on the slides used to identify the cycle stage and were able to find the cell type ratios associated with each stage.
For the STS portion of the project, I investigated the use of artificial intelligence in healthcare and looked into various examples of where it was used in positive and negative ways. Further, I looked into various governments and their views on artificial intelligence being used in clinical applications. I found that there were many concerns regarding this technology including potential privacy and data usage concerns. The different governments also do not have a proper artificial intelligence standard that has specific guidelines on how a medical company should create their technology. The SCOT framework was used to delve into this topic and look into the motivations and influences of the different social groups involved including the medical companies, public, government, and hospitals. There could have been more to contribute to the technical problem, however due to the lack of time, we were not able to create an accurate model for our data as the concentration values had errors. Future researchers should look more into the concentration values and the different variables that might affect it. We were able to look into the image classification data and gather ratios associated with each stage of the cycle which can be used to help researchers save time during their work. I was able to contribute to the STS problem through literary analysis and could see that there were benefits to artificial intelligence if it was used in limitation with proper regulations however, as of now there isn’t a strict standard for how these technologies should be made and in what manner they should be used. I would like to thank my advisors Jacob Dunn and Megan Haase of the M3 Lab at UVA and my Capstone professor, Dr. Timothy Allen ,for guiding me on my technical project. I would like to thank Professors Travis Elliott and Caitlyn Wylie for guiding me on my STS project. I would also like to thank my team for the technical project for helping me get the data and putting together our results.

BS (Bachelor of Science)
Predictive Modeling, Estradiol Concentration, Artificial Intelligence, Healthcare

School of Engineering and Applied Science

Bachelor of Science in Biomedical Engineering

Technical Advisor: Jacob Dunn, Megan Haase

STS Advisor: Caitlin Wylie

Technical Team Members: Mikayla Jackson, Khatiana Perez, Samiyah Syeda, Gregory Lawrence

All rights reserved (no additional license for public reuse)
Issued Date: