Classifying Tissue Samples using Topological Data Analysis – Dr. Ross Sweet
Data science is a discipline that draws on a wide range of mathematical, statistical, and computational techniques, with the goal of applying them broadly in science, technology, industry. One class of techniques, topological data analysis, relies on the mathematical field of topology. In topology, we study geometric features that persist despite deformations to the original shape. Since data science problems can involve noisy or otherwise messy spatial data, topology provides a natural connection to data analysis.
In this project, we will investigate if one topological data analysis tool, persistent homology, can be used to classify microscopy images of tissue samples. In particular, we will determine if persistent homology can be an effective tool for classifying cancer tissues. Students will work in the R programming language to implement a classification algorithm while also learning the underlying theoretical math that drives persistent homology.
Using Data Science to Investigate Speech Therapy – Drs. Mark and Carolyn Brodie
This project will explore the use of data science methods to help people practice and improve their language skills. We will create an app that discovers which areas of speech someone is having difficulties with and creates examples to help them get better. Such an app could be useful for patients recovering their speech after traumatic brain injury, for children (or anyone) trying to learn a language, for detecting early signs of the effect of dementia (or other conditions) on speech, and many other situations. If such an app was widely deployed, comparing data across different people might provide information about which brain circuits are responsible for common speech mistakes.
The app will present the user with some text to say, record their response, convert the audio to text, and compare that to the original text. If the user didn't say the text correctly, the app will categorize what type of mistake it was. (We will start with some predefined categories, knowing that these categories may need to be made more sophisticated in the future.) By monitoring the user's mistakes over time, the app will diagnose which areas the user should focus on, and will provide examples similar to the ones that the user is often making mistakes on, in order to facilitate greater improvement.
Representation Counts: Latinx Numbers of Iowa – Dr. Molly Tun
This research experience will introduce students to the field of Ethnomathematics – the study of mathematics and culture. Students will consider the ways that math practices are products of their cultural frameworks and the ways that numbers transform and structure societies. Together, we will collect, interpret, and display numeric data and statistics of social importance regarding the Latinx experience in Iowa, in relation to national data trends. Students will receive training of the US Census Bureau database in order to locate and interpret specific data fields from the American Community Survey; this survey released new information this past semester regarding census information regarding identity factors.
In connection with this database, we would identify variables of interest (such as language, education, salary) and work to collect and create visual representations of the target data. This numeric analysis would be a useful resource for public policy makers and those who work on issues of identity, representation, and underserved populations. Our project will help put students in conversation with prominent community members and advocacy groups. This research would reflect on numeric discrepancies between different social groups in our communities in the hopes of working towards more equity in higher education and increased representation of minoritized groups in various sectors of society.