OAI Research Initiative

As part of my research for my Master's in Applied Biostatistics I worked under Dr. Micheal LaValley with data from the BMC Osteoarthritis Initiative. The goal was to analyze the characteristics of 4796 patients with or at risk of osteoarthritis and compare those who had knee replacements to those who did not. This research has been an exercise of my statistical programming and machine learning skills by using unsupervised learning techniques (e.g. random forests and k-means clustering) in R to determine osteoarthritis patients most at risk for knee replacements using a multinomial logistic regression model. Since we began at the beginning of February 2023, we have complete analysis of the clustered data, the random forest clusters, and a model which identifies variables with the most predictive value.

Overall, this research has been the most influential practice in education for growing my skills with R and SAS. In addition to getting experience with biomedical research data set from a real longitudinal study I encountered problems that can arise from real data. In the final weeks I experimented with R Shiny as a side project of personal interest to build a dashboard for the organization of the many charts and models we generated through the course of the 100 hours of research, which I may later add to my portfolio site. I am grateful for the conversations my mentor, Mike LaValley, and I have had regarding how to normalize the data, optimize the model selection and prediction as well as the measure accuracy of our clusters and multinomial model.

If the preview above isn't loading correctly then view the full app here