Parkinson’s Keylogging Study

             I began working in the lab during the summer after my sophomore year. Up till that point, I’ve never been involved in research, but I signed up for the Accelerated Research Initiative, and I had the option to pick from several labs. I picked the DIY Diagnostics stream because I wanted to avoid wet lab work, and I wanted to get better at data analytics and coding. I was a little overwhelmed by all the separate projects going on simultaneously in the lab over summer, but I was given an option to pick from several existing projects or come up with my own diagnostic. I decided to join the Parkinson’s keylogging study.

             Over summer, I worked with a small group of friends to start a clinical research trial to determine the differences in how those diagnosed with Parkinson’s Disease typed compared to a control group. We would collect the data by visiting an exercise group for Parkinson’s patients and asking them to type on a laptop which has InputLog, a keylogging software. Afterwards, we would analyze the collected data using R and Python to look for patterns.

             I initially thought that I would spend most of my time coding, but I spent most of the time working with others to set up other aspects of the study. This involved figuring out how to get a new phone number to call the patients, completing IRB training, announcing our study to participants (pictured), and preparing the folders we would hand out to participants. I had a great experience seeing all the work involved in getting a research trial up and running, and when we actually collected data for the first trial from our first participant, I felt a huge sense of accomplishment.

             This experience still helped improve my coding skills, as I’d hoped. InputLog provided the action time (time from key up to key up) and pause time (time from key down to key down) for all the letters typed and presented the data in a CSV file. I used Python to create a script to remove unnecessary columns and count the total number of backspaces pressed in a typing session, and another script to measure overall accuracy. It was difficult at first because I wasn’t used to this level of autonomy, and I would sometimes spend hours stuck on the same problem. But over time, I got more used to working with data frames and navigating Stack Overflow for quick answers.

             I initially wanted to join DIY diagnostics because I was interested in computer science, but I wasn’t sure if I wanted to pursue a career in the field. But due to my experiences in the stream, I’m confident that I picked the right career path, and I was able to spend my time helping with research that could one day lead to earlier diagnoses of Parkinson’s Disease.