As usual, I spent most of my day working on the ultrasound model. I tried experimenting with some odd new materials, and I've had some intriguing results! One fun thing about developing new technology is that sometimes the most surprising materials can be the most effective. It's very fun to be in a lab like this one where nearly anything scientific I want to do is possible: I can heat things, cool things, vacuum things, inject air, use a wide variety of chemicals, 3D print things, and more.
Another thing I began today was a project in which I work with data! Data science is one of my biggest interests, and I've been studying it independently for over a year now, so I was extremely excited to get a chance to provide helpful services with data. The first step in any data science project is to understand the data. This data is a confidential dataset of every operation that took place in the hospital between mid 2017 and 2019, and it was provided to me by Dr. John Sheehy. The first thing I'm investigating is overlapping surgery. After talking with Dr. Sheehy and Dr. Bohl about overlapping surgery, I read this article to gain a better understanding of the practice and criticisms of it, which I summarize below. Overlapping surgery is when one senior surgeon (known as the attending) supervises two cases at the same time. Much of the mechanical work on a surgery is performed by residents, and the attending is only in the room for the most technical parts. According to every surgeon I've talked to, this practice allows more surgeries to happen while also giving residents real-world training in surgery without the attending standing over them. This both allows more people to access care more cheaply and leads to better trained surgeons in the future, since with overlapping surgeries, and resident's first experience working alone occurs before they become an attending. However, this Boston Globe article alleges that overlapping surgeries decrease the quality of surgery. The article references several individual cases in great detail to stir up emotion but provides little effective data to support its allegation. Unfortunately, this article provoked tremendous public backlash against overlapping surgery, and federal laws were nearly passed that would ban the practice. Almost all doctors were outraged that the practice was nearly banned, because they are in almost unanimous agreement that the positives of overlapping surgery outweigh the risks. My job is to help answer this question - using an actual, massive dataset. My first task is to make a new column for each row (which represents a surgery) that contains the fraction of the surgery that overlaps with another surgery supervised by the same attending. For instance, if an attending is also working on another surgery for half the time they're working on one surgery, the overlap_fraction column's value for that row would be 0.5. Dr. Sheehy says that he currently has a way to figure this out in excel, but it takes a whopping 48 hours for his computer to calculate the solution. This is a situation in which more versatile data analysis tools like python and pandas are superior, because they aren't encumbered with loading the data graphically for all operations, and they offer the user far more functions. This is one thing I'll work on when I get back to school next week, and it will be a great way for me to test my data cleaning skills!
0 Comments
Leave a Reply. |
Jeremy MahoneyI'm a high school student at Maumee Valley Country Day School, and I'm currently doing a neurosurgery-focused independent study at Barrow with Dr. Michael Bohl. ArchivesCategories |