The news is currently filled with images of crowded ER rooms, long lines, and overwhelmed hospital staff. I wanted to find a way to reduce the number of patients coming into hospital to 1) reduce further exposure to COVID-19 and 2) decrease stress for medical staff.
What it does
Using a compiled list of symptoms and risk factors from the CDC and other medical resources, I ran R-sentiment analysis on patient-inputted text to output a COVID-19 likelihood score. This score can then be presented to a physician, who will determine whether or not the patient should come in for treatment. A higher score indicates a more severe case.
Challenges I ran into
Since I am not in the medical field, I was not sure how to analyze the score, as in what score determines a "severe" case versus a "mild" case. However, I was able to break down the score into a ratio of positive versus negative factors, which helps to understand the score breakdown more. An example of this would be if two people had a score of 2: one person has 16 positive factors/8 negative factors whereas another person may have 2 positive factors/1 negative factor.
Accomplishments that I'm proud of
This was my first time using R-sentiment analysis, and I am proud that I was able to use a very effective program to hopefully give back to healthcare during this stressful period. I am also proud to have participated in my very first Hackathon!
What I learned
I learned how to look at a problem, design an idea to find a solution, and execute this idea.
What's next for COVET
With more research coming out about COVID-19, more detailed symptoms and risk factors can be added to the "Medical Terms Dictionary" I have created for this project. In addition, the code can be cleaned up more to be more inclusive for patients (such as not having to hyphenate compound-word symptoms).
r, sentiment-analysis, tidyverse