Conclusion
Summarizing the results
The overall results
Migraines are a debilitating condition that affects many. As this project showed, there is no clear guideline as to what causes migraines. Many symptoms are common among the group but using these symptoms as a way to classify all cases of migraines is difficult. The linear regressions tab showed there is strong evidence of a link between the age of the first headache and the duration of care in years after a chronic migraine diagnosis. However, this doesn't give a clear picture as to why this happens. Principal component analysis also confirmed that there are common links between symptoms like phonophobia and photophobia. Additionally, it showed relationships between intensity, character, and location. From the k-means analysis, the algorithm showed that there was a relationship between high BMI and age of diagnosis. The higher the BMI later in life, the earlier the individual was diagnosed with chronic migraines.
How could this improve?
Migraine datasets are not very common currently, it was a struggle to find decent data. Largely due to the privacy policy of HIPAA, medical records are hard to obtain or use outside of a medical setting. Realistically, it would be nice if there was a place where de-identified personal health information (PHI) could be stored on this condition across multiple racial diversities and genders. Additionally, it would be beneficial to have patients track their symptoms on onset to have an accurate picture of how symptoms progress through the four stages of a migraine. However, asking someone who is suffering from a migraine with brain fog to remember to write down their symptoms is highly unlikely.
Personal reflection
I enjoyed working on this project! I feel I learned a lot about the key concepts of machine learning. The portion of the project that stuck with me was the exploratory data analysis and clustering. Creating simple but informative ways to display data is a critical skill of a data analyst! I plan to take the information gained into my application development career to provide clean, beautiful, and explanatory visual representations of data. Personally, from this experience, I have gained a newfound appreciation for the amount of work and thought that goes into these algorithms and the knowledge an individual needs to know to draw accurate conclusions. I think I teeter towards visualizations and initial exploratory data analysis, but this stuff is very cool and powerful with what it is able to accomplish.
​
As an individual diagnosed with chronic migraines, this was an important topic to me personally to see if I could figure out if some of my symptoms lined up with the results. This condition is a mystery and has stumped some doctors I have seen. Investigating these issues can help individuals like me find some sense of normalcy in all the chaos going on in our health. I hope that there will be more data collected in the future and that it is given the proper analysis and maybe there's something out there that can figure out this conundrum!