Week 7(Late Post): Finishing the model.
Jinay M -
Hello all! Sorry for the late blog post once again, I have massive updates for you guys,
I finally finished making the model, and using a python library that can be found within VSC, I was able to get a current accuracy of the model to be 83.33 percent, meaning it was able to classify 83.33 percent of all the spiral drawings it was given from the test set into two different categories: either a Parkinson’s diagnosis, or a healthy person. I want to try and debug some code this week so that I am able to increase my accuracy to where it is slightly better than some of the currently clinically approved models that use MRI scans of patients. I think that this project can really work in clinics because this method of having Parkinson’s patients draw spirals to assess the progression of their disease works because it is non-invasive, works faster than most standard clinically approved detection techniques, and is completely cost effective. This method costs a couple of cents for the cost of paper!
I have talked to MD First Research Chandler and come up with a plan to try and implement this model when prospective patients come to get their tremors looked at. Hopefully the model is able to tell apart who has Parkinson’s and who has just benign tremors.
This week, I am going to try and work on classifying other neurodegenerative disease types as well because I want my model to tell apart other dementia related diseases as well as Parkinson’s. This journey has been really rewarding and it’s given me a chance to look at how diagnosis and prognosis of diseases that don’t have a fully accurate detection system work.
That’s all for week 7! Stay tuned
Bye bye!
Jinay Mehta
Comments:
All viewpoints are welcome but profane, threatening, disrespectful, or harassing comments will not be tolerated and are subject to moderation up to, and including, full deletion.