Alzheimers Prediction Application

The aim of this project is to create a Machine Learning model that can take the MRI data of a patient and generate a prediction regarding the diagnosis of Alzheimer's i.e. whether the patient is Demented or Non-demented. Using the data from the user to predict if the patient has Alzheimer's or not

app.py: Contains the fastapi code for the ML model
.ipynb: Contains the machine learning model for Alzheimer's detection
inputs.txt: Contains single example entry

Inputs: Visit, MR Delay, M/F, Hand, Age, EDUC, SES, MMSE, CDR, eTIV, nWBV, ASF

Output: Demented(0) or Nondemented(1) (Group)

Steps followed for creating this project:

1. Generate a new environment and install necessary libraries
2. Generate Machine Learning model for the patient data and create pickle file
3. Create an API using fastapi and test it on http://127.0.0.1:8000/predict or http://127.0.0.1:8000/docs to use the OpenAPI for interactive API.
4. Create Procfile and Deploy the app on Heroku web server
5. Use this ML model API link to send a POST request with the inputs in the format specified in inputs.txt as shown below.
6. If prediction = 0 then patient is Demented and if prediction = 1, patient is Nondemented

Prolems faced

1. Needed to add the CORS headers to the api to allow access to the html page. This resulted in a 400 bad request error. In order to fix this http://localhost:5000 was added to the whitelist.

---

References

1. https://www.geeksforgeeks.org/deploying-ml-models-as-api-using-fastapi/
2. https://github.com/krishnaik06/FastAPI
3. How To Deploy Machine Learning Models Using FastAPI-Deployment Of ML Models As API’s
4. How To Deploy FastAPI Machine Learning Models In Heroku