Problem Statement

Nowadays cancer clinical trials are creating tremendous problem especially in India as cancer diagnosed patients are not aware of the trials/treatment happening in their or nearby cities.

Clinical Trials means the treatment programs which is offered sometimes by a hospital or sometimes by government's aid, these offer good medical facilities and they often promote individuals to get their health checked

Introduction about the problem

As cancer is a very serious disease and it in any case should not be taken lightly, we figured out a machine learning model that tells the user the best cancer treatment hospital available as per one's disease, stage, gender etc.

The Solution

We created a machine learning model that asks the following queries from user :

Age
Weight
Height
Location
Blood Group
Disease
Stage
Cancer(0/1)
Gender

Using these parameters we trained a Scikit-Learn based machine learning model which predicts the best hospital or center that is available to a patient, the model's prediction is not always all accurate and it may give false output

Presentation

Google Slides

Technologies used:

Pandas, Numpy, Scikit-Learn, React, Django, Node, MongoDB

Team Members:

Information about the project's directory

Outer files

The data.csv file consists of all the data that the model is trained on, you may open to analyze the model further. designed_model.ipynb is the notebook file which is used for the complete data science part. Note: OneHotEncoding is being used while training the model.

The backend files

You may need to install all the necessary dependencies

Install cors and all the necessary libararies by typing

pip install django djangorestframework django-cors-headers scikit-learn==1.2.2 pandas numpy

NOTE: Remember to check paths wherever imports are there

The folder api consists of the backend files which are created using django framework of python. To run the backend server, go one step into api and type

python manage.py runserver

Inside the subdir api of api, there is a views file in which all python function are written, they gets executed when the particular api is called. To define those particular parameters we use urls file, all the links that are to be hitted and used are defined there. Changing the settings file is generally not recommended if everything is working but if it's not then you may need to make the necessary changes to setup cors.

For Node.js backend:

First install all the dependencies by typing in the node_backend folder:

npm install

To run the server, type

npm start

Our main file app.js consists of all the necessary API endpoints which helps us further to load and display the data. Then there are Schemas files like citiesSchema.js, trialcenterSchema.js and userSchema.js which helps us to define the schema for our data that has to be stored, as we have a MongoDB database

The Frontend files

React has been used for the complete Frontend part, to make the API calls in React Fetch and axios have been used.

First install all the necessary dependencies by going inside the frontend folder and type:

npm install

To start React server, open the main file in terminal and then type

npm start

and it should begin the server. Note: There are 3 servers running for the proper working of our model.

Project's Features

It offers a great help (if not accurate) to cancer diagnosed patients
The website is completely full stack, it means things like auth, cookies and sessions are implemented along with login and register
Due to the presence of two backend languages, it offers us multiple things to do simultaneously, even API contacting API!

Limitations of the project as of now

Due to lack of data, the model overfits a little.
As of now the model only shows output for some cities because of the limitations of dataset.

We took part in following MLH competitions

Best use of MongoDB Atlas
Most creative use of GitHub

aakarsh-2004/cancer_clinical_trials_prediction