Assignment 0

Overview

Welcome to Assignment 0, our setup assignment! This assignment will not be graded, but it will be necessary for completion of Assignments 1, 2 and 3. You will be going through the steps to get access to several necessary components.

General

  1. Sign up for our Gradescope. The entry code is 4V2J7N. You will be submitting all homework assignments as well as project updates to Gradescope.
  2. Sign up for our Ed if you have not already! This will be an essential form of communication in the course.
  3. Fill out the brief class check-in describing what you want to learn most in the class.

Data Access Setup

  1. Create a Kaggle account. Kaggle is a website with a large number of open-sourced datasets and challenges published. This will allow you to access data needed in future assignments.
  2. Create a PhysioNetWorks account for access to MIMIC-III clinical data.
    1. Follow the instructions here to complete the required training course. (Be careful to choose the correct training course specified in the instructions). When you finish the training course, download your full completion report (this is not the certificate: we want the form with the list of modules, dates, and scores) in PDF form, which you will submit as part of this assignment.
    2. When filling out the PhysioNetWorks application, fill in your role as student, with organization name as Stanford University, and job as student. You should list Serena Yeung as supervisor, with syyeung@stanford.edu as the email and Assistant Professor as the title. Include the course number and name, BIODS 220: Artificial Intelligence in Healthcare, in the research topic field.
    3. Continuing with the instructions, request access to MIMIC-III by submitting an application for credentialed access with your full completion report.
  3. Upon creating a PhysioNetWorks account, please fill out this form with your PhysioNetWorks information so we can grant you access to MIMIC-III. We will do this within a few days of the assignment due date.

Google Cloud Setup

If you are new to Google Cloud Platform (GCP), please follow these instructions . We will use GCP to run GPUs on VM instances. This setup includes creating a free trial with your personal Gmail account, upgrading to a paid account (in order to use GPUs), claiming a GCP coupon code for BIODS 220 students which will cover all GCP costs needed for the course, requesting an increase in GPU quota and setting up your deep learning VM instance. All assignment code has been tested using the settings recommended in this guide, so follow the instructions carefully.

If you are experienced with Google Cloud Platform (GCP) then still check the instructions to get the link to redeem coupons in step 2, and to copy the recommended VM configuration in step 3.

Jupyter Notebook Setup

Future assignments will be completed in Jupyter notebooks running from GCP. To set this up, follow these instructions. You'll be asked to launch a Jupyter notebook on the VM instance and access it in your browser. We'll also cover how to keep Jupyter sessions alive after exiting the GCP terminal by running them in tmux/screen. Finally you'll run a simple 'hello world' program in tensorflow.

What to hand in for this assignment

Submit a PDF of your CITI completion report (from Part 2 of the Data Access Setup) on Gradescope under Assignment 0. You do not need to hand in anything corresponding to the other parts of the assignment but it is important to do them now so that you will be ready for Assignments 1, 2 and 3. Looking ahead, please note that Assignment 1 will require ~1 hour to download data, so please plan accordingly.