Where to Start

I often get questions about where to start with ML/AI. It’s a big and technical field, so I put together a list of some of the group’s favorites for folks to work through. Some of the best resources have been compiled by a16z in their AI Canon, which I highly recommend as a starting place. They highlight intro to expert references, as well as commentary, landmark papers, a market overview, and practical tools for building products. I’ve added and highlighted some of my personal favorites, but it’s hard to go wrong if you work your way through their suggestions. 

Here I’ve focused more on:

• Healthcare AI-specific resources 
• Textbooks (we got so good at learning via textbooks in med school!)
• Low/no code resources recommended by the community not listed in the a16z document
• Low-commitment ways to stay up to date on the latest healthcare AI developments

The usual caveats apply:

• I’m sure there are amazing resources I’m not aware of/didn’t include. People are producing new, amazing content every day
• This list is aimed at physicians interested in getting a better technical understanding of deep learning and AI, not necessarily clinical use
• The field is changing quickly and some aspects of some of the courses may not have the newest advances
• These resources are focused more on understanding the technology; there are emerging fields that focus solely on the implementation/limitations/bias/etc that aren’t covered in detail here (yet!)

If you’re new to the field and it all seems like too much

1. Read the two papers I list below for the quick intro
2. Read 2-3 of the “gentle introduction” links from the a16z document
3. Complete one of the short video courses below
4. Sign up for 2-3 of the medical and/or general newsletters listed below. 
5. Figure out what your specific interests are
6. Look at the resources on a16z or here to dive deeper into those interests

For people who want a quick background:

A short guide for medical professionals in the era of artificial intelligence – written in 2020 so doesn’t cover some of the newest advances, but is an easy read with great graphics to understand the kinds of AI, kinds of learning, and high-level ideas on how to evaluate AI papers.

A Surgeon’s Guide to Machine Learning – published in 2021, this paper covers some of the practical aspects that a physician (not just a surgeon) might encounter when trying to develop a study using machine learning: data manipulation/preparation, choosing a model, and evaluating model performance

Short, free video intro courses:

MIT Introduction to Deep Learning: Videos of a compressed one week intro to ML class at MIT. It’s a really nicely explained, concise overview of ML with some interesting guest speakers toward the end.

Google’s Generative AI Learning Path – 10 courses that can be taken individually or separately, each about 45 min. Lets you “play” in the Google generative AI studio at the end.

Deeplearning.ai courses on Coursera – Free series of courses taught by Andrew Ng, very well-reviewed overall

Longer but excellent free courses:

MIT’s Machine Learning for Healthcare – 25 lectures with associated slides/pdfs about fundamental issues in the field. No videos.

Harvard’s CS197 – AI Research Experiences – More advanced course with a textbook and lecture notes, great resources and fun layout. No videos.

MIT’s Linear Algebra – Deep dive videos on linear algebra (which underpins much of ML) in a fun way (yes, fun!).

Udemy Machine Learning A-Z: Very user-friendly introduction to ML, probably need a small amount of programming experience but not much. Can use Python or R. About $20.

Free e-books for people comfortable with math and/or programming:

The Little Book of Deep Learning – free pdf explaining deep learning concepts. Heavy on the math but concise and has some good visuals to support the equations

Langchain AI ebook – From Pinecone, great free practical coding resource, definitely need a coding background

More general healthtech resources:

Sam Altman’s Startup school – great free videos from the original Stanford lecture series and other resources

Y Combinator Startup Library – Great easy to read resources on business topics we weren’t taught in medical school

a16z list of healthtech companies –  a nice list of healthtech companies and what they do. Worth checking out if you have an interest in a specific area of digital health to see (some of) what’s already out there

Medical AI news

Doctor Penguin – Great medical AI newsletter, also from Eric Topol

Ground Truths Substack by Eric Topol: Covers medical innovation, often including AI/ML topics 

Getting Clinical Substack from ScienceIO– Health tech, policy, and AI

General AI news

TLDR AI-Up to the minute news about AI

AI Explained – YouTube channel about new developments in AI

Super Data Science Podcast: A wide variety of topics including several with a medical focus

Towards Data Science – On Medium (so only a few articles/month if not a member), but has some great explanations of data science concepts

Machine Learning Mastery – Also great intro to ML concepts in blog posts that are very readable and have great visuals

XPC – Great resource for primary care-related health tech news

Communities

ML for MDs – all-physician group focused on impact of ML/AI in healthcare. Amazing group of tech-savvy physicians

Health Tech Nerds – many people in the digital health space – majority non-clinical. $20/month or $200/year

Textbooks

The AI Revolution in Medicine – Not really a textbook, but an overview of the field by some major names including Peter Lee and Isaac Kohane

Intro to Statistical Learning: Basic ML introductory textbook

Deep Learning with Python by Chollet: A great, very readable textbook for those with some Python background

Hands on Machine Learning: More in-depth ML textbook