AI in Medicine Roadmap

Learn AI in Healthcare in 5 Steps: Roadmap for Physicians

AI in healthcare has been getting popular in recent years. As a physician and Ph.D. student who moved from software development to the artificial intelligence field, I want to create a clear roadmap for physicians who want to dive deep into this field. I am telling this roadmap to all my friends and colleagues who ask by email. Now, this post will answer all their questions.

Why Doctors Should Learn AI in Healthcare?

Should all doctors learn AI in Healthcare? Obviously, no. Which doctors should learn? Definitely, the physicians who want to solve clinical problems from a multidisciplinary aspect and believe in solving these problems with AI. But, you are already too busy in the hospital, and you are spending too much time following new guidelines and research. Why should you spend extra time on AI?

In my opinion, yes. Because AI has already started to convert clinical practice in many fields. 5-6 years ago, I was offered to do research with my professors when I was in medical school, but I heard from them only sentences like “Human is the best computer in the world!” or “Computers can’t do that!”. Now, they are saying, “I saw an AI software in the congress that I’ve participated in lately; it was amazing!” or “Is there any research we can participate in? Can we work together?”. Sorry, dear professors, karma is a bitch! Briefly, if you don’t want to miss the next medical revolution, yes, you should learn AI in healthcare.

Radiology, Pathology, and Pharmacology are the most popular divisions but are not limited to these. Ophthalmology, Dermatology, and even Psychiatry can do so much work with AI. Ophthalmologists detect diabetic retinopathy with AI, dermatologists classify skin lesions as benign or malign, psychiatrists analyze patients’ disorders, and cardiologists detect arrhythmia from ECGs. I believe that you will find a new application field in AI in healthcare, or you will make better existing applications.

I want to mention some investments from companies. One of the biggest investors is NVIDIA. No, they are not only producing gaming GPUs. NVIDIA dedicated a supercomputer for life sciences called Cambridge-1 in the UK. Also, NVIDIA supports more than 1000 healthcare startups in their Inception program. In addition to NVIDIA, the biggest drug companies are investing in AI to reduce drug development costs and time.

After a brief explanation and summary of the general situation, we can continue with AI in the medicine roadmap.

Start Your AI in Healthcare Roadmap

First of all, you should be aware that you are a physician and should be able to look from the clinical aspect to AI. You should learn technical aspects and implementations but never deep dive into hard technical details. You will understand what I am saying when you gain some level in AI.

I will mention some essential skills and learning resources for AI in healthcare. All of them are almost equally good. So, it is up to you which one is better for you and your budget. Also, you don’t have to do these step by step; you can continue in parallel. If you have any resource suggestions, you can share them in the comment section below.

Learn Software Development Basics with Python

I think learning Python is a great starting point for learning AI in healthcare. Because while you are learning Python, you will understand computer basics and learn how to think like a programmer. Also, Python is easy to learn, can work on almost any computer, and can learn from many resources.

Learning Python book is one of the most comprehensive books for Python. It is a long book, but you don’t have to read all of it once. Also, don’t forget to do practice exercises at the end of every chapter.

Understand Matrices and Derivatives

Generally, machine learning and deep learning applications are based on matrices and derivatives. We don’t use matrices and derivatives directly in daily development practice because we use common machine learning frameworks in our code. The frameworks do that for us effortlessly but understanding the basic idea is essential to understanding how your code does work.

I assume that you passed the math classes with high degrees in high school because you are a doctor now 🙂 If so, this level will be sufficient to continue to the next step. If not, I recommend learning these. If you still say, “I want to learn mathematical explanations in detail!” there is a book for you. But, I would recommend you to read all these roadmaps.

Learn Machine Learning and Deep Learning Frameworks

Machine learning and deep learning frameworks are bricks of AI development. All calculations and implementations are would be extremely hard if they did not exist. As I mentioned before, we don’t do the hard math calculations manually, thanks to these frameworks like TensorFlow and PyTorch.

You can use one of these three books below to learn ML and DL basics and frameworks. I read all of them, and I think these are the best option to start ML and DL journey. They start with ML and DL basics and dive into frameworks with code examples.

Learn Healthcare-Specific AI in Healthcare Applications

I didn’t mention any healthcare-specific book or recommendation up to now. Because you should learn theoretical backgrounds first. Now, we can specialize in the field of AI in healthcare. The first book is Deep Medicine, written by Eric Topol, MD. This book doesn’t contain any technical information or code example, but it will definitely open your mind. You will look at different aspects to AI in healthcare.

The second book is a technical book that contains code examples and implementations. You will learn healthcare-specific methods and data processing methods.

Learn MONAI Framework

TensorFlow and PyTorch are great frameworks in general, but we need some specific solutions and methods for AI in healthcare. MONAI does it for us. It is based on PyTorch and developed for healthcare applications. So, you should learn MONAI, obviously. You can read my post about MONAI to learn more about it.


This is a roadmap for physicians who want to learn AI in healthcare. It covers essential programming skills in Python, machine learning and deep learning basics, and finally, healthcare-specific machine learning and deep learning applications.

If you applied these five steps with patience, you have already completed the hardest part. Most people start with excitement and quit after a few weeks. Because it is not as easy as reading here. Only a small percent of learners complete the roadmap. Next, I would recommend you to read “How to Get Better in AI: 6 Suggestions for Beginners” post.

Leave a Reply

Your email address will not be published.