Note: This article is being steadily continued and fitted to my current state of learning.
Overview of my learning materials
Recent developments in AI are likely to lead to a quantum leap in the medical field. Though it will probably not replace doctors, it will definitely change the way we have to see our work.
One of the key areas where AI is being used in healthcare is medical imaging. However, also drug development and various specialties, including intensive care, neurology, and psychiatry will benefit from progress in the field of AI.
The fast advancement of this technology is ubiquitous in today’s media and seems to be disruptive for medicine 10 years from now. Yet, university doesn’t prepare medical students for this kind of future. This is why I took matters into my own hands. Here is a short summary of my learning journey so far.
Interest Awakes (Semester 1)
Self-Learning Material (Semester 2)
- Taking the “AI for medics” online course of Charité Berlin. It provides a good overview and an easy introduction to the core concepts of AI. While the “basic course” touches on some technological explanations, the “clinics course” explains how AI is and will be used in medical care and research.
- Learning Basic Python coding on kaggle.com for free.
- I quickly noticed that I often couldn’t completely follow up with technological explanations as I didn’t know some of the mathematical concepts they were referring to. That’s why I took the course “Mathematics for Machine Learning” which I can highly recommend.
Free Sources of Information I found useful
- 3Blue1Brown for perfect explanation and animation of maths concepts in linear algebra, analysis, or statistics
- KhanAcademy as a free workbook for maths questions
- StatQuest is a YT channel with neat explanations for statistical topics, such as linear regression
- A free crowd-sourced list for medical data here (source: Chrislovejoy)
Project-Based Learning (Semester 3)
- After getting down the maths and coding basics, I started applying for summer internships.
- During semester break, I am working and learning at a gastroenterology lab that uses AI to improve endoscopy procedures.
- Here, I started jumping into the field of image processing via CNNs (Convolutional Neural Networks)
- Thereby, I tried both: Implementing multi-class classification models and regression models
- Also, I learned the basics of developing a data pipeline: from anotating raw video data, to cleaning up the data, chosing the right model architecuture as well as training and evaluating a model with the right metrics
- Getting assigned an own project and having the opportunity to work with experts on the field of medical AI fast-tracked that learning process in a way that I’d never have assumed.
- If you want to hear more about my experiences with the project, here is a report I’ve published on Thieme’s student blog.
Free Sources of Information I found useful
- Fresh-up handbook on “Machine Learning for Scientist” (Web-Based)
- 3h-Headstart Course into Deep Learning on Kaggle.com
- Awesome Tutorials on MachineLearningMastery