Hello! I’m Tim, a consultant (attending) in Neurointensive care and Anaesthesia in the UK. I’m looking to develop my understanding of AI and ML techniques pertinent to my field.
I am a relative newcomer and would be grateful for any tips on where to begin. I have approximately 6-8h per week written into my job plan dedicated specifically to this and am really looking forward to getting my teeth into it!
Hi, Tim.
Welcome to the DeepLearning.AI forums. There are a number of ways to approach this, but it would help to know more about what your goals are and a bit more about your background.
A lot of the courses here are oriented towards developers, meaning engineers who want to learn how to build new AI/ML applications at some level. If your goal is to help develop ML applications in the medical field, then the next question is how much math background you have and whether you already know how to program computers. To take the developer path, you need to know at least high school level linear algebra (matrix multiplication) and mathematical analysis and analytic geometry (what are mathematical functions and how to use them) and you need to be reasonably proficient in python programming. If you don’t already have that background but want to take the developer path, there are courses here that can provide a lot of that background. When I was an undergrad here in the US, I had lots of friends who were “premeds”, meaning studying to go to med school upon graduation, and they all had to take at least sophomore level organic chemistry, which required pretty rigorous math training. So if it also works that way in the UK, you’re probably fine on the math and python would be the only question.
The other general approach is as a user of ML applications. What is available, how to use them and what are their limitations and wider implications. E.g. what are LLMs, how are they trained, what are there current limitations and what can we expect in the near future. That would be a different path that would not require the math and programming background I mentioned above.
Hi Paul, thanks so much for your reply and the advice you’ve given.
Re my background, my maths will be fine (a little brushing up necessary I’m sure), but programming will need some work. I wrote some websites back in the day using HTML and JavaScript (ancient history!) so have a tiny bit of knowledge there, but nothing like Python.
As to application and usage, the unit I work in has data spanning 10 years of continuous monitoring (down to every 5s I think!) of scores if not hundred of variables on somewhere between 4 and 5 thousand patients. It’s clearly a mountain to climb!
What I want to achieve once I have a solid basis is
A. Insights to be gained from current data we have (kinda data science principles)
B. Generation of deep learning algorithms to analyse patient data in real time. I expect these to be specific outputs to begin with.
C. (Super long term) Generation of iterative AI-clinician interface to assist with decision making for patients on NeuroICU
I’ve got some time and I’m looking at resources (like Coursera and Datacamp etc) so I’m really keen to get going!
Hi, Tim.
Thanks for the additional information about your situation and your goals. It sounds like taking the developer path is the way to go. You will want to learn python in one way or another. There are various ways to go about that either on Coursera or elsewhere. Once you’ve gotten python added to your toolkit, then the recommendation would be to do the Machine Learning Specialization (MLS) here followed by the Deep Learning Specialization (DLS).
MLS is an introduction to Machine Learning that introduces you to a number of different types of ML algorithms.
DLS is a more advanced set of courses that covers all the major types of Deep Neural Networks and gives you lots of examples of the types of problems that can be addressed with different Neural Network architectures.
With that as background, you will have a good picture of what’s possible with ML and can then consider what problems you can solve using the data that you have.
Note that MLS and DLS do not really deal with how to prepare data for use with ML algorithms. That is a separate topic under the general heading of Data Science. They have just introduced a new course from DeepLearning.AI that gives an introduction to Data Science. Definitely worth a look, although I personally have not yet checked what it covers.