Career advice for ECE undergrad interested in ML4Health research

I received a email about career advice from an undergrad student majoring in ECE at a Chinese university. With the student’s permission, I have rephrased his questions and will write my answers here.

1. As a phd student in Data Science, what advantages and disadvantages do you see for ECE students pursuing a research career in DS or ML?

• Advantages:
  • If you choose signal processing as your concentration, there is a lot of concepts (e.g., pattern recognition, compression, denoising) related to DS/ML. This provides additional perspectives and tools for solving DS/ML problems.
  • If hardware, circuit and computer system is part of your requirement, you have a good understanding how DS/ML algorithms work on a lower level. This is useful for debugging and designing efficient large scale systems.
  • The engineering skills (coding, project management, debugging) are handy for planning/running DS/ML experiments.
• Disadvantages:
  • Although there is a lot of similar concepts , they require adaptation (e.g., control theory -> system controllability)
  • If you apply to DS/ML PhD programs, publication/connections from ECE conferecnes/journals (e.g., IEEE) might receive less recognition than DS/ML conferences/journals (e.g., neurips)
  • If you’re interested in DS/ML theory, you might need to take additional math/algorithm courses.

2. If I wanted to start working on my research career ahead of time, particularly in the use of AI in healthcare, what course content or skills would you recommend I self-study to better prepare myself?

This depends on what kind of research you want to do. For my current line of work, I would recommend the following:

a. Clinical LLM

• Programming: imperative, functional, parallel programming; computer system
• Software engineering: a mid-scale, collaborative project (e.g., being able to write the Carcassone game from scratch). Know how to use github, unit test, PR etc.
• Math: proofs, linear algebra, discrete math
• Data: database (SQL, handling large dataset)

b. Privacy

• More advanced math courses (e.g., analysis, advanced algorithms)
• Some law courses (I haven’t taken it as an undergrad, but I wish I had :p)
• Ethics/philosophy courses

c. Both

• Scientific writing
• A collaborative capstone course or a project management course

3. Are there any additional advice or recommendations you could provide to ECE students like me who are interested in the application of AI in medicine?

I recommend choosing a graduate program / company (academia is not the only place to do AI for medicine) with connections to healthcare industry (e.g., hospital, pharma companies). This way you would have access to data and the opportunity to talk to the end users about their need :)