75 points by data_medic 6 months ago flag hide 15 comments
user1 6 months ago next
I've been working with machine learning in healthcare for a few years now. It's a challenging but rewarding field. I have used it for predicting patient readmissions and helping with diagnosis. AMA!
user2 6 months ago next
That's really interesting! Can you talk a bit more about the types of model architectures you've been using? Have they been mostly traditional models or deep learning architectures?
user1 6 months ago next
We've used both traditional and deep learning models, depending on the problem. Traditional models like Random Forests are great for faster iteration, but deep learning models like CNNs and RNNs can yield better results for certain tasks like image/time series analysis.
user3 6 months ago prev next
Have you faced any regulatory challenges in implementing ML in healthcare? I'm considering a move to this field from software engineering and would love to hear your thoughts.
user1 6 months ago next
Yes, there are definitely regulatory challenges. It's crucial to maintain privacy and security. We follow HIPAA and FDA guidelines. Adopting technologies like federated learning can be helpful to meet guidelines while handling sensitive healthcare data.
user4 6 months ago prev next
Can you share some coursework, textbooks, or other resources for getting started in this field? I am interested in pursuing this as a career.
user5 6 months ago next
Coursera's 'Applied Data Science in Python' is a good start. Also, the book 'Deep Learning' by Ian Goodfellow, Yoshua Bengio, and Aaron Courville is a classic text in the deep learning space.
user4 6 months ago next
Thanks! Do you think self-paced online courses are enough to prepare for a career transition or should I consider more formal education paths like Masters degrees?
user5 6 months ago next
I think self-paced courses and hands-on projects give a good starting point. Practical, real-world experience is crucial, especially working on projects and contributing to open-source. A combination of both self-learning and formal education is ideal.
user6 6 months ago prev next
<shameless_plug>Check out our MedTech Fellowship (</shameless_plug>) that combines the knowledge of machine learning and healthcare with mentorship and industry projects.
user6 6 months ago prev next
Stanford's CS230 is a renowned course in Deep Learning. Additionally, the 'Machine Learning Mastery' blog offers valuable content and tutorials on machine learning.
user7 6 months ago prev next
With many healthcare organizations adopting ML, I believe it will become a necessary skill. Which libraries and frameworks would you suggest for beginners?
user1 6 months ago next
scikit-learn, TensorFlow, and Keras are great starting points for beginners. They offer simple APIs and abundant resources.
user8 6 months ago prev next
FastAI and PyTorch are other great tools you can start exploring as you gain more experience. They have gained popularity recently, and the communities are doing a great job at producing educational content.
user9 6 months ago prev next
Definitely learn SQL for data manipulation. You'll most likely work with large datasets, and knowing how to handle them efficiently is essential. Also, don't underestimate the power of regular expressions. They are handy when processing unstructured text data.