The course is useful but the grading is terrible. In case of testing the code based on test cases, the grader looks into the code and only passes the code if it's written in a way consistent with the hints. This means that vectorized computations are out and one has to implement loops by hand. Furthermore, since its not clear what the grader is complaining about, one ends up wasting a lot of time if one really cares about "completing" the course. Furthermore, despite a lot of complaints the instructors have not fixed this issue.

What I am finding in most Coursera courses: 1) Important topics are just glossed over in often < 3 minutes. That itself make me wonder if the courses are worth the price even.

2) Code assignments seem to take most time figuring out what the provided code does, and how I am expected to fill in a few blanks. And after completion I am far from being able to, for example, build a basic x-ray classifier from beginning to end.

As other reviews have mentioned, though the course introduces important concepts for evaluating models in a medical context (confusion matrix, ROC curves), the concepts and exercises were too basic and surface level. Keep in mind the medical context is solely from the point of view of medical imaging. The autograder was also buggy and I was unable to start new topics in Discourse to ask questions about it.

This is interesting topic and I learnt how these things are done in medicine. However, from technical point of view, there are many issues. Bugs, typos, unexplained terms (dear learner, now please calculate background ratio) make this course messy and leaves the taste of "rushed product of corona crisis".

The course is very shallow. It explains in detail some simple concepts like Sensitivity and Specificity and then immediately touches complex topics like image recognition architectures, without much explanation. The course materials are unclear and the auto-grader is buggy.

The course was awesome and practical. I used the content so much in my work. However, there were some errors in the assignments or the labs which made me a little confused. It would be better to fix them in order to clarify everything in the course.

Instructions in the graded assignment did not have clear instructions. Sometimes, correct implementation was graded 'incorrect' by the grader. Also, videos of the ROC curve was not clear about why it is needed or what does it say about a model.

It is cool to walk through the CheXNet model and implement it. I am motivated to check out the original paper and train the model on my computer later. But since the remaining two courses are too bad, I only give 3 stars for this course.

Hi Sir/Madam,

i took this course then realised it is beyond my understanding. I am a grade 12 student . please help me to cancel this ,so, i can take another course which can benefit me.

i will appreciate your help.

thanking you

The final assignments should included end to end projects like e.g. hypertuning of parameters.

Most of the utility code especially for understanding the entire image from patch is not clear and should be explained separately.

Too much task was given but less explanation. It was really hard to complete all the tasks. It would be better if easiest tasks are given or more explanation with huge explanation.

Expected content that would be new but found content which I was already familiar with. Disappointed a little on that. Course could have some more interesting and new content.

Good instructor but concepts were not taught in-depth. The assignments gave only a superficial understanding of the subject and cannot prepare one for working in the industry.

The course touches on several aspects of ML for medical. However, the content seems too little and narrow. Only a few cases and architectures are explored.

Out dated course, they should have give more information on how to train model locally then just giving the weights and finished with the assignment.

Would like it if the projects were more in-depth. We should understand the end-to-end pipeline: from preprocessing to deploying in production

Some concepts used in the assignments are note explained in the videos e. g. the calculation of AUC.

Its a good introduction course, but it spends alot of time going over a lot general concepts.

The lectures are good, but I experienced many issues on homeworks.

Content is too shallow, could have gone deeper into some topics.

The codes should have be explained ...

Too general for practical usage

not a deep course. very basic

good as an introduction

Fairly shallow.