is it possible for me to publish work that is based on someone else's patented tech?

A patent provides the owner of the patent with the exclusive right to prevent any other party, not authorized by the patent owner, to make, sell, or use the patented technology. I assume from your post that you are not actually making, using, or selling the patented product, device, whatever, but just writing about it. Patents per se are public documents, so you are free to use any written material contained in the patent. You may also improve the patented device but beware that if, for example, you physically embody the improvement, you may be infringing the patent even though you yourself may be entitled to a patent upon the improvement. Here is a very basic situation - let's say that you are the first one to invent a chair - so you claim and patent a chair comprising at least one leg (could be a pedestal-type chair), a seat, and a back. Someone comes along and wants to improve the chair by adding a seat cushion, or arms to the chair, for more comfort. They then receive their patent. However, they cannot make, use, or sell their own chair because they would be infringing your patent.
On the other hand, you become aware of their improved chair and you want to make, use, or sell the improved chair because it is in fact more comfortable, but you cannot make, use, or sell the improved chair because you will infringe that patent. What to do? You both cross-license each other permitting each to manufacture the improved chair without infringing each other's patents.

I fully agree with the basic principles of patent law outlined by Steven, with respect, I suppose, of an issued patent or published patent application (since you accessed freely there). Published patent titles are public documents its a fact, you are perfectly free to refer to their contents under your academic study.

And of course, if you propose improvements to the patented invention, I guess these will be displayed in your study only and that they will not be implemented "physically".

Under these reservations, the question of infringement of the concerned granted patent or published patent application does not arise in this context, whatever the law to which they belong. One does not infringe a (published) patent title only by offering to perfect (improve) the covered invention in a publication and just writing about it. This would be contrary to the very philosophy of patents that is to advance research through disclosure of the patented invention, in consideration of the private monopoly granted by the state.

However, I would especially like to draw your attention to another potential risk specifically incurred by technical and scientific publications. This risk is insidious in that it focuses on the patented invention in itself and falls under another area of ​​intellectual property. You have to know indeed that the content of a patent, notably its description (text) and drawings (figures), is a intellectual work by nature. And like any work, it can be considered protected by Copyright law despite its essentially technical and scientific aspect, as it shows a certain (supposed) originality (in the choice of words, their arrangement, the expression of ideas, artwork of operation of the invention ...). The content of a patent or a patent application is comparable to that of a classical book and any publications. Thus in France (my country) the courts have already had occasion in the past to recall that the text of a patent constituted an original work governed by the law of Copyright, as it could do it also about a manual, a technical notice or a software license agreement for example. And this position is not unique to France in that it is intended to be universally accepted.

Therefore, I recommend you as well to take some precautions in writing your study, in the event that you would have to reproduce "verbatim" whole extracts of the description of the invention (supporting the claims) and / or figures (explaining how its works) : act as if you were referring to the contents of a book, put all the reproduced passages in inverted commas and specify their origin well (patent references, identification of the inventor(s)), so as to avoid claims and possibly lawsuits from the author or co-authors of the patented invention for infringing - not the concerned patent right per se - but their supposed copyright on certain text elements and / or graphics (Figs) contained therein.

Your sincerely,
Simon