Choosing a Nursing Research TopicHow AI Helps You Narrow to One

How to find a topic — capturing the friction you notice in daily practice and shaping it into a question — is covered in a separate article. If you don't have candidates yet, start there.

This article covers what comes next. You have two or three candidates and can't choose. Or you have one, but you're not sure it's right. What the deciding stage requires isn't inspiration — it's inspection.

Established checklists for judging a research topic look at feasibility, interest, novelty, ethics, and relevance (the framework known as FINER is the classic example).

But for unit-based or hospital nursing research, the yardstick that matters most is simpler: "Can we truly finish this in a year, with the team we have now?" A topic you can carry to the finish line beats an impressive one. Admit that up front, and the choice gets dramatically easier.

For each candidate, write out these four points. 1) Can you recruit the population (how many such patients do you see per year)? 2) Can you finish in time (how many months will data collection take)? 3) Can you measure it (what will you use to evaluate the outcome, and is that instrument available to you)? 4) Is the ethical burden acceptable (how will you obtain consent, and how does the intervention differ from usual care)?

If you feel candidates dropping away, that isn't failure — it's progress. The candidate that survives inspection is the topic your team can actually finish.

This is where AI helps. Show it a candidate and ask: "I want to stress-test the feasibility of this research topic. List the points I should check regarding sample size, timeline, measurement, and ethics." It will surface considerations you had overlooked.

It's also good for narrowing questions like "What changes if I restrict the population from all postoperative patients to postoperative hip-fracture patients?" That said, AI can only answer in generalities. How many eligible patients pass through your ward each year, whether your colleagues will cooperate — the realities that actually decide the question live only in your workplace.

Your other piece of evidence is the state of prior research. Run a light literature search on each candidate topic and see how much similar work exists.

Lots of studies? Then think about whether you can articulate why doing it in your setting still matters. Almost none? Then be suspicious about why (perhaps it's genuinely hard to measure, or the population is hard to recruit). An empty space is sometimes a gold mine — and sometimes the site where previous researchers gave up and withdrew. AI summaries work well for pre-reading here too, but if you cite something, checking the original text remains the rule.

Once the inspection is done, the decision belongs to you, your team, and your research supervisor. Never hand the decision itself to AI.

The reason isn't only ethics. Genuine conviction about your topic is the fuel that carries you through a year of research. A topic chosen by someone else — including an AI — won't sustain you through the weeks when data collection gets hard. Let AI gather the material for the inspection; keep the responsibility and the conviction for yourselves. That division of labor is the healthy one.

Once you've decided, write down why you chose this topic and why you dropped the others. Your research protocol and your ethics review will ask exactly these questions.

With Paperfy, the papers you read while weighing candidates stay organized — PDF and AI summary on one page per paper — and you can add decision notes right onto the summaries: "supports candidate A," "different intervention from ours." With your reasoning stored alongside the literature, the protocol practically starts writing itself.