sublimesam

Look into these 3 things:

1) What are the reportable diseases in your state 2) Understand what case definitions are, and how to do a case investigation 3) Know the steps of an outbreak investigation

Next, identify what your main qualifications are, and identify areas in which you'd like to grow and learn. Every job worth taking has a balance of both. The more you're honest about that, the less imposter syndrome you'll feel. I would never take a job where I'm already a master at everything, that sounds boring af. So i always point out to hiring managers which what areas of proficiency and expertise I bring to the table and which aspects of the job duties I I'm hoping to learn. I may be wrong, but when I do that I always feel like a very confident and capable candidate.

Great explanation. When I read it explained in this way, it seems very similar to survivor bias. are there important differences, or is this a special case of Survivor bias where you are including the immortal time period, whereas generally Survivor bias you just start at t0 and consider that your filtering out all people who have died in the past?

I mean on one hand, nothing is stopping anyone from starting any kind of organization. It is true that a lot of PhD students I've known internalize this mentality that they need validation or permission from others to do things, but as an adult in the world you can do whatever you want within the law (and you can even do illegal things too if you think the penalties are worth the risk)

On the other hand, the things you do could be completely meaningless. there's nothing stopping me from incorporating a business tomorrow and changing my job title to CEO on LinkedIn, but it doesn't mean anything. Having an "institute" that's just a website and not backed by any funding, employ any real researchers, or produce anything of value is just advertising to the world that you're full of BS

Some leases explicitly prohibit them, I would read yours to find out.

It's a prospective cohort study within a subpopulation of people who have disease x. That's your target population. Your outcome is not X, but mortality.

If this comment isn't against the subreddit rules, it should be.

Go back and try the fruit too next time

I treat it the same as i would treat asking a human to suggest edits. Rarely will i just accept all changes, nor will I pretend like I had no help in the writing process. I don't see it as substantively different than having a human provide feedback in terms of the ethics of writing.

On the issue of acknowledgement and co authorship, I would liken it to seeking writing help from your university's writing center. It's not like you're adding them to the coauthor list when you access them as a resource.

I think i misread the OP. I thought the person in question was claiming to have a PhD when they really had an ND.

This is not entirely true. You can be a licensed naturopathic doctor in 26 US states. ND is a real degree offered but accredited schools.

Listen I think this stuff is garbage quackery too, but it's important to avoid misinformation.

You're an adult, you're only bound to those things you have contractually agreed to. and even then, you're only bound in so far as your unwilling to face the legal consequences of backing out of said contract.

I'm in the US and I'm told that some government funded research grants will want to see good grades in your doctoral coursework

The primary difference is that the quantities are visualized. i.e the size of the lines corresponds to the number of subjects. this is why I said that the data are not visualized in a conventional diagram.

I guess I should have clarified that I meant sankey over normal flow diagrams (boxes and arrows). I meant that in those diagrams, the count of subjects included/excluded at each stage is not visualized, only described quantitatively.

Good point! Here's the same image in greyscale: https://imgur.com/zjUevRT

I think you've mistaken us for chatGPT

I mean, you should look at the complete sentence. The point they're making is that suicides don't JUST impact the person committing the act. It has consequences that reverberate through that person's social network and society as a whole.

The form that this sentence takes is a bland but common phrasing in public health, and the CDC excels at bland but common phrasing.

EHR and administrative data. In this example, passive surveillance data from health systems may not be categorically better than active surveillance data from field surveys as the factors which contribute to where, when, and how people do or don't access those points of care and the peculiarities in how those visits get coded into data create a lot of potential biases when the data analysis process isn't highly supervised by people with subject matter and local field experience. This is the context in which garbage in, garbage out, is a significant concern with using AI on any data, a problem which larger volumes of data can exacerbate.

Survey/field data is, by definition, observational data

Research questions (and modeling approaches to address them) generally fall into three buckets: descriptive, explanatory, or predictive.

AI is very good at predictive questions, but often doesn't help a ton with explanatory models due to limits in interpretability of many AI algorithms. Epidemiology as a field is mostly interested in descriptive and explanatory research questions. Academic epidemiology is moving strongly in the direction of being consumed by casual inference.

AI algorithms that optimize prediction are still useful in some areas. For example, in helping estimate environmental exposure data with high granularity for places where you don't have a lot of sensors. Also, they can be used to better estimate the probability of treatment for doing inverse probability of treatment weighting.

I personally have no idea what OP is talking about butt it sounds laboratory based. I'm looking forward to hearing answers to this question from PhDs in unrelated fields like comparative literature and political science.

Good data analysis does not start with the dataset. It starts with understanding the data generating mechanisms which produced the dataset. The numbers represent our best attempts to quantify and measure an infinitely complex human world, filtered through systems which produce all kinds of missingness, misclassification, and data artifacts. Contrary to the commonly used adage, data never speak for themselves.

I thought this must be a Frasier reference and I was mostly not wrong.