How can I do this?

Sorry for the “catchy” title - I simply can’t name what I want to do. So here we go:
am particularly interested in rare diseases. I want to establish a kind of “database” with four levels:

  1. name of disease (one)
  2. symptoms (many)
  3. lab findings associated with disease
  4. examination findings associated with disease

Later I wat to be able to select symptoms or lab findings (from the pool of all symptoms/lab findings etc. in the database) and then get a list of possible diseases from that database.

I have no idea if this is possible with DTP. If not I’d be more than happy of someone could point me to a possible solution.

Thanks, Michael

  1. probably would be a text item (go for Markdown or plain text, I’d say). The name of the item is the name of the disease. The text body:? Maybe some description of the disease?

  2. could be tags.

But 3. and 4.? Could you provide us with some sample data? The amount of content of boths findings is unclear to me. As is their format and how they are associated to a disease. Does only one of both exist for a disease? Probably not.

So they more likely might be separate files linked by item links or WikiLinks to the disease file.

However you do it, I am pretty sure a database of this kind is possible in DEVONthink.

I’m going to watch this thread for what I hope will be brilliant ideas coming from the community. If I were going to try and do what you have described with DT, I would start with a document named for the disease, say alcoholic liver disease. I would then use tags for sections 2-4 of your list above; you might want to see the sections headed TAGGING SOURCES in the handbook though - it might be viable to create documents with either keywords or hashtags and have those converted to tags automatically.

I would run into problems doing this, though; I’ll try an illustrate with an example:

  1. Alcoholic liver disease
  2. Jaundice, appetite loss
  3. Elevated transaminases, deranged coagulation tests
  4. Jaundice, enlarged liver

There can be an overlap of signs and symptoms, so use of nested tags might be necessary to differentiate; or it may be simpler to group signs and symptoms as one; you might opt to use nested tags anyhow, so as create a cleaner database structure. But the bigger problem in my implementation would be nomenclature: is it “deranged coagulation tests”, “increased INR”, “decreased Quick”, “increased PT”, “prolonged PT”…? Consistency here is key: both hemorrhagic shock and alcoholic liver disease cause raised INR - if one is tagged “increased INR” and the other “prolonged PT” then your database won’t work.

If you manage to jump those hurdles, then you would be left with a system which allowed you to search for a combination of tags and be presented with every disease combining those pathologies, which would be brilliant.

(and yes, I’m aware none of the diseases presented in this post are rare - but it was easier than trying to think of a disease with symptom overlap with Goldenhar syndrome.)

1 Like

To me, that sounds like the typical use case for a relational database. You have several “tables” (disease, symptom, lab finding, examination finding) related to each other (one disease: many symptoms, many lab findings, many examination findings).

DT is not a relational database. So while what you want to do might be possible with it, in my mind it would not be the right tool (in the sense of “optimal”). OTOH, relational databases are not known to be particularly user friendly. Maybe Framemaker would be useful in this context or Ninox (which exists for macOS and iPadOS and syncs its data between both platforms).


I was torn on whether I should write that; if I understand things™ correctly, then the great advantage of a relational database is data accessibility or data sharing. I ended up wondering whether in effect the structures created wouldn’t be overkill. And then I found that there is definition drift as to what a relational database is. Then I was confused. And then I wasn’t confused. And then I wondered which tool would provide for a sufficiently “basic” relational database model to be useable at the level I understand the OP to want (as opposed, say, to relational access by multinational corps spanning continents). Mhm.

1 Like

To judge that, we’d have to know about what # of diseases, symptoms, lab and examination findings we’re talking here: If it were only two diseases with sufficiently small sets of symptoms, DT might be ok. If we’re talking about 10s of diseases and hundreds of symptoms, I’d certainly go and look elsewhere.

Although I know that a relational database is in general not as easy to manage as DT, its stricter handling of data are an advantage in this context. One could easily enforce strict rules for diseases (so as not to have AIDS and SIDA as two different ones). And it is of course very easy to search for all diseases matching a certain subset of symtoms.

DT is, I think, better for unstructured data.


whilst I’m not the OP, we are almost certainly talking a complex database here; you might get a feeling for the sort of thing we are talking about by taking a quick browse around orphananesthesia for example - I think your conclusion may well be correct.

The OP wants to create a database of rare diseases. So you’re totally missing the point here!

Just kidding.


I knew somebody would do that :stuck_out_tongue:

1 Like

Hahaha! Well played, @suavito :smiley:

Chronically suffering of Ineptiae garrulae nervosae makes me predictable.


I’d say this depends on the audience. If it’s a database used by others, I’d use the specific language accepted and taught. If it’s for solo use, I’d use terms I’d commonly say as those are the ones that would come to mind when searching. However you can also add aliases to tags to cover both and potentially cross that boundary.

Just a thought.

I am overwhelmed by the speed and quality of answers!! Thanks to all.
You are right, there are going to be more than just a dozen of diseases and symptoms, so maybe a relational database would make sense (I have to learn that from scratch though…). You are right pointing to the importance of having a strict catalog of symptoms, lab findings etc. in order to make this work.
What a pity that haven’t included that kind of search!

Even if DTP probably isn’t the right tool to use here, the community certainly has been the right place to ask!
As the evening is approaching I will slowly go get my liver entertained :slight_smile:


The examples I used conform to both those criteria - they are basically all the same thing. And that was just one example of thousands. Difficult…

that only sounds rare :smiley:

1 Like

I’m going to guess lack of funding; it’s a great project though!

A relational database might be a better choice, but you could get some mileage from Devonthink. You want the correlations to just appear. You don’t want to manually maintain any more than you have to.

Here’s what comes to mind. Turn off “exclude groups from tagging” in File->Database properties.

Make groups for Diseases, Symptoms, Lab findings, and Examination findings.

When you add something, consider putting it in a specific group under one of those main topics, even if you add a new group for just one file.

So, let’s say you have a group called IQ of a Bivalve under symptoms. In there you’ve got notes about IQ testing, desire to eat crayons, belief we never went to the Moon, and whatever else goes with mollusk-level cognition.

Under Diseases, you’ve got a group for the terrible outbreak of Dunning-Kruger Upper Left.

If you tag IQ of a Bivalve with Dunning-Kruger Upper Left, the D-K group will sprout a subgroup, IQ of a Bivalve, which is a second instance of your Bivalve IQ symptom group.

As you add or edit notes about bivalves to either appearance of the IQ of a Bivalve group, the changes appear in both instances.

The D-K group shows everything with a connection to D-K. If you look at the symptom group, IQ of a Bivalve, the list of tags shows what it relates to. There, you can use the reveal tag feature to navigate to D-K Upper Left, Capitol Storming Syndrome, Higher Education Via Facebook, or any of the other diseases you’ve tagged as presenting with sub-vertebrate brain wattage.

Hope that helps. Please excuse any slant that may have come through in my comments. I just read a few posts elsewhere from some ardent science-deniers. Ewww.

No offense intended to octopuses. They are smart, and they are, in truth, mollusks.

Thanks! Nevertheless, it sounds more straight forward to use a database, and is certainly more portable for later use ore integration elsewhere.

I liked the slant a lot :rofl:

1 Like

Could Numbers or Excel also do the job? :thinking:

I thought about that, too - but ended up discarding the thought. For a small number of diseases it’d be ok, but otherwise you aren’t going to get the data presented in a useful way, me thinks.