Aggregate data from different collections

Question

. 。 Answers to this question are eligible for a +100 reputation bounty. thatguy wants to to this question. 对这个问题的。

I am currently working on a project that has the following schema using mongoose.

User schema

const userSchema = {
name: string
email: string
medicalVisits: [{type: Schema.ObjectId, ref: "records"}]
createdAt: Date
}

Records schema

const recordSchema = {
medication: [String],
rating: Number
user: [{type: Schema.ObjectId, ref: "user"}]
tests: [{type: Schema.ObjectId, ref: "tests"}]
createdAt: Date
}

Tests schema

testScore: Number
answers: Object
user: [{type: Schema.ObjectId, ref: "user"}] 
createdAt: Date

From the little schema above, I have a setup where a patient can take tests multiple times and their respective tests are saved in the Tests collection . Also, the date is recorded for all tests they take. A doctor can request to see a patient's record, in this case, the patient has only one record document that has their tests records embedded in them. Currently, I am faced with the problem of getting a patient's newest and oldest test score alongside their initial details.

I can do a mongoose populate to get all information regarding a user, eg

await User.findById(userId).populate({
  path: "medicalVisits"
  model: "records"
  populate: {
    path: "tests"
    model: "test"
 }
})

And that operation returns the patient's record and all the tests they have taken since they signed up to date. But when I make such a call to the Database, I just want to retrieve the patient's newest and oldest score. In other words, I want to get the patients, Initial test score , and their most recent test score. I am new to Mongoose aggregation , I tried to use the Mongoose aggregate function, but it returns an empty array, I guess I am missing something.

Currently, this is what my aggregate pipeline looks like.

const user = await Doctor.aggregate([
      { $match: { _id: docId } },
      {
        $lookup: {
          from: "users",
          localField: "patients",
          foreignField: "_id",
          as: "patients",
        },
      },
      { $unwind: "$patients" },
      { $unwind: "$patients.medicalVisits" },
      {
        $lookup: {
          from: "records",
          localField: "patients.user",
          foreignField: "_id",
          as: "patientRecord",
        },
      },
      { $unwind: "$patientRecord" },
      // { $sort: { createdAt: 1 } },
      {
        $group: {
          _id: docId,
          user: { $last: "$patients" },
          record: { $last: "$patientRecord"}
        },
      },
    ]);
    return user[0];

From the above snippet, my intention is: given a doctor Id , they can see a list of their patients and also see their newest and oldest test score.

Expected Output

const output = {
userId: 6e12euido....
name: "John doe"
email: "john@john.com"
rating: 2
initialTestScore: 10
recentTestScore: 30
}

How do I go about this? Or what could be a better alternative? Thank you very much.

Answer 1

tried my best to understand your case, and I think your aggregation pipeline should be like:

const patientsWithNewestRecord = await Doctor.aggregate([
      { $match: { _id: docId } },
      {
        $lookup: {
          from: "users",
          localField: "patients",
          foreignField: "_id",
          as: "patients",
        },
      },
      // one patient, per doc
      { $unwind: "$patients" },
      // one patient with all his/her visit records, per doc
      {
        $lookup: {
          from: "records",
          localField: "patients.medicalVisits",
          foreignField: "_id",
          as: "patientRecords",
        },
      },
      // one patient with one visit record, per doc
      { $unwind: "$patientRecords" },
      // sort by patient first, createdAt second
      { $sort: { 'patientRecords.user': 1, 'patientRecords.createdAt': 1 } },
      {
        $group: {
          _id: { patient: '$patientRecords.user' },
          user: { $last: "$patients" },
          record: { $last: "$patientRecords"}
        },
      },
    ]);

this pipeline return a list of a doctor's patients and also see their newest test record. Oldest test record should be in similar war.

Answer 2

Based on these collections (as I understand them from your question):

// doctor collection:
{ _id: "doc1", patients: ["user1"] }
// user collection:
{
  _id: "user1", name: "John", email: "john@gmail.com",
  medicalVisits: ["record1", "record2"]
}
// record collection:
{ _id: "record1", rating: 2, tests: ["test1", "test2"] }
{ _id: "record2", rating: 4, tests: ["test3"] }
// test collection:
{ _id: "test1", testScore: 12, createdAt: ISODate("2021-12-04") }
{ _id: "test2", testScore: 9, createdAt: ISODate("2021-12-05") }
{ _id: "test3", testScore: 15, createdAt: ISODate("2021-12-24") }

we can apply:

db.doctor.aggregate([

  { $match: { _id: "doc1" } }

  { $lookup: {
    from: "user",
    localField: "patients", foreignField: "_id",
    as: "patients"
  }},

  { $unwind: "$patients" }, { $unwind: "$patients.medicalVisits" },

  { $lookup: {
      from: "record",
      localField: "patients.medicalVisits", foreignField: "_id",
      as: "records"
  }},

  { $unwind: "$records" }, { $unwind: "$records.tests" },

  { $lookup: {
    from: "test",
    localField: "records.tests", foreignField: "_id",
    as: "tests"
  }},

  { $unwind: "$tests" },

  { $sort: { "tests.createdAt": 1 } },

  { $group: {
    _id: "$patients._id",
    name: { $first: "$patients.name" },
    email: { $first: "$patients.email" },
    rating: { $first: "$records.rating" },
    initialTestScore: { $first: "$tests.testScore" },
    recentTestScore: { $last: "$tests.testScore" }
  }},

  { $set: { "userId": "$_id" } }, { $unset: "_id" }
])

in order to extract:

{
  userId: "user1",
  name: "John",
  email: "john@gmail.com",
  rating: 2,
  initialTestScore: 12,
  recentTestScore: 15
}

---

Differences compared to your query:

• I $lookup the test collection as it seems you information from there to get both test dates and test scores.
• I $sort by test date ( createdAt ) before the $group by user such that we'll be able to define the right order for selecting the $first and $last test scores.
• I extract user's information by using a $first on each group on user's field (since all unwind records for a given user have the same user information): for instance email: { $first: "$patients.email" }
• I extract the $first and $last test scores for a user as defined by the $sort order: initialTestScore: { $first: "$tests.testScore" } and recentTestScore: { $last: "$tests.testScore" } .
• I finally $set / $unset to rename the _id field into userId

Answer 3

I would suggest to do the following once you have the userId/patientId: -

1. Get their tests(all) from the DB in a sorted order.
2. Take the first and last element of the array for your initial and final test report based on the sorting order you have applied (ascending or descending)

If you can just retireve the user details and all the tests without any sorting, then you can proceed the following way : -

1. Run a loop through all the tests and sort the tests according to test date.
2. Take the first and last element of the array for your initial and final test report based on the sorting order you have applied (ascending or descending)

You will not be performing the operations on DB end, so there might be a minor speed issue, but the difference would still come out to be in milliseconds unless a user takes a billion tests.

Let me know if this helps, let me know if it doesnt