To use summarised_result format

[catalamarti]

Describe the solution you'd like
It would be nice that the output is formatted to the summarised_result format, happy to help, but for management of the results it is easier if all the results are in the same format. Happy to discuss if the is any difficulty.

[mvankessel-EMC]

Hi @catalamarti as long as the SummarisedResult is not compatible in a database, I cannot switch fully to it, without breaking compatibility with HADES.

What I can do however, is add methods to the TreatmentPatternsResults object that overrides i.e. transformToSummarisedResult(), exportSummarisedResult(), and importSummarisedResult(), from omopgenerics.

You should then be able to use the omopgenerics functions on the class, as if it were a data.frame.

[catalamarti]

Hmmm adding summarised result into database is sth I want to do but failed so far due to the splitting and settings separation. I am happy to try again, but it is tricky.

I would feel more confortable designing a new result model (summarised result 2.0) that would solve this issues. Althought this would be quite a breaking change that I dont know if we are willing to do.

For the moment if we would have a function to trun the TP result into a summarised result object I think it would be ideal. What do you think @mvankessel-EMC ?

[mvankessel-EMC]

@catalamarti I agree. If you'd like to contribute feel free to add any omopgenerics related function in a PR that converts the fields from the TreatmentPatternsResults object, as a public method to the object.

[mvankessel-EMC]

@catalamarti, something like this:

tpr <- TreatmentPatterns::TreatmentPatternsResults$new(
  filePath = system.file(package = "TreatmentPatterns", "DummyOutput/output.zip")
)

tpSummarisedDf <- tpr$treatment_pathways |>
  head() |>
  dplyr::inner_join(tpr$analyses, by = "analysis_id") |>
  dplyr::inner_join(tpr$cdm_source_info, by = "analysis_id") |>
  dplyr::mutate(
    freq = as.character(.data$freq)
  ) |>
  dplyr::rename(
    result_id = "analysis_id",
    additional_level = "target_cohort_id",
    cdm_name = "cdm_source_name",
  ) |>
  tidyr::pivot_longer(
    cols = c("age", "sex", "index_year"),
    names_to = "strata_name",
    values_to = "strata_level"
  ) |>
  tidyr::pivot_longer(
    cols = c("pathway", "freq"),
    names_to = "estimate_name",
    values_to = "estimate_value"
  ) |>
  dplyr::mutate(
    estimate_type = dplyr::case_when(
      .data$estimate_name == "pathway" ~ "character",
      .data$estimate_name == "freq" ~ "numeric"
    ),
    additional_name = "target_cohort_id",
    variable_name = "Treatment Pathway",
    variable_level = dplyr::row_number()
  ) |>
  tidyr::pivot_longer(
    cols = "target_cohort_name",
    names_to = "group_name",
    values_to = "group_level"
  ) |>
  dplyr::select(
    "result_id", "cdm_name", "group_name", "group_level", "strata_name",
    "strata_level", "variable_name", "variable_level", "estimate_name",
    "estimate_type", "estimate_value", "additional_name", "additional_level"
  )

summarisedTreatmentPathways <- omopgenerics::newSummarisedResult(x = tpSummarisedDf)
#> `result_type`, `package_name`, and `package_version` added to settings.
#> ! The following column type were changed:
#> • variable_level: from integer to character
#> • additional_level: from integer to character

visOmopResults::visTable(
  result = summarisedTreatmentPathways,
  groupColumn = c("strata_name", "group_level"),
  header = c("cdm_name", "estimate_name"),
  type = "gt"
)

								Cdm name
								Synthea synthetic health database
Result id	Group name	Strata level	Variable name	Variable level	Estimate type	Additional name	Additional level	Estimate name
								pathway	freq
age; viralsinusitis
1	target_cohort_name	all	Treatment Pathway	1	character	target_cohort_id	8	acetaminophen-amoxicillin+clavulanate	-
				2	numeric	target_cohort_id	8	-	103.00
				7	character	target_cohort_id	8	acetaminophen-penicillinv-amoxicillin+clavulanate	-
				8	numeric	target_cohort_id	8	-	55.00
				13	character	target_cohort_id	8	aspirin-acetaminophen-amoxicillin+clavulanate	-
				14	numeric	target_cohort_id	8	-	52.00
				19	character	target_cohort_id	8	acetaminophen-penicillinv	-
				20	numeric	target_cohort_id	8	-	49.00
				25	character	target_cohort_id	8	acetaminophen-amoxicillin+clavulanate-doxylamine	-
				26	numeric	target_cohort_id	8	-	42.00
				31	character	target_cohort_id	8	acetaminophen-aspirin-amoxicillin+clavulanate	-
				32	numeric	target_cohort_id	8	-	41.00
sex; viralsinusitis
1	target_cohort_name	all	Treatment Pathway	3	character	target_cohort_id	8	acetaminophen-amoxicillin+clavulanate	-
				4	numeric	target_cohort_id	8	-	103.00
				9	character	target_cohort_id	8	acetaminophen-penicillinv-amoxicillin+clavulanate	-
				10	numeric	target_cohort_id	8	-	55.00
				15	character	target_cohort_id	8	aspirin-acetaminophen-amoxicillin+clavulanate	-
				16	numeric	target_cohort_id	8	-	52.00
				21	character	target_cohort_id	8	acetaminophen-penicillinv	-
				22	numeric	target_cohort_id	8	-	49.00
				27	character	target_cohort_id	8	acetaminophen-amoxicillin+clavulanate-doxylamine	-
				28	numeric	target_cohort_id	8	-	42.00
				33	character	target_cohort_id	8	acetaminophen-aspirin-amoxicillin+clavulanate	-
				34	numeric	target_cohort_id	8	-	41.00
index_year; viralsinusitis
1	target_cohort_name	all	Treatment Pathway	5	character	target_cohort_id	8	acetaminophen-amoxicillin+clavulanate	-
				6	numeric	target_cohort_id	8	-	103.00
				11	character	target_cohort_id	8	acetaminophen-penicillinv-amoxicillin+clavulanate	-
				12	numeric	target_cohort_id	8	-	55.00
				17	character	target_cohort_id	8	aspirin-acetaminophen-amoxicillin+clavulanate	-
				18	numeric	target_cohort_id	8	-	52.00
				23	character	target_cohort_id	8	acetaminophen-penicillinv	-
				24	numeric	target_cohort_id	8	-	49.00
				29	character	target_cohort_id	8	acetaminophen-amoxicillin+clavulanate-doxylamine	-
				30	numeric	target_cohort_id	8	-	42.00
				35	character	target_cohort_id	8	acetaminophen-aspirin-amoxicillin+clavulanate	-
				36	numeric	target_cohort_id	8	-	41.00

^{Created on 2025-11-07 with reprex v2.1.1}

I set the variable_level to be the row_number because I ran into an issue where different estimates are not allowed for identical grouping columns. Its kind of a hacky work-around, I'm not sure how you usually deal with this.

[mvankessel-EMC]

Here's the snippet what it would look like as a method on the class. If there is a generic function in omopgenerics (i.e. transformToSummarisedResult() or asSummarisedResult()), you would be able to directly put the object in the omopgenerics function aswell.

tpr <- TreatmentPatterns::TreatmentPatternsResults$new(
  filePath = system.file(package = "TreatmentPatterns", "DummyOutput/output.zip")
)

tpSummarisedDf <- tpr$transformToSummarisedResult()

# If transformToSummarisedResult() would be a generic, we could do:
# tpSummarisedDf <- transformToSummarisedResult(tpr)

summarisedTreatmentPathways <- omopgenerics::newSummarisedResult(x = tpSummarisedDf)
#> `result_type`, `package_name`, and `package_version` added to settings.
#> ! The following column type were changed:
#> • variable_level: from integer to character
#> • additional_level: from integer to character

visOmopResults::visTable(
  result = summarisedTreatmentPathways,
  groupColumn = c("strata_name", "group_level"),
  header = c("cdm_name", "estimate_name"),
  type = "gt"
)

[catalamarti]

Hi @mvankessel-EMC if you had to set the variable_level as row number to avoid erroring it means that you have two rows that the only difference is the row number but that contain different values in estimate_value. Although it makes it work I would not recommend that because probably this means that there is some information missing (e.g. outcome cohort, pathway, ...)
Happy to take a look together if needed

[mvankessel-EMC]

@catalamarti It seemed to be because of the counts. Multiple identical counts (estimate_value) can belong to different pathways. I have now included the pathway itself as a variable_level. Still not sure if that would be the ideal way to do this, as I'm now effectively storing all the pathways twice in the result.

EDIT:
Putting the pathways in the variable_x columns and the counts in estimate_x columns also works:

tpr <- TreatmentPatterns::TreatmentPatternsResults$new(
  filePath = system.file(package = "TreatmentPatterns", "DummyOutput/output.zip")
)

tpSummarisedDf <- tpr$treatment_pathways |>
  dplyr::inner_join(tpr$analyses, by = "analysis_id") |>
  dplyr::inner_join(tpr$cdm_source_info, by = "analysis_id") |>
  dplyr::mutate(
    freq = as.character(.data$freq)
  ) |>
  dplyr::rename(
    result_id = "analysis_id",
    additional_level = "target_cohort_id",
    cdm_name = "cdm_source_name",
  ) |>
  tidyr::pivot_longer(
    cols = c("age", "sex", "index_year"),
    names_to = "strata_name",
    values_to = "strata_level"
  ) |>
  tidyr::pivot_longer(
    cols = c("freq"),
    names_to = "estimate_name",
    values_to = "estimate_value"
  ) |>
  tidyr::pivot_longer(
    cols = c("pathway"),
    names_to = "variable_name",
    values_to = "variable_level"
  ) |>
  dplyr::mutate(
    estimate_type = dplyr::case_when(
      .data$estimate_name == "pathway" ~ "character",
      .data$estimate_name == "freq" ~ "numeric"
    ),
    additional_name = "target_cohort_id"
  ) |>
  tidyr::pivot_longer(
    cols = "target_cohort_name",
    names_to = "group_name",
    values_to = "group_level"
  ) |>
  dplyr::select(
    "result_id", "cdm_name", "group_name", "group_level", "strata_name",
    "strata_level", "variable_name", "variable_level", "estimate_name",
    "estimate_type", "estimate_value", "additional_name", "additional_level"
  )

summarisedTreatmentPathways <- omopgenerics::newSummarisedResult(x = tpSummarisedDf)
#> `result_type`, `package_name`, and `package_version` added to settings.
#> ! The following column type were changed:
#> • additional_level: from integer to character

^{Created on 2025-11-11 with reprex v2.1.1}

[catalamarti]

How and why are the pathways stored twice? @mvankessel-EMC

[mvankessel-EMC]

@catalamarti In this implementation, but it is solved in my comment above:

tpr <- TreatmentPatterns::TreatmentPatternsResults$new(
  filePath = system.file(package = "TreatmentPatterns", "DummyOutput/output.zip")
)

tpSummarisedDf <- tpr$treatment_pathways |>
  dplyr::inner_join(tpr$analyses, by = "analysis_id") |>
  dplyr::inner_join(tpr$cdm_source_info, by = "analysis_id") |>
  dplyr::mutate(
    freq = as.character(.data$freq)
  ) |>
  dplyr::rename(
    result_id = "analysis_id",
    additional_level = "target_cohort_id",
    cdm_name = "cdm_source_name",
  ) |>
  tidyr::pivot_longer(
    cols = c("age", "sex", "index_year"),
    names_to = "strata_name",
    values_to = "strata_level"
  ) |>
  dplyr::mutate(
    `Treatment Pathway` = .data$pathway
  ) |>
  tidyr::pivot_longer(
    cols = c("pathway", "freq"),
    names_to = "estimate_name",
    values_to = "estimate_value"
  ) |>
  tidyr::pivot_longer(
    cols = c("Treatment Pathway"),
    names_to = "variable_name",
    values_to = "variable_level"
  ) |>
  dplyr::mutate(
    estimate_type = dplyr::case_when(
      .data$estimate_name == "pathway" ~ "character",
      .data$estimate_name == "freq" ~ "numeric"
    ),
    additional_name = "target_cohort_id"
  ) |>
  tidyr::pivot_longer(
    cols = "target_cohort_name",
    names_to = "group_name",
    values_to = "group_level"
  ) |>
  dplyr::select(
    "result_id", "cdm_name", "group_name", "group_level", "strata_name",
    "strata_level", "variable_name", "variable_level", "estimate_name",
    "estimate_type", "estimate_value", "additional_name", "additional_level"
  )

summarisedTreatmentPathways <- omopgenerics::newSummarisedResult(x = tpSummarisedDf)
#> `result_type`, `package_name`, and `package_version` added to settings.
#> ! The following column type were changed:
#> • additional_level: from integer to character

^{Created on 2025-11-11 with reprex v2.1.1}

[catalamarti]

Is then Treatment pathways included in settings? @mvankessel-EMC