Skip to content

Add example 1 #2

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ymnliu wants to merge 3 commits into
base: master
Choose a base branch
from
Open

Add example 1 #2

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
b52942b
Create evaluate_ohnlp_pipeline_run.sql
ymnliu Feb 25, 2022
c52099d
Update README.md
ymnliu Feb 25, 2022
02f9007
Update evaluate_ohnlp_pipeline_run.sql
ymnliu Feb 25, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,3 +13,6 @@ Installation and Preparation Scripts for setting up OHNLPTK for the RECOVER Task
**For a Local Run:** Run `run_pipeline_local.sh`. You will be prompted for the configuration to use. Alternatively, run `run_pipeline_local.sh your_config_name.json` for non-interactive execution

**For other environments:** Executed `package_modules_and_configs.sh` and edit the `run_pipeline_###` appropriate to your environment and replace declared variables as appropriate.

## Example result analytics
For an example NLP output summary script against PostgreSQL, please refer to `evaluate_ohnlp_pipeline_run.sql`
152 changes: 152 additions & 0 deletions evaluate_ohnlp_pipeline_run.sql
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,152 @@
--Queries written against PostGreSQL; queries should be adaptable to other SQL DIALECTS
--OPTIONAL: indicates that query uses additional data stored in OHDSI
--ADVANCED: indicates uses PostGreSQL specific syntax/functions which would need to be rewritten

--Author: Janos Hajagos (jhajagos)

--Set the search path
set search_path to sbm_covid19_documents, sbm_covid19_hi_cdm_build;

--Get the count of notes
select count(note_id) as n_notes from note;

--OPTIONAL: Get counts for person and visit_occurrences
select count(note_id) as n_notes,
count(distinct person_id) as n_person,
count(distinct visit_occurrence_id) as n_visits
from note;


--Get the count of annotations that have been been made
select count(*) as n_annotations,
count(distinct note_id) as n_notes,
count(distinct lexical_variant) as n_lexical_variants,
count(distinct note_nlp_concept_id) as n_concepts
from note_nlp;


--OPTIONAL: Join to back to notes to get number of visits and persons with annotations
select count(*) as n_annotations,
count(distinct nn.note_id) as n_notes,
count(distinct lexical_variant) as n_lexical_variants,
count(distinct note_nlp_concept_id) as n_concepts,
count(distinct person_id) as n_person,
count(distinct visit_occurrence_id) as n_visits
from note_nlp nn join note n on nn.note_id = n.note_id;


--Get counts for note_nlp_concept_id across
select note_nlp_concept_id,
count(*) as n_annotations,
count(distinct note_id) as n_notes,
count(distinct lexical_variant) as n_lexical_variants
from note_nlp group by note_nlp_concept_id order by count(*) desc;

--OPTIONAL: Join back to notes to get number of visits and number of options
select note_nlp_concept_id,
count(*) as n_annotations,
count(distinct nn.note_id) as n_notes,
count(distinct lexical_variant) as n_lexical_variants,
count(distinct person_id) as n_person,
count(distinct visit_occurrence_id) as n_visits
from note_nlp nn join note n on nn.note_id = n.note_id
group by note_nlp_concept_id order by count(*) desc;

--OPTIONAL Join in concept table to get concept names
select note_nlp_concept_id,
c.concept_name,
count(*) as n_annotations,
count(distinct nn.note_id) as n_notes,
count(distinct lexical_variant) as n_lexical_variants,
count(distinct person_id) as n_person,
count(distinct visit_occurrence_id) as n_visits
from note_nlp nn join note n on nn.note_id = n.note_id
left outer join concept c on c.concept_id = nn.note_nlp_concept_id
group by note_nlp_concept_id, c.concept_name order by count(*) desc;

--Get counts for which term modifiers match
select note_nlp_concept_id, term_modifiers,
count(*) as n_annotations,
count(distinct note_id) as n_notes,
count(distinct lexical_variant) as n_lexical_variants
from note_nlp group by note_nlp_concept_id, term_modifiers order by note_nlp_concept_id, count(*) desc;


--OPTIONAL
select note_nlp_concept_id,
c.concept_name,
term_modifiers,
count(*) as n_annotations,
count(distinct nn.note_id) as n_notes,
count(distinct lexical_variant) as n_lexical_variants,
count(distinct person_id) as n_person,
count(distinct visit_occurrence_id) as n_visits
from note_nlp nn join note n on nn.note_id = n.note_id
left outer join concept c on c.concept_id = nn.note_nlp_concept_id
group by note_nlp_concept_id, term_modifiers, c.concept_name order by note_nlp_concept_id, count(*) desc;

--ADVANCED: Contains PostGreSQL (PSQL) specific functions
--Get positive and negative term counts
--Get top lexical variants
with note_nlp_enhanced as (
select
split_part(split_part(term_modifiers, ',', 1), '=', 2) as certainty, --psql
split_part(split_part(term_modifiers, ',', 2), '=', 2) as experiencr, --psql
split_part(split_part(term_modifiers, ',', 3), '=', 2) as status, --psql
length(snippet) as snippet_length, --psql
length(lexical_variant) as lexical_variant_length, --psql
nn.*
from note_nlp nn)
,
ranked_lexical_variants as
(
select z1.*, rank() over (partition by note_nlp_concept_id order by n_annotations desc) as lexical_variant_rank from (
select lower(lexical_variant) as lower_lexical_variant, note_nlp_concept_id, count(*) as n_annotations
from note_nlp_enhanced
group by lower(lexical_variant), note_nlp_concept_id) z1
)
select
nne1.note_nlp_concept_id,
c.concept_name, c.vocabulary_id,
c.concept_code,
nne1.n_annotations,
pos.n_annotations as pos_n_annotations,
neg.n_annotations as neg_n_annotations,
nne1.n_notes,
pos.n_notes as pos_n_notes,
neg.n_notes as neg_n_notes,
nne1.n_lexical_variants,
nne1.n_lower_lexical_variants,
top_vars.top_five_lower_lexical_variants,
nne1.average_lexical_variant_length,
nne1.average_snippet_length
from (
select
note_nlp_concept_id,
count(*) as n_annotations,
count(distinct note_id) as n_notes,
count(distinct lexical_variant) as n_lexical_variants,
count(distinct lower(lexical_variant)) as n_lower_lexical_variants,
avg(snippet_length) as average_snippet_length,
avg(lexical_variant_length) as average_lexical_variant_length
from note_nlp_enhanced group by note_nlp_concept_id
) nne1
left outer join
(select note_nlp_concept_id, count(*) as n_annotations, count(distinct note_id) as n_notes
from note_nlp_enhanced
where certainty = 'Positive' group by note_nlp_concept_id) pos -- positive mentions
on nne1.note_nlp_concept_id = pos.note_nlp_concept_id
left outer join (
select note_nlp_concept_id,
array_agg(lower_lexical_variant) as top_five_lower_lexical_variants --psql
from ranked_lexical_variants where lexical_variant_rank <= 5 group by note_nlp_concept_id
) top_vars --top lower cases lexical variants matched
on nne1.note_nlp_concept_id = top_vars.note_nlp_concept_id
left outer join
(select note_nlp_concept_id, count(*) as n_annotations, count(distinct note_id) as n_notes
from note_nlp_enhanced where certainty = 'Negated' group by note_nlp_concept_id) neg --negative mentions
on nne1.note_nlp_concept_id = neg.note_nlp_concept_id
left outer join concept c on c.concept_id = nne1.note_nlp_concept_id --get concept name and codes
order by nne1.n_annotations desc
;