Job request: 23846

version: '3.0'

# Ignore this`expectation` block. It is required but not used, and will be removed in future versions.
expectations:
  population_size: 1000

actions:
## Short data report work ##

  generate_study_population_AnyCode_SDR:
    run: ehrql:v1 generate-dataset analysis/dataset_definition_any_antenatal.py --output output/dataset_anycode.csv.gz
    outputs:
      highly_sensitive:
        dataset: output/dataset_anycode.csv.gz


### tables for any antenatal code 5 year cohort
  describe_SDR_cohort_with_any_antenatal_code:
    run: r:latest analysis/AntenatalCohort_baseline_table_SDR.R
    needs: [generate_study_population_AnyCode_SDR]  
    outputs:
      moderately_sensitive:
          patient_numbers: output/patient_numbers_antenatal_cohort_SDR.csv
          cohort_contin_overall: output/SDR_table_continuous_vars_antenatal_cohort.csv
          cohorts_catagorical_overall: output/SDR_table_categorical_antenatal_cohort.csv

# plots of antenatal data 
  describe_antenatal_cohort_plots: 
    run: r:latest analysis/AntenatalCohort_timeseries_plots.R
    needs: [generate_study_population_AnyCode_SDR]
    outputs:
      moderately_sensitive:
          antenatal_first_csv: output/first_antenatal_codes_by_month_year.csv
          antenatal_first_plot: output/first_antenatal_codes_by_month_year.jpeg
          antenatal_plot_code_freq: output/antenatalCohort_box-and-whisker_code_counts.jpeg

  describe_SDR_cohort_with_any_antenatal_code_minmax_meanSD:
    run: r:latest analysis/AntenatalCohort_baseline_table_SDR_minmax.R
    needs: [generate_study_population_AnyCode_SDR]
    outputs:
      moderately_sensitive:
          patient_nums: output/patient_numbers_AntenatalCohort_SDR_minmax.csv
          cohorts_minmax: output/SDR_table_minmax_AntenatalCohort.csv
          cohorts_meanSD: output/SDR_table_meanSD_AntenatalCohort.csv

# describe delivery code episodes - Antenatal Cohort
  describe_antenatal_cohort_delivery_episodes:
    run: r:latest analysis/AntenatalCohort_num_deliveries_each.R
    needs: [generate_study_population_AnyCode_SDR]
    outputs:
      moderately_sensitive:
          by_NAs: output/df_count_NAs_for_del_episodes.csv
          not_missing_dates: output/df_count_del_episodes.csv

# describe outcomes for delivery episodes 
  describe_antenatal_cohort_outcomes:
    run: r:latest analysis/AntenatalCohort_outcomes.R
    needs: [generate_study_population_AnyCode_SDR]
    outputs:
      moderately_sensitive:
          outcome_plot_data: output/monthly_rates_outcomes.csv
          outcome_plot: output/monthly_outcome_rates.jpeg

# describe covid positive and vaccination for delivery episodes 
  describe_antenatal_cohort_covidvacc:
    run: r:latest analysis/AntenatalCohort_covid_and_vacc_overtime.R
    needs: [generate_study_population_AnyCode_SDR]
    outputs:
      moderately_sensitive:
          covid_plot_data: output/monthly_rates_covid_and_vaccs.csv
          covid_plot: output/monthly_covid_rates.jpeg