AegisFunc

AegisFunc is a tool for Spatio-temporal epidemiology based on OHDSI CDM v5.4.0.
AegisFunc originated from the Aegis project that is an open-source spatial analysis tool based on CDM. https://github.com/ABMI/Aegis.

Requirements

OHDSI CDM v5.4.0

We use latitude and longitude fields in Location table (updated on v5.4.0).

INLA package

INLA package is not supported by CRAN.
You can install manually from the r-inla repository

install.packages("INLA", repos = c(getOption("repos"), INLA = "https://inla.r-inla-download.org/R/stable"), dep = TRUE)

R v4.2

Latest version of INLA requires R v4.2 or above

Installation

You can install the development version of AegisFunc like so:

devtools::install()

Load

Load AegisFunc package before use.

library(AegisFunc)

Check your CDM database

Database connection

Connection string of SQL Server: “jdbc:sqlserver://localhost:1433;databaseName=cdm;user=user;password=password”
Connection string of PostgreSQL: “jdbc:postgresql://localhost:5432/cdm?user=user&password=password”

We have tested with SQL Server 2019.

dbms <- "sql server"
path_to_driver <- getwd()
connection_string <- "jdbc:sqlserver://[SERVER_IP]:[SERVER_PORT];user=[USER_ID];password=[USER_PW];databaseName=[CDM_DB_NAME]"

conn_info <- get_connection_details(
  dbms = dbms,
  path_to_driver = path_to_driver,
  connection_string = connection_string
)

Check CDM version

You must confirm your CDM version.
Because we only support for CDM v5.4.0 or above.

conn_info <- conn_info
cdm_database_schema <- "[CDM_DB_SCHEMA]"

cdm_source <- get_cdm_source(
  conn_info = conn_info,
  cdm_database_schema = cdm_database_schema
)
cdm_version <- cdm_source[c("cdm_version")]

Load cohort list

Query defined cohort list.

conn_info <- conn_info
result_database_schema <- "[RESULT_DB_SCHEMA]"

cohort_list <- get_cohort_list_table(
  conn_info = conn_info,
  result_database_schema = result_database_schema
)

Forecasting (temporal model)

Step 01. Prepare Data

Get cohort table from CDM/Atlas database

model <- "temporal"
conn_info <- conn_info
cdm_database_schema <- "[CDM_DB_SCHEMA]"
result_database_schema <- "[RESULT_DB_SCHEMA]"
target_cohort_definition_id <- "1"
outcome_cohort_definition_id <- "2"

cohort_table <- get_cohort_analysis_table(
  model = model,
  conn_info = conn_info,
  cdm_database_schema = cdm_database_schema,
  result_database_schema = result_database_schema,
  target_cohort_definition_id = target_cohort_definition_id,
  outcome_cohort_definition_id = outcome_cohort_definition_id
)

Step 02. Calculate forecasting

Calculate forecasting

model <- "temporal"
table <- cohort_table
observation_end_date <- "2008-01-01"
prediction_end_date <- "2009-08-01"
variables_type = "day,season,month,week"

deriv <- calculate_forecasting(
  model = model,
  table = table,
  observation_end_date = observation_end_date,
  prediction_end_date = prediction_end_date,
  variables_type = variables_type
)

Step 03. Plot

Plot with data

data <- deriv

plot <- get_forecasting_plot(
  data = data
)
plot

Disease Map/Cluster (spatial model)

Step 01. Prepare Data

Get cohort table from CDM/Atlas database

model <- "spatial"
conn_info <- conn_info
cdm_database_schema <- "[CDM_DB_SCHEMA]"
result_database_schema <- "[RESULT_DB_SCHEMA]"
target_cohort_definition_id <- "1"
outcome_cohort_definition_id <- "2"
cohort_start_date <- "2020-01-01"
cohort_end_date <- "2020-12-31"
time_at_risk_start_date <- "0"
time_at_risk_end_date <- "0"
time_at_risk_end_date_panel <- "cohort_start_date" # "cohort_start_date" or "cohort_end_date"

cohort_table <- get_cohort_analysis_table(
  model = model,
  conn_info = conn_info,
  cdm_database_schema = cdm_database_schema,
  result_database_schema = result_database_schema,
  target_cohort_definition_id = target_cohort_definition_id,
  outcome_cohort_definition_id = outcome_cohort_definition_id,
  cohort_start_date = cohort_start_date,
  cohort_end_date = cohort_end_date,
  time_at_risk_start_date = time_at_risk_start_date,
  time_at_risk_end_date = time_at_risk_end_date,
  time_at_risk_end_date_panel = time_at_risk_end_date_panel
)

Read geo data

name <- "KOR" # "GADM" or "KOR"
country <- "KOR"
level <- "2"

geo <- get_geo_data(
  name = name,
  country = country,
  level = level
)

Map cohort table (lat/long) with geo data

latlong <- cohort_table
geo <- geo

geo_map <- map_latlong_geo(
  latlong = latlong,
  geo = geo
)

Arrange table

model <- "spatial"
table <- geo_map

table_arr <- calculate_count_with_geo_oid(
  model = model,
  table = table
)

Step 02. Adjustment

Adjustment for age and sex

model <- "spatial"
table <- table_arr
mode <- "std" # "std" or "crd"
fraction <- "100000"
conf_level <- "0.95"

table_adj <- calculate_adjust_age_sex_indirectly(
  model = model,
  table = table,
  mode = mode,
  fraction = fraction,
  conf_level = conf_level
)

Step 03-1. Calculate disease map

Generate graph file from geo data

geo <- geo

graph_file_path <- trans_geo_to_graph(
  geo = geo
)

Calculate disease map

model <- "spatial"
table <- table_adj
graph_file_path <- graph_file_path

deriv <- calculate_disease_map(
  model = model,
  table = table,
  graph_file_path = graph_file_path
)

Step 03-2. Calculate disease cluster

Calculate disease cluster

model <- "spatial"
table <- table_adj

deriv <- calculate_disease_cluster(
  model = model,
  table = table
)

Step 04. Plot

Merge geo data with derivatives

geo <- geo
deriv_arr <- deriv$arranged_table

data <- merge_geo_with_deriv(
  geo = geo,
  deriv = deriv_arr
)

Plot with data

data <- data
stats <- deriv$stats
color_type <- "colorQuantile"
color_param <- base::list(
  palette = "Reds",
  domain = NULL,
  bins = 7,
  pretty = TRUE,
  n = 4,
  levels = NULL,
  ordered = FALSE,
  na.color = "#FFFFFF",
  alpha = FALSE,
  reverse = FALSE,
  right = FALSE
)

plot <- get_leaflet_map(
  data = data,
  stats = stats,
  color_type = color_type,
  color_param = color_param
)
plot

Disease Map/Cluster (spatio-temporal model)

Step 01. Prepare Data

Get cohort table from CDM/Atlas database

model <- "spatio-temporal"
conn_info <- conn_info
cdm_database_schema <- "[CDM_DB_SCHEMA]"
result_database_schema <- "[RESULT_DB_SCHEMA]"
target_cohort_definition_id <- "1"
outcome_cohort_definition_id <- "2"
cohort_start_date <- "2020-01-01"
cohort_end_date <- "2020-12-31"
time_at_risk_start_date <- "0"
time_at_risk_end_date <- "0"
time_at_risk_end_date_panel <- "cohort_start_date" # "cohort_start_date" or "cohort_end_date"

cohort_table <- get_cohort_analysis_table(
  model = model,
  conn_info = conn_info,
  cdm_database_schema = cdm_database_schema,
  result_database_schema = result_database_schema,
  target_cohort_definition_id = target_cohort_definition_id,
  outcome_cohort_definition_id = outcome_cohort_definition_id,
  cohort_start_date = cohort_start_date,
  cohort_end_date = cohort_end_date,
  time_at_risk_start_date = time_at_risk_start_date,
  time_at_risk_end_date = time_at_risk_end_date,
  time_at_risk_end_date_panel = time_at_risk_end_date_panel
)

Read geo data

name <- "KOR" # "GADM" or "KOR"
country <- "KOR"
level <- "2"

geo <- get_geo_data(
  name = name,
  country = country,
  level = level
)

Map cohort table (lat/long) with geo data

latlong <- cohort_table
geo <- geo

geo_map <- map_latlong_geo(
  latlong = latlong,
  geo = geo
)

Arrange table

model <- "spatio-temporal"
table <- geo_map

table_arr <- calculate_count_with_geo_oid(
  model = model,
  table = table
)

Step 02. Adjustment

Adjustment for age and sex

model <- "spatio-temporal"
table <- table_arr
mode <- "std" # "std" or "crd"
fraction <- "100000"
conf_level <- "0.95"

table_adj <- calculate_adjust_age_sex_indirectly(
  model = model,
  table = table,
  mode = mode,
  fraction = fraction,
  conf_level = conf_level
)

Step 03-1. Calculate disease map

Generate graph file from geo data

geo <- geo

graph_file_path <- trans_geo_to_graph(
  geo = geo
)

Calculate disease map

model <- "spatio-temporal"
table <- table_adj
graph_file_path <- graph_file_path

deriv <- calculate_disease_map(
  model = model,
  table = table,
  graph_file_path = graph_file_path
)

Step 03-2. Calculate disease cluster

Calculate disease cluster

model <- "spatio-temporal"
table <- table_adj

deriv <- calculate_disease_cluster(
  model = model,
  table = table
)

Step 04. Plot

Check names of derivatives

years <- names(deriv)

Merge geo data with derivatives

data <- base::list()

for(i in 1:length(years)) {
  geo <- geo
  deriv_arr <- deriv[years[i]][[1]]$arranged_table

  data <- append(data, list(merge_geo_with_deriv(
    geo = geo,
    deriv = deriv_arr
  )))
}

names(data) <- years

Make plots

plot <- base::list()

for(i in 1:length(years)) {
  data <- data[years[i]][[1]]
  stats <- deriv[years[i]][[1]]$stats
  color_type <- "colorQuantile"
  color_param <- base::list(
    palette = "Reds",
    domain = NULL,
    bins = 7,
    pretty = TRUE,
    n = 9,
    levels = NULL,
    ordered = FALSE,
    na.color = "#FFFFFF",
    alpha = FALSE,
    reverse = FALSE,
    right = FALSE
  )


  plot <- append(plot, list(get_leaflet_map(
    data = data,
    stats = stats,
    color_type = color_type,
    color_param = color_param
  )))
}

names(plot) <- years

Check names of derivatives

idx <- names(plot)

Plot with index

plot[idx[1]][[1]]

jungsungjae/AegisFunc

AegisFunc

Requirements

OHDSI CDM v5.4.0

INLA package

R v4.2

Installation

Load

Check your CDM database

Database connection

Check CDM version

Load cohort list

Forecasting (temporal model)

Step 01. Prepare Data

Step 02. Calculate forecasting

Step 03. Plot

Disease Map/Cluster (spatial model)

Step 01. Prepare Data

Step 02. Adjustment

Step 03-1. Calculate disease map

Step 03-2. Calculate disease cluster

Step 04. Plot

Disease Map/Cluster (spatio-temporal model)

Step 01. Prepare Data

Step 02. Adjustment

Step 03-1. Calculate disease map

Step 03-2. Calculate disease cluster

Step 04. Plot