TOML config and schema files that replace the YAML versions.

test/config/construct_spine_from_influenza_cases.toml

@@ -0,0 +1,25 @@
+projectname = "influenza-cases-spine"
+description = "Construct a spine containing only people who have had influenza"
+output_directory = "output"  # During testing this expands to: /path/to/SpineBasedRecordLinkage.jl/test/output/
+spine = {datafile = "", schemafile = "schema/spine.toml"}
+append_to_spine = true
+construct_entityid_from = ["firstname", "lastname", "birthdate"]
+tables = [{tablename = "influenza_cases", datafile = "data/influenza_cases.csv", schemafile = "schema/influenza_cases.toml"}]
+
+[[criteria]]
+tablename  = "influenza_cases"
+exactmatch = {firstname = "firstname", middlename = "middlename", lastname = "lastname", birthdate = "birthdate"}
+
+[[criteria]]
+tablename  = "influenza_cases"
+exactmatch = {firstname = "firstname", lastname = "lastname", birthdate = "birthdate"}
+
+[[criteria]]
+tablename   = "influenza_cases"
+exactmatch  = {firstname = "firstname", birthdate = "birthdate"}
+approxmatch = [{datacolumn = "lastname", spinecolumn = "lastname", distancemetric = "levenshtein", threshold = 0.3}]
+
+[[criteria]]
+tablename   = "influenza_cases"
+exactmatch  = {lastname = "lastname", birthdate = "birthdate"}
+approxmatch = [{datacolumn = "firstname", spinecolumn = "firstname", distancemetric = "levenshtein", threshold = 0.5}]

test/config/link_all_health_service_events.toml

@@ -0,0 +1,45 @@
+projectname = "health-service-usage"
+description = "Construct a spine from 3 health service usage tables and link the tables to the spine."
+output_directory = "output"  # During testing this expands to: /path/to/SpineBasedRecordLinkage.jl/test/output/
+spine = {datafile = "", schemafile = "schema/spine.toml"}
+append_to_spine = true
+construct_entityid_from = ["firstname", "lastname", "birthdate"]
+tables = [
+    {tablename = "hospital_admissions",     datafile = "data/hospital_admissions.csv",     schemafile = "schema/hospital_admissions.toml"},
+    {tablename = "emergency_presentations", datafile = "data/emergency_presentations.csv", schemafile = "schema/emergency_presentations.toml"},
+    {tablename = "influenza_cases",         datafile = "data/influenza_cases.csv",         schemafile = "schema/influenza_cases.toml"}
+]
+
+[[criteria]]
+tablename  = "hospital_admissions"
+exactmatch = {firstname = "firstname", lastname = "lastname", birthdate = "birthdate"}
+
+[[criteria]]
+tablename  = "emergency_presentations"
+exactmatch = {firstname = "firstname", lastname = "lastname", birthdate = "birthdate"}
+
+[[criteria]]
+tablename   = "emergency_presentations"
+exactmatch  = {birthdate = "birthdate"}
+approxmatch = [
+        {datacolumn = "firstname", spinecolumn = "firstname", distancemetric = "levenshtein", threshold = 0.3},
+        {datacolumn = "lastname",  spinecolumn = "lastname",  distancemetric = "levenshtein", threshold = 0.3}
+]
+
+[[criteria]]
+tablename  = "influenza_cases"
+exactmatch = {firstname = "firstname", middlename = "middlename", lastname = "lastname", birthdate = "birthdate"}
+
+[[criteria]]
+tablename  = "influenza_cases"
+exactmatch = {firstname = "firstname", lastname = "lastname", birthdate = "birthdate"}
+
+[[criteria]]
+tablename   = "influenza_cases"
+exactmatch  = {firstname = "firstname", birthdate = "birthdate"}
+approxmatch = [{datacolumn = "lastname", spinecolumn = "lastname", distancemetric = "levenshtein", threshold = 0.3}]
+
+[[criteria]]
+tablename   = "influenza_cases"
+exactmatch  = {lastname = "lastname", birthdate = "birthdate"}
+approxmatch = [{datacolumn = "firstname", spinecolumn = "firstname", distancemetric = "levenshtein", threshold = 0.5}]

test/config/link_influenza_cases_to_events.toml

@@ -0,0 +1,44 @@
+projectname = "influenza-cases-link"
+description = "Link the emergencies, admissions and influenza tables to the spine that was constructed from the influenza cases table."
+output_directory = "output"  # During testing this expands to: /path/to/SpineBasedRecordLinkage.jl/test/output/
+spine = {datafile = "output/spine.tsv", schemafile = "schema/spine.toml"}
+append_to_spine = false
+tables = [
+    {tablename = "hospital_admissions",     datafile = "data/hospital_admissions.csv",     schemafile = "schema/hospital_admissions.toml"},
+    {tablename = "emergency_presentations", datafile = "data/emergency_presentations.csv", schemafile = "schema/emergency_presentations.toml"},
+    {tablename = "influenza_cases",         datafile = "data/influenza_cases.csv",         schemafile = "schema/influenza_cases.toml"}
+]
+
+[[criteria]]
+tablename  = "emergency_presentations"
+exactmatch = {firstname = "firstname", lastname = "lastname", birthdate = "birthdate"}
+
+[[criteria]]
+tablename   = "emergency_presentations"
+exactmatch  = {birthdate = "birthdate"}
+approxmatch = [
+    {datacolumn = "firstname", spinecolumn = "firstname", distancemetric = "levenshtein", threshold = 0.3},
+    {datacolumn = "lastname",  spinecolumn = "lastname",  distancemetric = "levenshtein", threshold = 0.3}
+]
+
+[[criteria]]
+tablename = "hospital_admissions"
+exactmatch = {firstname = "firstname", lastname = "lastname", birthdate = "birthdate"}
+
+[[criteria]]
+tablename  = "influenza_cases"
+exactmatch = {firstname = "firstname", middlename = "middlename", lastname = "lastname", birthdate = "birthdate"}
+
+[[criteria]]
+tablename  = "influenza_cases"
+exactmatch = {firstname = "firstname", lastname = "lastname", birthdate = "birthdate"}
+
+[[criteria]]
+tablename   = "influenza_cases"
+exactmatch  = {firstname = "firstname", birthdate = "birthdate"}
+approxmatch = [{datacolumn = "lastname", spinecolumn = "lastname", distancemetric = "levenshtein", threshold = 0.3}]
+
+[[criteria]]
+tablename   = "influenza_cases"
+exactmatch  = {lastname = "lastname", birthdate = "birthdate"}
+approxmatch = [{datacolumn = "firstname", spinecolumn = "firstname", distancemetric = "levenshtein", threshold = 0.5}]

test/schema/emergency_presentations.toml

@@ -0,0 +1,13 @@
+name        = "emergency_presentations"
+description = "Each row specifies a presentation at a hospital emergency department"
+primarykey  = ["hospitalid", "campusid", "presentationdate", "patientid"]
+columns = [
+    {name = "hospitalid",  description = "Hospital ID", datatype = "Int",    iscategorical = false, isrequired = true, isunique = false, validvalues = "1:9999"},
+    {name = "campusid",    description = "Campus ID",   datatype = "Int",    iscategorical = false, isrequired = true, isunique = false, validvalues = "1:9999"},
+    {name = "patientid",   description = "Patient ID",  datatype = "Int",    iscategorical = false, isrequired = true, isunique = false, validvalues = "Int"},
+    {name = "presentationdate", description = "Presentation date",  datatype = "Date", iscategorical = false, isrequired = true, isunique = false, validvalues = "Date"},
+    {name = "firstname",   description = "First name",  datatype = "String", iscategorical = false, isrequired = false, isunique = false, validvalues = "String"},
+    {name = "middlename",  description = "Middle name", datatype = "String", iscategorical = false, isrequired = false, isunique = false, validvalues = "String"},
+    {name = "lastname",    description = "Last name",   datatype = "String", iscategorical = false, isrequired = false, isunique = false, validvalues = "String"},
+    {name = "birthdate",   description = "Birth date",  datatype = "Date",   iscategorical = false, isrequired = false, isunique = false, validvalues = "Date"}
+]

test/schema/hospital_admissions.toml

@@ -0,0 +1,14 @@
+name        = "hospital_admissions"
+description = "Each row specifies an admission to a hospital"
+primarykey  = ["hospitalid", "campusid", "admissiondate", "dischargedate", "patientid"]
+columns = [
+    {name = "hospitalid",  description = "Hospital ID", datatype = "Int",    iscategorical = false, isrequired = true, isunique = false, validvalues = "1:9999"},
+    {name = "campusid",    description = "Campus ID",   datatype = "Int",    iscategorical = false, isrequired = true, isunique = false, validvalues = "1:9999"},
+    {name = "patientid",   description = "Patient ID",  datatype = "Int",    iscategorical = false, isrequired = true, isunique = false, validvalues = "Int"},
+    {name = "admissiondate", description = "Date of admission",  datatype = "Date", iscategorical = false, isrequired = true, isunique = false, validvalues = "Date"},
+    {name = "dischargedate", description = "Date of discharge",  datatype = "Date", iscategorical = false, isrequired = true, isunique = false, validvalues = "Date"},
+    {name = "firstname",   description = "First name",  datatype = "String", iscategorical = false, isrequired = false, isunique = false, validvalues = "String"},
+    {name = "middlename",  description = "Middle name", datatype = "String", iscategorical = false, isrequired = false, isunique = false, validvalues = "String"},
+    {name = "lastname",    description = "Last name",   datatype = "String", iscategorical = false, isrequired = false, isunique = false, validvalues = "String"},
+    {name = "birthdate",   description = "Birth date",  datatype = "Date",   iscategorical = false, isrequired = false, isunique = false, validvalues = "Date"}
+]

test/schema/influenza_cases.toml

@@ -0,0 +1,12 @@
+name = "influenza_cases"
+description = "Each row specifies report of a case of influenza"
+primarykey = "reportid"
+columns = [
+    {name = "reportid",   description = "Report ID", datatype = "Int", iscategorical = false, isrequired = true, isunique = true, validvalues = "Int"},
+    {name = "reportdate", description = "Date of case notification", datatype = "Date", iscategorical = false, isrequired = true, isunique = false, validvalues = "Date"},
+    {name = "firstname",  description = "First name",  datatype = "String", iscategorical = false, isrequired = false, isunique = false, validvalues = "String"},
+    {name = "middlename", description = "Middle name", datatype = "String", iscategorical = false, isrequired = false, isunique = false, validvalues = "String"},
+    {name = "lastname",   description = "Last name",   datatype = "String", iscategorical = false, isrequired = false, isunique = false, validvalues = "String"},
+    {name = "birthdate",  description = "Birth date",  datatype = "Date",   iscategorical = false, isrequired = false, isunique = false, validvalues = "Date"},
+    {name = "patient_postcode", description = "Postcode of the patient's residential address", datatype = "Int", iscategorical = false, isrequired = true, isunique = false, validvalues = "3000:3999"}
+]

test/schema/spine.toml

@@ -0,0 +1,11 @@
+name = "spine"
+description = "Linkage spine. Each row specifies a person"
+primarykey = "EntityId"
+columns =[
+    {name = "EntityId",    description = "Entity ID",   datatype = "UInt",   iscategorical = false, isrequired = true, isunique = true,  validvalues = "UInt"},
+    {name = "firstname",   description = "First name",  datatype = "String", iscategorical = false, isrequired = true, isunique = false, validvalues = "String"},
+    {name = "middlename",  description = "Middle name", datatype = "String", iscategorical = false, isrequired = true, isunique = false, validvalues = "String"},
+    {name = "lastname",    description = "Last name",   datatype = "String", iscategorical = false, isrequired = true, isunique = false, validvalues = "String"},
+    {name = "birthdate",   description = "Birth date",  datatype = "Date",   iscategorical = false, isrequired = true, isunique = false, validvalues = "Date"},
+    {name = "patient_postcode", description = "Postcode of the patient's residential address", datatype = "Int", iscategorical = false, isrequired = true, isunique = false, validvalues = "3000:3999"}
+]