Skip to main content

Table 6 Selection of semantic rules used for extraction in GATE format, each with example phrases they match

From: Automated screening of research studies for systematic reviews using study characteristics

Characteristic

 

Study design

Example

We

conducted

a

hospital-based

prospective cohort

study

Rule

 

(verb)

({Token.string==~“(?i)a”}|{Token.string==~“(?i)an”})

({Token})[0,1]

(study)

(types)?

Population

Example

Cohort

of

665 Danish pregnant women

Rule

({Token.string==~“(?i)cohort”}|{Token.string==~“(?i)sample”}|{Token.string==~“(?i)total”}|{Token.string==~“(?i)samples”}|{Token.string==~“(?i)cross-sectional”}|{Token.string==~“(?i)subsample”})

{Token.string==~“(?i)of”}

(population)

Exposure

Example

Perfluorinated compounds

in relation to

birth weight

Rule

(folic_variations)

{Lookup.majorType==“relations”}

(birthoutcomes)

Outcome

Example

association

with

 

miscarriage

Rule

{Lookup.majorType==“relations”}

({Token.string==~“(?i)between”}|(with))

({Token.string==~“(?i)the”})?

(birthoutcomes)

Confounding factor

Example

Adjusting

for

 

covariates, including maternal pre-pregnancy BMI, smoking, education, and birth weight

Rule

(adjustment)

({Token.string = ~“(?i)for”}|{Token.string = ~“(?i)by”})

({Token})[0,3]

(cnf)

Country

Example

In

Alberta

,

Canada

Rule

{Token.string==~“(?i)in”}

({Token}) [1]

{Token.string==“,”}

({Lookup.majorType==“countries”}|{Lookup.majorType==“states”})