AI models during Data extraction (from text)

Model name

Cooperative models*

Type of SRs/topic

Description

Models that support extraction of study characteristics

Study identifier 

-

 Universal

Unique code used to identify research work, like NCT number 

Study name

-

 Universal

Refers to the trial name, added by the authors of the study, e.g. coming from the first letters of the study name  

Study objective

-

 Universal

Extracts the study objective.

Study conclusion

-

 Universal

Extracts the main conclusions of the study 

Study limitations

-

 Universal

Extracts the study limitations as reported by the study authors

Country name

-

 Universal

Extracts countries where the study was conducted

Was this study multi-center or single-center

-

 Standard/human

Defines if the study was single- or multicenter 

Study design e.g. RCT, prospective, observational, cross-sectional etc

-

 Universal

Refers to the methodologies used in research to gather the data needed to explore a specific question, e.g. observational or randomized-controlled trial. 

Study design comment

-

 Universal

More detailed information about study design

Study phase e.g I, II, III, IV

-

 Standard/human

Describes the clinical trial phase (I-IV)

Statistical methodology

-

 Standard/human

Describes the statistical methodology which was used in the study

Efficacy analysis type

-

 Standard/human

Defines the type of efficacy analysis - ITT, mITT, per protocol

Safety analysis type

-

 Standard/human

Defines the type of safety outcomes analysis - ITT, mITT, per protocol

Cross-over design (yes/no)

-

 Standard/human

Specifies whether crossover design was used in the study

Data reported for cross-over

-

 Standard/human

Was the data reported separately for the cross-over? (yes or no)

Study funding/sponsor information

-

 Universal

Funder or sponsor of the study  

 Study period

-

 Universal

Describes study duration (range of years, as reported in the study)

Timepoint (follow-up length)

-

 Universal

One instance for every time point/follow-up length reported in the study

Support critical appraisal

Validity of statistical analysis

-

 Standard/human

Did the analysis include an intention-to-treat analysis? If so, was this appropriate and were appropriate methods used to account for missing data?

Recommended for ROB Domain 3: Missing outcome data

SQ [D3.1] Were data for this outcome available for all, or nearly all, participants randomised?

Validity of outcome reporting

-

 Standard/human

Is there any evidence suggesting that the authors measured more outcomes than reported?

Recommended for Domain 5: Risk of bias in the selection of the reported result

SQ [5.1] Were the data that produced this result analysed following a pre-specified analysis plan finalized before unblinded outcome data were available for analysis?

Validity of withdrawals

-

 Standard/human

Were there any unexpected imbalances in drop-outs between groups? If so, were they explained or adjusted for?

Recommended for ROB Domain 3: Missing outcome data

SQ [D3.1] Were data for this outcome available for all, or nearly all, participants randomised?

Validity of baseline characteristics

-

 Standard/human

Were the groups similar at the outset of the study in terms of prognostic factors, for example, severity of disease?

Recommended for ROB: Domain 1: Risk of bias arising from the randomization process

SQ [D1.3] Did baseline differences between intervention groups suggest a problem with the randomization process?

Study allocation blinding e.g. open-label, single, double, triple-blind

-

 Standard/human

Were the care providers, participants and outcome assessors blind to treatment allocation? If any of these people were not blinded, what might be the likely impact on the risk of bias (for each outcome)?) 

Recommended for ROB Domain 2: Risk of bias due to deviations from the intended interventions 

SQ [D2.1] Were participants aware of their assigned intervention during the trial?

SQ [D2.2] Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?

Study assessment blinding (yes, no)

-

 Standard/human

Recommended for ROB Domain 2: Risk of bias due to deviations from the intended interventions 

SQ [D2.1] Were participants aware of their assigned intervention during the trial?

SQ [D2.2] Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?

Study allocation concealment

-

 Standard/human

Was randomisation carried out appropriately?

Recommended for ROB: Domain 1: Risk of bias arising from the randomization process

Models that support extraction of concepts related to study populations

Study population

-

 Standard/human

Extracts data related to the study population, such as disease.

Inclusion criterion

-

 Universal

1. Specifies what the inclusion criteria were for patients in the study.

2. For Systematic reviews- extract inclusion criteria

Exclusion criterion

-

 Universal

1. Specifies what the exclusion criteria were for patients in the study.

2. For systematic reviews - extract exclusion criteria

Number of patients screened

-

 Standard/human

Extract the number of patients screened for eligibility before the study begins.

Number of patients included 

-

 Standard/human

Extracts the number of patients who were randomized. To extract the number of patients included in each study group, please use the model: ‘Study arm: Number of patients in the study arm’.

Number of patients that completed the study

-

 Standard/human

Extracts the number of patients who completed the study.

Previous therapy details

-

 Standard/human

Refers to the information if any prior therapies were used 

Subgroup: Name of the subgroup

-

 Standard/human

Extracts subgroups that were evaluated in the study

Baseline characteristic: name of the baseline characteristic 

Yes

 Standard/human

Extracts all the variable names that are included in the baseline characteristics (e.g. sex, age, Body mass index, region etc)

Baseline characteristic: Unit

Yes

 Standard/human

Extracts the units of available baseline variables (e.g. centimetres for height)

Baseline characteristic: Central tendency type 

Yes

 Standard/human

Extracts the available type of the central tendency (e.g. mean, median etc.) for the arm comparison 

Baseline characteristic: Central tendency value 

Yes

 Standard/human

Extracts the central tendency value for the arm comparison

Baseline characteristic: Variance type 

Yes

 Standard/human

Extracts the available type of variance (e.g. range, IQR, SE  etc.) for the arm comparison

Baseline characteristic: Variance value

Yes

 Standard/human

Extracts the variance value for the arm comparison

Models that support extraction of concepts related to interventions/study arms

Study arm: Name of the intervention

Yes

 Standard/human

Extracts the detailed name of the intervention

Study arm: Dose of the intervention

Yes

 Standard/human

Extracts the dose for each of the study arms

Study arm: Unit of the dose of the intervention

Yes

 Standard/human

Extracts the dose unit for each of the study arms

Study arm: Route of administration of the intervention

Yes

 Standard/human

Extracts the route of administration for each of the study arms

Study arm: Frequency of administration of the intervention

Yes

 Standard/human

Extracts the frequency of administration for each of the study arms

Study arm: Name of the study arm

Yes

 Standard/human

Extracts the names of each of the study arms

Study arm: Number of patients in the study arm

Yes

 Standard/human

Extracts the number of patients who were randomized in each of the study arms.

Treatment time

-

 Standard/human

Duration of treatment, e.g. 10 days and details on when the treatment was discontinued, e.g. 10 days or until discharge

Concomitant treatment

-

 Standard/human

Concomitant treatment administered along with the studied intervention

Models that support extraction of outcome data

Outcome: Name of the outcome

Yes

 Standard/human

Extracts the name of the outcome from the article

Outcome: What was the outcome measured with

Yes

 Standard/human

How was the outcome measured? E.g. questionnaire (specify its name), blood test, etc.

Outcome: Unit

Yes

 Standard/human

Extracts the outcome unit (e.g. points, cm,  etc)

Outcome: Timepoint (follow-up length)

Yes

 Standard/human

Extracts the specific time point or window when authors collected data for the outcome 

Models that support extraction of outcome data - Comparison level

Outcome: Effect measure

Yes

 Standard/human

Extracts the effect measure type (e.g. define if it’s OR/RR/HR etc) for the arm comparison

Outcome: Effect measure value 

Yes

 Standard/human

Extracts the value of effect measure (e.g. the value for OR/RR/HR etc) for the arm comparison

Outcome: p-value

Yes

 Standard/human

Extracts the p-value (for the specific outcome) for the arm comparison

Outcome: Variance type

Yes

 Standard/human

Extracts the type of the variance (e.g. define if it’s 95%CI, range etc) for the arm comparison

Outcome: Variance value

Yes

 Standard/human

Extracts the value of variance (e.g. value for 95%CI) for the arm comparison

Models that support extraction of outcome data - Arm  level

Outcome: Central tendency type

Yes

 Standard/human

Extracts the available type of the central tendency (e.g. mean, median etc.) for the arm comparison for the outcomes 

Outcome-Study arm: value

Yes

 Standard/human

Extracts the outcome value for each of the study arms