Table 1

Definition of inclusion criteria

Inclusion criteriaDefinition
Any study designArticles that report the development and/or validation of a prediction model based on experimental studies or observational studies. This includes randomised controlled trials, prospective and retrospective cohort, case–control studies and case–cohort studies.
Using at least 2 predictors for risk predictionArticles that report the development and/or validation of a prediction model using at least two predictors. Articles that use imaging or speech parameters as structured data plus other predictors such as clinical, demographics, histological and genetic risk scores features will be included.
Any data sourcesArticles that report the development and/or validation of a prediction model using any structured data source, for example, electronic medical records, administrative claims data and individual patient data meta-analysis data.
Any supervised ML techniqueArticles that report the use of any ML technique for development and/or validation of a prediction model. We will consider as a ML technique, a statistical technique based on advanced computational capacity and lower human intervention. More specifically, we will focus on supervised ML techniques.
Patient health-related outcomesArticles that report the development and/or validation of a prediction model whose main outcome is on an individual patient level. We will include articles assessing diagnosis, prognosis and health services performance, such as length of stay or triage assessment.
All outcome measures formatArticles that report the development and/or validation of a prediction model whose main outcome has one of the following formats: continuous, binary, ordinal, multinomial and time-to-event.
  • ML, machine learning.