Harmonising frailty index deficits

This repository gives detail of the harmonisation process for variables (deficits) included in frailty indices from randomised controlled trials and from observational cohort studies. The file "Original_and_harmonised_FI_terms.csv" contains each of the frailty index deficits from each of the included studies. The columns are as follows: Study: The name of the study from which the frailty index was taken study_type: Observational (cohort) study or Randomised Controlled Trial Original_term: This is the original term used within the description of the frailty index, extracted verbatim from the published version of the included study Mapped_concept: This is the Chat-GPT suggested term when prompted to assign synonyms to the full list of deficits within a given category. The prompts used to derive these terms are included within this repository Harmonised_concept: This column was created following manual review of each of the 'mapped concepts' against each of the original terms. Where this is the same as the mapped concept, this means that the ChatGPT suggested term was retained after manual review. Where it differs, this means that the term was manually changed following review. This could be for one of several reasons: (1) the ChatGPT term had altered the meaning of the term, (2) Two or more ChatGPT suggested terms corresponded to the same overall concept, and were therefore edited to the same term, (3) Two or more distinct original concepts had been mapped to the same concept by ChatGPT, in which case these were separated. Type: Detailed the type of deficit (e.g. functional deficits, symptoms, health conditions, labotory deficits). These were all manually assigned from the original term Category: This details the category of deficit in terms of organ system (e.g. cardiovascular, cognitive, urological) or other category (e.g. physical function) Cat: This is the category of the study (observational cohort, cardiovascular trial, other trial) order_y: This is a composite varaible comprising the category and the harmonised concept, used to order the y axis of figure 1 by the frequenct of the deficit within the dataset and within categories in_order: Numerical variable used in plotting figure 1 to order the axis. The file "Script for figures" gives the R script use to produce the manuscript figures from the file "Original_and_harmonised_FI_terms.csv". The remaing files, each with the prefix "chatgpt" give the text used for ChatGPT prompts in order to derive the mapped concepts.