LLM-Assisted Title/Abstract Screening for Systematic Reviews

Documentation for package ‘screenllm’ version 0.1.0

Help Pages

audit_disagreements	Surface strong LLM-human disagreements for a manual audit
backend_mock	Mock backend for tests
backend_ollama	Ollama backend
build_prompt	Build a screening prompt for one record
check_setup	Verify local setup for 'screenllm'
clear_cache	Clear cached LLM scores for a project
custom_ensemble	Define a custom LLM ensemble
default_ensemble	The paper's default LLM ensemble
default_ensemble_light	A "light" ensemble that runs on a laptop
define_criteria	Define screening inclusion criteria
delete_artefact	Delete an artefact (or a whole project)
delete_model	Delete an Ollama model to free up disk space
detect_gpu	Detect a usable GPU for Ollama inference
estimate_runtime	Estimate wall-clock time for a ranking run
export_report	Render the screening report as a self-contained HTML document
export_worksheet	Export the human-screening worksheet
find_duplicates	Detect and remove duplicate records
gpu_status	Live NVIDIA GPU status snapshot
install_prereqs	One-stop setup for a fresh machine
launch_app	Launch the full end-to-end Shiny app
launch_screening_app	Launch the interactive screening Shiny app
list_projects	List existing project names
list_project_artefacts	Canonical artefact filenames
load_artefact	Load an artefact from a project directory
load_toy_cbfm_criteria	Load the ready-made criteria for the toy CBFM corpus
ollama_catalog	Curated catalog of Ollama models useful for screening
ollama_health	Ping the Ollama server
ollama_installed_models_detail	Installed Ollama models with disk-space metadata
plan_screening	Plan the human screening set with the SAFE stopping rule
pull_model	Pull an Ollama model to the local server
rank_records	Rank a corpus with an LLM ensemble
read_decisions	Read completed screening decisions
read_records	Read a corpus of records from disk
save_artefact	Save an artefact into a project directory
summarise_screening	Summarise a completed screening session