Skip to contents

Liver Markers

Source: vignettes/liver_markers.Rmd
liver_markers.Rmd

Scope

Compute liver-related indices (FLI, NFS, APRI, FIB-4, BARD, ALBI, MELD-XI) with configurable handling for missing data and extreme values.

When to use

  • You have routine liver labs and anthropometry (BMI, waist, TG, GGT, AST, ALT, platelets, albumin, bilirubin, creatinine, diabetes flag) and want standard steatosis/fibrosis scores.
  • You need built-in NA policy controls, numeric coercion, and optional extreme-value scanning/capping.
  • You want MELD-XI, ALBI, and common non-invasive fibrosis indices in one call for screening or modeling.

Inputs

  • data: data.frame/tibble with required inputs.
  • col_map: named list mapping required keys to columns (units assumed, no conversion): BMI (kg/m^2), waist (cm), TG (mg/dL), GGT (U/L), age (years), AST (U/L), ALT (U/L), platelets (10^9/L), albumin (g/L), diabetes (0/1 or logical), bilirubin (mg/dL), creatinine (mg/dL).
  • na_action: keep (default) propagates NA; omit drops rows missing required inputs; error aborts; ignore treated as keep; warn keeps but issues high-missingness warnings via na_warn_prop.
  • check_extreme: set TRUE to scan inputs with extreme_action (warn, cap, error, ignore, NA) using default bounds unless extreme_rules overrides them. Defaults: BMI 10-70; waist 40-200 cm; TG 10-1500 mg/dL; GGT 1-2000 U/L; age 18-120; AST/ALT 1-5000; platelets 10-1000 (10^9/L); albumin 15-60 g/L; bilirubin 0.1-40 mg/dL; creatinine 0.2-20 mg/dL.
  • Numeric coercion: mapped columns (except diabetes) are coerced to numeric; non-finite values become NA before policies. diabetes is coerced to integer 0/1 with a warning if values are outside {0,1,TRUE,FALSE}.

Quick start 

library(HealthMarkers)
library(tibble)

liver <- tibble::tibble(
  BMI           = c(24, 30, 33),
  waist         = c(80, 100, 110),
  triglycerides = c(150, 260, 90),   # mg/dL
  GGT           = c(30, 55, 18),
  age           = c(45, 58, 67),
  AST           = c(25, 60, 42),
  ALT           = c(20, 45, 28),
  platelets     = c(250, 190, 210),  # 10^9/L
  albumin       = c(42, 38, 40),     # g/L
  diabetes      = c(FALSE, TRUE, 1),
  bilirubin     = c(1.0, 0.9, 1.3),  # mg/dL
  creatinine    = c(0.9, 1.1, 1.0)   # mg/dL
)

liver_markers(
  data = liver,
  col_map = list(
    BMI = "BMI", waist = "waist", TG = "triglycerides", GGT = "GGT", age = "age",
    AST = "AST", ALT = "ALT", platelets = "platelets", albumin = "albumin", diabetes = "diabetes",
    bilirubin = "bilirubin", creatinine = "creatinine"
  ),
  na_action = "keep"
)
#> # A tibble: 3 × 7
#>     FLI   NFS  APRI  FIB4  BARD  ALBI MELD_XI
#>   <dbl> <dbl> <dbl> <dbl> <int> <dbl>   <dbl>
#> 1  27.9 -27.5 0.25   1.01     1 -2.76    8.20
#> 2  87.0 -21.8 0.789  2.73     3 -2.45   10.0 
#> 3  73.8 -22.6 0.5    2.53     3 -2.52   10.8

Handling missingness and extremes

  • Missing values: keep/ignore propagate NA; omit drops rows missing required inputs; warn keeps but logs high-missingness by column; error aborts on missing required inputs.
  • Extreme values: with check_extreme = TRUE, scan against defaults (or extreme_rules); choose extreme_action = "warn"|"cap"|"error"|"ignore"|"NA" to warn, cap in-range, abort, skip, or blank-out flagged inputs. Non-finite values become NA before checks.

Outputs

  • Returns a tibble with: FLI, NFS, APRI, FIB4, BARD, ALBI, MELD_XI.
  • Bilirubin is converted to umol/L internally for ALBI; other inputs use provided units.
  • Zero or nonpositive platelets, TG, GGT, bilirubin, creatinine, or negative ALT trigger warnings and yield NA in affected indices.

Verbose diagnostics 

old_opt <- options(healthmarkers.verbose = "inform")
liver_markers(
  data = liver,
  col_map = list(
    BMI = "BMI", waist = "waist", TG = "triglycerides", GGT = "GGT", age = "age",
    AST = "AST", ALT = "ALT", platelets = "platelets", albumin = "albumin", diabetes = "diabetes",
    bilirubin = "bilirubin", creatinine = "creatinine"
  ),
  verbose = TRUE
)
#> liver_markers(): preparing inputs
#> liver_markers(): column map: BMI -> 'BMI', waist -> 'waist', TG -> 'triglycerides', GGT -> 'GGT', age -> 'age', AST -> 'AST', ALT -> 'ALT', platelets -> 'platelets', albumin -> 'albumin', diabetes -> 'diabetes', bilirubin -> 'bilirubin', creatinine -> 'creatinine'
#> liver_markers(): results: FLI 3/3, NFS 3/3, APRI 3/3, FIB4 3/3, BARD 3/3, ALBI 3/3, MELD_XI 3/3
#> # A tibble: 3 × 7
#>     FLI   NFS  APRI  FIB4  BARD  ALBI MELD_XI
#>   <dbl> <dbl> <dbl> <dbl> <int> <dbl>   <dbl>
#> 1  27.9 -27.5 0.25   1.01     1 -2.76    8.20
#> 2  87.0 -21.8 0.789  2.73     3 -2.45   10.0 
#> 3  73.8 -22.6 0.5    2.53     3 -2.52   10.8
options(old_opt)

Tips

  • Confirm triglycerides are in mg/dL (not mmol/L) and albumin in g/L.
  • Enable check_extreme = TRUE with extreme_action = "cap" to bound implausible values before scoring; use warn during QA to see flags.
  • For modeling-ready outputs, prefer na_action = "omit"; use keep or warn when inspecting data quality.