Derive percentile-based biomarker cutoffs using adjusted risk — derive_cutoffs

Implements a two-cutoff approach to classify subjects into rule-out, rule-in, or indeterminate zones using confounder-adjusted risk scores.

Usage

derive_cutoffs_percentile(
  adjust_result,
  biomarker,
  low_thresh = 0.05,
  high_thresh = 0.7,
  low_pct = 0.8,
  high_pct = 0.2
)

Arguments

adjust_result: Output list from adjust_score(), containing the dataset with biomarker, adjusted_risk, and outcome.
biomarker: Character, name of biomarker column.
low_thresh: Numeric, threshold for defining the low-risk group (default = 0.05). Very small datasets may produce too few low-risk subjects at extreme thresholds, triggering fallback cutoffs.
high_thresh: Numeric, threshold for defining the high-risk group (default = 0.70). Very small datasets may produce too few high-risk subjects at extreme thresholds, triggering fallback cutoffs.
low_pct: Numeric, percentile used to derive the rule-out cutoff within the low-risk group (default = 0.80).
high_pct: Numeric, percentile used to derive the rule-in cutoff within the high-risk group (default = 0.20).

Value

List with rule-out cutoff (c_L), rule-in cutoff (c_H), classification vector, and augmented dataset.

Details

Method

The function uses adjusted_risk (from adjust_score()) to define clinically meaningful risk groups.
Low-risk group: subjects with adjusted risk < low_thresh (default 0.05), representing individuals unlikely to have disease. The rule-out cutoff (c_L) is defined as the 80th percentile of biomarker values within this group.
High-risk group: subjects with adjusted risk greater than high_thresh (default 0.70), representing individuals likely to have disease. The rule-in cutoff (c_H) is defined as the 20th percentile of biomarker values within this group.
If either group is too small, fallback cutoffs are used based on the global biomarker distribution (5th and 95th percentiles).
Final classification: biomarker < c_L "Negative"; biomarker > c_H "Positive"; otherwise "Indeterminate".

References

Giacomucci, G., Crucitti, C., Ingannato, A., et al. (2025). The two cut-offs approach for plasma p-tau217 in detecting Alzheimer's disease. Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring, 17:e70116.

Hazan, J., Liu, K.Y., Isaacs, J.D., Howard, R. (2025). Cut-points and gray zones: The challenges of integrating Alzheimer's disease plasma biomarkers into clinical practice. Alzheimer's & Dementia, 21:e70113.

Author

Bhrigu Kumar Rajbongshi, Seyed Ehsan Saffari and Nastaran Marzban

Examples

n <- 500
set.seed(123)

# outcome
DISEASE <- rbinom(n, 1, 0.35)

# biomarker depends on disease
PTAU <- rnorm(n, mean = 25 + 10 * DISEASE, sd = 8)

# confounders (optional noise)
AGE <- round(rnorm(n, 73, 7))
SEX <- sample(c(0, 1), n, replace = TRUE)
BMI <- rnorm(n, 25, 4)
HTN <- rbinom(n, 1, 0.3)
DM <- rbinom(n, 1, 0.2)

df <- data.frame(PTAU, AGE, SEX, BMI, HTN, DM, DISEASE)

res <- adjust_score(
df,
biomarker = "PTAU",
outcome = "DISEASE",
confounders = c("AGE","SEX","BMI","HTN","DM")
)
#> number of iterations= 704 
#> Setting levels: control = 0, case = 1
#> Setting direction: controls < cases

cutoffs <- derive_cutoffs_percentile(res, biomarker = "PTAU")
#> Number of cases below low_thresh = 0.05 is too small, so fallback to 5th percentile of biomarker was used. Consider using another low_thresh value if your dataset is small.
cutoffs$c_L; cutoffs$c_H
#> [1] 13.30872
#> [1] 36.84693
table(cutoffs$class)
#> 
#> Indeterminate      Negative      Positive 
#>           373            25           102