P.R.I.S.M. — Pipeline for Real-time Infection Signal Mining

# P.R.I.S.M. — R Sandbox
# Select biomarkers in the right panel, then click Run.
# Your selected markers are available as the 'markers' data frame.

# View loaded markers
print(markers)

# Combined d' under binormal model (independence assumed)
d_combined <- sqrt(sum(markers$da^2))
auc_combined <- pnorm(d_combined / sqrt(2))

cat("\n=== Composite Biomarker Performance ===\n")
cat("Combined d':", round(d_combined, 3), "\n")
cat("Combined AUC:", round(auc_combined, 4), "\n")

# Optimal operating point
opt_sn <- pnorm(d_combined / 2)
opt_sp <- pnorm(d_combined / 2)
cat("Sensitivity:", round(opt_sn * 100, 1), "%\n")
cat("Specificity:", round(opt_sp * 100, 1), "%\n")

# Likelihood ratios
lr_pos <- opt_sn / (1 - opt_sp)
lr_neg <- (1 - opt_sn) / opt_sp
cat("LR+:", round(lr_pos, 2), "\n")
cat("LR-:", round(lr_neg, 4), "\n")

# Fire score assessment
cat("\n=== Fire Score ===\n")
cat("Avg fire score:", round(mean(markers$fire_score), 1), "/100\n")
cat("Classification:", ifelse(mean(markers$fire_score) > 50, "FIRE-dominant", "SMOKE-dominant"), "\n")

Console

Click "Activate R Runtime" to initialize WebR (~20MB download, cached after first load).\nOnce ready, selected biomarkers will auto-sync as the 'markers' data frame.

The God Particle

🔬

Select biomarkers to begin composite analysis