README.md

In r4subrisk: Risk Quantification Engine for Clinical Submission Readiness

r4subrisk

r4subrisk is the risk quantification engine in the R4SUB ecosystem.

It uses an FMEA-inspired framework (Probability x Impact x Detectability) to quantify submission risk, build risk registers, track mitigations, and emit standardized R4SUB evidence rows.

It answers the question:

What are the key risks to submission readiness, how severe are they, and are they being addressed?

Core Concepts

FMEA Risk Model

Each risk is scored on three dimensions (1--5 scale):

• Probability: likelihood the issue persists to submission
• Impact: regulatory consequence if unresolved
• Detectability: difficulty of catching the issue (1=easy, 5=hard)

RPN (Risk Priority Number) = Probability x Impact x Detectability (range 1--125)

RPN Bands

| RPN | Level | Interpretation | |-----|-------|----------------| | 80--125 | critical | Immediate action required | | 40--79 | high | Must resolve before submission | | 15--39 | medium | Plan mitigation | | 1--14 | low | Monitor |

Installation

pak::pak(c("R4SUB/r4subcore", "R4SUB/r4subrisk"))

Quick Start

library(r4subcore)
library(r4subrisk)

# From a manual risk register
risks <- data.frame(
  risk_id       = c("R001", "R002"),
  description   = c("Missing SDTM variables", "Unmapped ADaM derivations"),
  category      = c("data_quality", "traceability"),
  probability   = c(4, 3),
  impact        = c(5, 4),
  detectability = c(2, 3)
)
rr <- create_risk_register(risks)
rr

# Or derive risks automatically from evidence
risk_items <- evidence_to_risks(evidence)
rr <- create_risk_register(risk_items)

# Compute scores and emit evidence
scores <- compute_risk_scores(rr)
ctx <- r4sub_run_context(study_id = "ABC123", environment = "DEV")
ev <- risk_register_to_evidence(rr, ctx = ctx)

Core Functions

| Function | Purpose | |---|---| | risk_config_default() | FMEA scales, RPN bands, severity mappings | | classify_rpn() | Classify an RPN value into a risk level | | create_risk_register() | Build a risk register with RPN + levels | | evidence_to_risks() | Derive risk items from r4subcore evidence | | compute_risk_scores() | Aggregate risk metrics (mean/max RPN, distribution) | | risk_indicator_summary() | Summary indicator table | | risk_register_to_evidence() | Emit r4subcore-compatible evidence rows | | apply_mitigations() | Update risks with mitigations, recompute RPN | | compare_risk_registers() | Trend analysis between snapshots |

Design Principles

• FMEA-grounded: industry-standard risk methodology
• Evidence-first: all risk items emit r4subcore evidence rows
• Mitigation-aware: track and measure risk reduction over time
• Transparent: every RPN is decomposable into P x I x D

License

MIT

r4subrisk documentation built on March 4, 2026, 1:07 a.m.