Technical Notes

Implementation verification documentation (tests, constraints, and evidence basis) for all RLT Platform calculators.

Validation Summary

All platform calculators are verified through automated testing to confirm implementation correctness. Tests verify that code behavior matches stated formulas, clinical thresholds, and expected directional relationships. This validation confirms technical accuracy but does not replace clinical validation or physician judgment.

Total Tests
112
Status
All Passing ✓
Module Coverage
17/17 (100%)
Last Validated
2026-01-24

What We Test

Baseline Verification

Every calculator is tested with standard clinical inputs to verify results render correctly and match expected ranges.

  • Result display visibility
  • Value ranges within clinical norms
  • Chart and table generation

Sensitivity Testing

Each input is varied independently to confirm outputs change in the clinically expected direction.

  • Higher activity → higher dose
  • Lower eGFR → higher nephrotoxicity risk
  • Higher Gleason → higher risk category

Edge Case Testing

Boundary values and extreme inputs are tested to ensure proper handling and conservative capping.

  • Minimum/maximum input values
  • Threshold boundaries (e.g., PSA at 10, 20)
  • Conservative caps enforced

100-Case Stratified Testing

Randomized inputs across the full clinical range to detect edge cases and regression issues.

  • Seeded random for reproducibility
  • All major input combinations covered
  • 15/15 calculator modules covered

Module Coverage

Calculator Test Methods Tests Key Validations Status
Organ Dosimetry Unit + edge + 100-case 7 Kidney/salivary doses, cycle scaling, activity effects
Voxel Dosimetry Unit + behavior + 100-case 6 S-value calculations, SUV effects, weight scaling, BED
Serial Dosimetry Unit + behavior + 100-case 5 Time-activity curves, mono/bi-exponential models, organ mass
AE Predictor Unit + behavior + 100-case 7 Xerostomia, nephrotoxicity, hematologic risk, conservative caps
PSA Trajectory Unit + behavior + 100-case 5 Nadir prediction, decline %, response types, monitoring schedule
Response Tool Unit + edge + 100-case 6 RECIST 1.1 criteria, PCWG3 2+2 rule, overall response
Risk Stratification Unit + 100-case 5 PSA50/PSA90 rates, PFS estimation, risk factor adjustments
Staging (NCCN/TNM) Unit + edge + 100-case 7 Risk categories, grade groups, response assessments
Prostate Cancer ADT Unit + 100-case 30 NCCN/STAMPEDE/LATITUDE/CAPRA scoring, drug selection, contraindications
Cost-Effectiveness Unit + edge + 100-case 9 ICER calculations, QALY modeling, region/comparator effects
Trial Simulation Unit + 100-case 4 VISION trial response curves, PSA/SUV effects
Cohort Simulator Unit + edge + 100-case 5 Power calculations, cohort size effects, dropout rates
Response Estimator Unit + edge + 100-case 5 Response probability, feature importance, uncertainty bounds
Genomic Stratification Unit + edge + 100-case 6 HRR mutation effects, BRCA status, therapy selection
Consensus Pathways Unit + edge + 100-case 5 Guideline concordance, treatment sequencing

Scientific Validation

Dosimetry Calculations

  • S-values: Lu-177 MIRD phantom data (ICRP 133, OLINDA/EXM)
  • Half-life: Physical 6.647 days; biological organ-specific
  • BED formula: Dale model for continuous low-dose-rate irradiation
  • α/β ratios: Kidneys 2.6 Gy, salivary 3 Gy, tumor 10 Gy

Clinical Decision Rules

  • RECIST 1.1: 30% decrease for PR, 20%+5mm increase for PD
  • PCWG3: 2+2 rule for bone progression confirmation
  • NCCN: PSA + Gleason + T-stage scoring algorithm
  • LATITUDE/STAMPEDE: High-volume disease criteria

Data Sources: S-values loaded from data/s-values-lu177.json containing MIRD reference organ self-dose and cross-dose values. Clinical thresholds match published guidelines (NCCN 2024, ESMO 2023).

Constraints & Limitations

Input Ranges

  • • PSA: 0.1 – 10,000 ng/mL
  • • Administered activity: 1 – 20 GBq
  • • Patient weight: 40 – 200 kg
  • • Age: 18 – 100 years
  • • eGFR: 15 – 150 mL/min/1.73m²
  • • Bone metastases: 0 – 100 lesions

Not Validated For

  • • Neuroendocrine prostate cancer (NEPC)
  • • Pediatric patients
  • • Non-PSMA-expressing tumors
  • • Combination with external beam RT dosimetry
  • • Real-time treatment guidance
  • • Regulatory submission data

Conservative Caps Applied

  • Hematologic AE risk: max 43%
  • Xerostomia risk: max 9%
  • Nephrotoxicity risk: max 6%

Reference Populations

  • VISION trial cohort (mCRPC)
  • TheraP trial data
  • ICRP reference adult male

Drug Safety Checks

  • Abiraterone: cardiac/liver contraindications
  • Enzalutamide: seizure history
  • GnRH agonists: spinal cord compression

Production Readiness Statement

This platform has been verified for implementation correctness within declared assumptions and input ranges. All 112 automated tests pass, covering 17/17 modules with 100-case stratified testing on all 15 clinical calculators.

Important: This validation confirms the code correctly implements stated formulas and logic. It does not constitute clinical validation, regulatory approval, or substitute for physician judgment. All outputs should be verified against published evidence and institutional protocols before clinical use.

RLT Platform v10 · Validation Run: 2026-01-24 · Test Framework: Node.js node:test + jsdom