Skip to content

reporting-checklist.md

Latest commit

 

History

History
68 lines (57 loc) · 3.53 KB

reporting-checklist.md

File metadata and controls

68 lines (57 loc) · 3.53 KB
Raw

BP-Card Reporting Checklist (Modules 0–3)

A practical checklist for completing the BP-Card. Modules are filled top to bottom; Module 3 is reported separately for each calibration pathway.

Module 0 — Clinical purpose and use context

  • Model name and version
  • Estimation family (hemodynamics-derived / data-driven BP regression / ABP waveform reconstruction)
  • Calibration paradigm (calibration-free / calibration-based)
  • Model type (ML / DL / hybrid / other)
  • Input data type (handcrafted features / time series / images / demographic / other)
  • Output variable (SBP / DBP / MAP)
  • Signal modality (PPG / ECG / BCG / BioZ / IPG / iPPG / other)
  • Sensor specification and location
  • Intended use (research / home or ambulatory / adjunctive clinical / clinical pilot)

Module 1 — Dataset and cohort reporting

  • Data source (public / private), cohort size, recording sessions, total paired samples
  • Population type (healthy / hypertensive / ICU / other)
  • Demographics (age, sex, BMI, height, weight, ethnicity / region)
  • Medical history / comorbidities
  • Medication (class / dose / regimen)
  • Known sensitivities (motion / low perfusion / arrhythmia / sensor displacement / skin tone)
  • Skin tone / Fitzpatrick and ambient light (PPG-based)
  • Measurement phases (%) and BP profile distributions (%)
  • SBP / DBP (mean ± SD, range)
  • Baseline SBP / DBP (mean ± SD, range)
  • ΔSBP / ΔDBP (mean ± SD; subject-specific or population-based)
  • Preprocessing pipeline
  • Signal quality filtering / exclusion criteria
  • Fiducial points / feature extraction
  • Excluded samples after cleaning
  • BP distribution reported before and after cleaning

Module 2 — Validation strategy

  • Evaluation method (subject-wise hold-out / subject-wise CV / LOSO / LOOCV / record-wise / sample-wise) — subject-wise is the minimum standard
  • Data split details
  • Unseen-subject test set (yes / no)
  • External validation, independent cohort / site (yes / no)
  • Reference standard and reference device (manufacturer / model)
  • Measurement conditions (resting / ambulatory / dynamic / long-term / ICU)
  • Body position, activity state, rest period prior to measurement
  • Baseline comparator and comparator result
  • Minimum acceptable standard met
  • Pathway-specific analyses reported separately (Module 3A and/or 3B)

Module 3A — Performance reporting, calibration-free models

  • Models evaluated (baseline model and proposed / final model)
  • Standardized performance table, stratified by BP group (All / Normal / Prehypertension / Stage 1 / Stage 2)
  • Metrics reported: ΔBP±SD, MD±SD, MAD, MAPD, CP5, CP10, CP15, R
  • Correlation / regression scatter plot
  • Bland–Altman plot (with normality check and 95% CIs on limits of agreement)

Module 3B — Performance reporting, calibration-based models

  • Models evaluated (baseline model and proposed / final model)
  • Calibration condition and calibration type (one-point / two-point / multi-point)
  • Static test and dynamic test (yes / no)
  • Time since calibration evaluated (and interval in days)
  • Recalibration trigger (if any)
  • Standardized performance table, stratified by condition (Static / Dynamic / Time after calibration)
  • Metrics reported: ΔBP±SD, MD±SD, MAD, MAPD, CP5, CP10, CP15, R
  • Correlation / regression scatter plot (BP change relative to calibration)
  • Bland–Altman plot (with normality check and 95% CIs on limits of agreement)