◆ Methodology
How the Vital Pulse Index is built
This page is the human-readable methodology. The machine-readable canonical copies live at /api/vital-pulse/methodology (current version) and /api/vital-pulse/methodology/{version} (audit access for any prior version).
Current version
Version 1.0.0. Published 6 May 2026. Approved by: Vitagri internal review (CEO Rob Ward acting as methodology approver). An independent methodology panel may be appointed in future and would review and amend this methodology accordingly; any such amendment bumps the version. Until then, the panel-chair authorisation step on the monthly publish flow is performed by the CEO via an out-of-band token.
Peer-reviewed methodology paper: not yet submitted. The DOI will surface here if and when a paper is accepted; until then the fullPaperUrl field is empty in the API response.
Composite construction
The Vital Pulse Index is a weighted average of six pillars on a 0–100 scale. Pillar weights sum to 100 and are versioned in lockstep with the methodology document.
| # | Pillar | Weight | Cadence |
|---|---|---|---|
| 1 | Soil health trajectory | 20% | Annual ingest, monthly publish (carried forward between annual readings) |
| 2 | Crop nutrient density | 25% | Monthly |
| 3 | Polyphenol and bioactive trend | 10% | Monthly recompute on rolling 60-month window |
| 4 | Public procurement quality | 15% | Annual school data + monthly NHS interim |
| 5 | Diet-related disease burden | 15% | Annual ingest, monthly publish (carried forward) |
| 6 | Policy and market momentum | 15% | Monthly |
Traffic-light thresholds
Status bands match the AHDB Soil Health Scorecard convention exactly: Investigate for 0–39 (red), Review for 40–69 (amber), On track for 70–100 (green). Pillar 5 inverts the underlying metric (lower disease cost → higher score) before banding so the colour direction is consistent with the other five.
Velocity
Vital Pulse is velocity-aware: every reading carries both a level (the score, 0–100) and a velocity (smoothed 12-month rolling change, expressed in points-per-month). Velocity is computed at publish time from the prior twelve months of immutable per-pillar readings:
velocity_t = (score_t − score_{t−12}) / 12
A qualitative direction label is derived from velocity: improving (≥ +0.17 pts/mo, equivalent to ≈ +2 pts/year), stable (between −0.17 and +0.17), deteriorating (≤ −0.17 pts/mo). Until twelve months of history exist, velocity is reported as null and the dashboard renders the badge as "history accumulating".
Acceleration (the second-order derivative — change in velocity month-on-month) is reserved for a future methodology version; in v1.0 the field is null in published readings.
Per-pillar velocity rules
Pillars 1 (Soil) and 5 (Disease cost) refresh annually rather than monthly. Velocity for those is computed only on actual annual readings; interpolated months carry the prior velocity forward unchanged, and the dashboard renders an explicit "annual data" note so readers do not misread interpolated values as fresh monthly movement. Pillars 2, 3 and 6 use the standard 12-month rolling computation on monthly readings. Pillar 4 (Procurement) blends monthly NHS interim readings with the annual school-food data.
Sources
Each pillar's primary sources are listed on its drill-down page and in the canonical configuration at /vital-pulse/assets/vp-pillar-config.json. Source status (available, partially-available, scoping, not-yet-approached, pending-pilot-launch) is part of the configuration; if a primary source is unavailable at the point of a scheduled publish, the publish flow refuses to interpolate the score and instead offers a "carried forward" or "data unavailable" option, surfaced explicitly to readers in the gauge tile.
Bionutrient Institute attribution
Pillar 2 (Crop nutrient density) uses Bionutrient Institute's 1–100 per-food scale as the underlying nutrient-density rating method. Whenever a reading on that pillar draws on Bionutrient Meter data, three attribution surfaces activate automatically: the dashboard tile renders a visible "Source: Bionutrient Institute" line; the API response sets pillars[1].biAttributed to true and emits a X-Source: Bionutrient Institute header; and the embed widget includes the attribution in its visible footer.
Governance
Vital Pulse v1.0 is published under Vitagri internal review. The CEO acts as methodology approver and issues the out-of-band authorisation token required for each monthly publish; the publish endpoint refuses without it. An independent methodology panel may be appointed in future and would assume both roles; any change to governance is recorded on this page and in /api/vital-pulse/panel. Vitagri Org Ltd discloses any material commercial interest that could shape a reading in the same place.
Retrospective readings (Jan 2025 – Apr 2026)
Vital Pulse begins its public publishing run in May 2026. To give the index a meaningful track record at launch — and to make the 12-month rolling velocity computable from day one — the months from January 2025 to April 2026 are populated as retrospective readings. Each retrospective reading is computed from publicly-available data that genuinely existed at the time: UKCEH Countryside Survey for soil; Pulse Brain literature meta-analysis for crop nutrient density and bioactives; Government Buying Standards plus dated NHS hospital food and school food compliance reports for procurement; NHS / OHID / ONS published series for diet-related disease cost; Defra and DHSC policy registers plus retailer ESG disclosures for momentum.
Every retrospective reading is explicitly tagged in the API response (retrospective: true on the Reading and on each PillarReading) and on the dashboard (a small "Retrospective" badge on the tile). Velocities derived from retrospective readings are valid because the underlying scores are computed from real, dated data; they are not extrapolations or model output.
Retrospective scores are reasoned editorial estimates against publicly-available data, signed off by the methodology approver. They are not mechanically generated from a fixed algorithm — different reasonable analysts could land on slightly different numbers within the bands. The numbers shown are Vitagri's transparent best read; corrections are welcome at rob.ward@vitagri.org.
Once a retrospective reading is written it is immutable in the same way as a forward-published reading. The retrospective tag is permanent and cannot be removed by re-running the backfill — the audit log records every retrospective write with the actor, timestamp and IP.
Audit and immutability
Once a monthly reading has been published, the contents of vp:current and the corresponding vp:reading:{year}:{month} Blob are immutable. CI rejects any migration script that mutates a published reading or per-pillar Blob. Every write to a vp:* key records an entry in the admin activity log, including actor, action, key, timestamp, IP, methodology version and approver name at time of write.
Version history
| Version | Published | Notes |
|---|---|---|
| 1.0.0 | 6 May 2026 | Initial published methodology. Six pillars, weights summing to 100, traffic-light bands matching AHDB. Monthly publishing cadence; smoothed 12-month rolling velocity at ±0.17 pts/month direction thresholds. Approved by Vitagri internal review; independent methodology panel deferred. |
Citation
Use the canonical citation string provided on the Data page. Once the methodology paper is accepted, the formal academic citation replaces it.