Defendable data is the core structure of the platform. Every number here traces to primary literature and is re-checkable. The data justifies the platform — and, more importantly, it proves why the owned longitudinal record is the moat.
Diabetic foot ulcers (DFUs) — the strongest initial proof base:
| Fact | Figure | Source |
|---|---|---|
| People with a DFU each year | ~18.6M global · ~1.6M US | JAMA 2023 (Armstrong, Tan, Boulton et al.) |
| Lifetime risk of a DFU (person with diabetes) | 19–34% | JAMA 2023; NEJM 2017 |
| Share of diabetes-related lower-limb amputations preceded by a DFU | ~80% | JAMA 2023 |
| DFUs ending in lower-extremity amputation | up to 20% | JAMA 2023 |
| Become infected | ~50% (per the review) | JAMA 2023 |
| 5-year mortality after a DFU | ~30% | JAMA 2023 + multiple |
| 5-year mortality after major amputation | >70% | JAMA 2023 + multiple |
| US direct treatment cost | $9–13B / year | JAMA 2023 + multiple |
A diabetic foot ulcer carries a 5-year mortality worse than many common cancers, and a major amputation worse than most. This is not a cosmetic problem. It is a survival problem.
The burden is rising — people live longer with more medical complexity, so lifetime risk climbs. A larger, older, more-complex diabetic population means more feet at risk for longer.
The single most important number for the product:
| After a DFU heals, share who recur | Source | |
|---|---|---|
| within 1 year | ~40% | 2024 systematic review + meta-analysis |
| within 3 years | ~60% | 〃 (corroborated, prospective cohort) |
| within 5 years | ~65% | 〃; Feb 2025 meta-analysis (42% @1y, 65% @5y); NEJM 2017 |
A healed wound is not the end of the story — it is the beginning of a higher-risk timeline. Two-thirds recur within five years. The clinically-correct model of a foot is not "healthy vs ulcer" — it is remission, a state that must be watched, indefinitely, because it reverts.
A one-time photo check cannot hold remission. A longitudinal record can. That is the product.
A generic model answers a weak question: "does this photo look bad?" The longitudinal record answers the strong one:
"What changed — compared with this same person, same foot, same wound site, same surgery, same shoe history, same risk profile?"
Foot risk is temporal. The signal is not the image; the signal is the delta. Change detection against a personal baseline is where the owned record beats any generic classifier — and it compounds: the record is worth more every week because there is more history to compare against. That is defensible in a way a model checkpoint never is.
The Foot Passport is the owned longitudinal foot record — not "upload a photo and let AI check it," but "build and own a portable record of your foot health over time." It preserves:
changes, swelling, redness, callus, pressure marks, nail changes, surgical sites.
offloading history, wound history, vascular workup, infection history.
shoes, pressure-relief devices, fit changes, wear patterns, clinician recommendations.
swelling, pressure, walking tolerance, caregiver notes.
was prescribed, what wound care was used, what improved, what came back.
files. The moat is the schema · habit · continuity · portability · interpretability · receipts — not the disk. So OpenFootLab is not "secure storage"; it is a foot-risk record system that helps people, caregivers, and care teams see change earlier and communicate better.
podiatry classifier, a future FDA-cleared model — all swappable inside the system. Models get cheaper and better and get replaced. The record remains. (See Getting-the-Most-from-MedGemma.md — the model's job is to read the delta against the baseline, not to be the whole system.)
The model can change. The storage vendor can change. The phone can change. The record remains.
The bigger, more defendable company is a longitudinal foot-risk platform with an owned portable record at the center — not a "diabetes app" (that box is too small).
Foot-Risk-Beyond-Diabetes.md).*The moat is not the NAS. Not the model. Not even the app. The moat is the trusted foot timeline — owned by the person, structured enough for care, portable enough for life, and useful before the crisis.*
Not medical advice. Figures cited from the linked peer-reviewed literature; they vary by population and method. Where a figure is stated in the JAMA 2023 review but not independently re-derived here, it is attributed to the review.