Inside the Healing Trajectory: How WoundScribe Turns Visit Data into a Predictive Curve
product
A deep dive into the healing analytics engine — how measurements, tissue, perfusion, and comorbidities become a predictive curve that flags stalled wounds early.
The healing trajectory isn't a chart of what happened — it's a prediction of what will happen if nothing changes. This is a look inside how that curve is built and what it does for the clinician at the point of care.
The inputs: seven signals per wound, per visit
Every visit contributes a structured record:
- Geometry — length, width, depth, area, volume from calibration-free imaging
- Tissue composition — percent granulation, slough, eschar, epithelium
- Exudate — character and volume
- Periwound — maceration, erythema, callus
- Perfusion context — ABI, TcPO₂, pulse exam when available
- Systemic context — HbA1c, albumin, comorbidities
- Treatment applied — dressing, debridement, offloading, CTPs
Each of these enters the model as a time-stamped feature, not a free-text sentence.
The model: expected healing versus observed healing
For a wound of a given etiology, location, size, and patient context, there is an expected trajectory — a curve of area reduction over time drawn from population-level healing data. The healing analytics dashboard overlays the patient's observed trajectory on that expected curve visit by visit.
Two lines matter:
- Expected — where a comparable wound in a comparable patient should be at week N.
- Observed — where this wound actually is at week N.
Divergence between them is the signal.
The trigger: early divergence, not late failure
Most stalled wounds are recognized late — when granulation stops, when infection appears, when the toe is already dusky. The trajectory model surfaces divergence one to three visits earlier, because it reads small changes in area velocity and tissue composition together, not one at a time.
When divergence crosses a threshold, the visit note carries a flag: reassess plan, check perfusion, consider referral.
The handoff: from flag to action
A flag is only useful if it lands in the workflow. The trajectory output is embedded in the AI-powered EMR for wound care so the next visit's assessment opens with the current trajectory status, the last three measurements, and the model's recommended reassessments.
Why calibration-free imaging matters here
A predictive curve is only as good as its measurements. Rulers and manual tracing introduce inter-clinician variance that swamps the signal the model is trying to detect. Point-and-capture wound measurement removes that variance so a 12% area reduction reads as 12%, not as noise.
What it means for outcomes
Earlier divergence detection means earlier plan changes — different offloading, earlier CTP, earlier vascular referral. For diabetic foot ulcers specifically, where 80% of amputations begin as an ulcer, the value of the curve is measured in limbs.