rPPG, FHIR, HIPAA, LOINC, POS, SDNN — our domain is heavy on three-letter abbreviations. Here's the running list we share with every new engineer, clinician, or customer on their first day.
A secret string that authenticates your SDK or server. Prefix ssp_live_ in production, ssp_test_ in sandbox. Rotate at least yearly and never ship a live key inside a client binary.
The ratio of low-frequency to high-frequency HRV power (LF/HF), a rough indicator of sympathetic vs. parasympathetic tone. We report it alongside HRV for stress dashboards.
Business Associate Agreement. A HIPAA contract that lets a covered entity share PHI with a vendor. SmartScanPro signs BAAs on the Pro plan and above.
A plot that shows the difference between two measurement methods against their mean. The standard way to report agreement between a new sensor and a gold-standard reference in clinical validation.
Beats per minute. Our default unit for heart rate. UCUM code /min. LOINC 8867-4 for "Heart rate by pulse oximetry".
A classical rPPG method (de Haan & Jeanne, 2013) that combines the red, green, and blue color channels to suppress motion-induced intensity changes while preserving the pulse signal.
A 0–1 value we return with every reading, reflecting signal-to-noise ratio, motion, and lighting quality. Below 0.6 we recommend you discard or re-capture.
US remote-patient-monitoring billing codes covering device setup, data transmission, and clinician review. Many of our RPM customers bill against these.
Fast Healthcare Interoperability Resources. The modern standard for exchanging clinical data. We produce FHIR R4 Observation resources out of the box.
The FHIR resource representing a single measurement (heart rate, SpO₂, etc.). Includes a LOINC code, a value, a unit, an effective time, and a subject reference.
A six-tier classification of skin phototype (I = very fair, VI = deeply pigmented). We stratify validation cohorts across all six and publish per-subgroup accuracy.
The US Health Insurance Portability and Accountability Act. Governs how PHI is handled. Our on-device pipeline keeps raw video from ever becoming PHI in the first place.
Heart Rate Variability. The beat-to-beat variation in time between heartbeats. A window into autonomic nervous system state. We compute both RMSSD and SDNN from a 30-second scan.
A unique client-supplied string attached to a POST so retrying it doesn't create a duplicate record. Pass Idempotency-Key header; we dedupe for 24 hours.
The international coding system for diagnoses. We return ICD-10 codes when parsing a discharge summary or prescription.
Logical Observation Identifiers Names and Codes. A universal vocabulary for lab and clinical observations. Every metric we return ships with its LOINC code.
The ratio of low-frequency to high-frequency spectral power in an HRV trace. A commonly used, debated proxy for sympathetic / parasympathetic balance.
Mean Absolute Error — the average magnitude of the difference between our reading and a reference sensor. Lower is better. We report MAE for every metric.
Millimetres of mercury. The unit for blood pressure. UCUM code mm[Hg].
Optical Character Recognition. Our medical-device OCR pipeline reads LCD and 7-segment displays on 220+ device models and returns structured fields with per-field confidence.
Running the ML model directly on the phone, mirror, or embedded device rather than a cloud server. The SmartScanPro SDK does this by default — raw video never leaves the device.
Instead of a single confidence score for a whole document, we return a confidence number for each extracted field (e.g. systolic 0.97, diastolic 0.92, pulse 0.88).
Protected Health Information. Any identifiable health data covered by HIPAA. On-device inference in zero-retention mode means no PHI persists server-side.
Plane-Orthogonal-to-Skin (Wang et al., 2017). An rPPG algorithm that projects pixel color onto a plane orthogonal to the skin-tone axis, which improves robustness to motion.
A small clip-on device that shines light through a finger or earlobe and measures blood oxygen saturation (SpO₂) and pulse. Our OCR pipeline reads any 7-segment or LCD oximeter display from a photo.
Breaths per minute. We derive it from a 30-second face scan by measuring subtle respiratory-induced motion and variation in the pulse envelope.
Root Mean Square of Successive Differences. A time-domain HRV metric sensitive to parasympathetic (vagal) activity. Usually 20–50 ms in healthy adults at rest.
Remote Photoplethysmography. Measuring a pulse signal from subtle color changes in the skin of someone's face using a regular camera — no contact, no sensor.
Remote Patient Monitoring. A care model where a patient uses at-home devices and apps; data flows to a clinician who reviews asynchronously and bills against CPT 99453–99458.
Standard Deviation of NN intervals. A time-domain HRV metric reflecting overall variability across a recording. Reported alongside RMSSD for a fuller HRV picture.
Signal-to-Noise Ratio. Drives our confidence score. Poor lighting, motion, or a heavily covered face lowers SNR and widens our returned uncertainty bands.
Peripheral oxygen saturation. The percentage of hemoglobin carrying oxygen. Normal at sea level is 95–100%. We estimate it from rPPG and report it with a ±1.6% MAE.
The provision of healthcare services remotely via video, phone, or messaging. SmartScanPro adds contactless vitals to a telehealth visit without new hardware.
TensorFlow Lite. The on-device model runtime we use on Android (alongside NNAPI and GPU delegates) to keep latency sub-second on mid-range phones.
A clinical measurement of a basic body function: heart rate, respiratory rate, body temperature, blood pressure, and SpO₂. We return all five from a single face scan.
An estimate of arterial "age" from pulse-wave shape. Higher vascular age relative to chronological age correlates with cardiovascular risk factors.
WebAssembly. The binary format our web SDK ships in so that rPPG inference runs at near-native speed directly in the browser, with no server round-trip for raw video.
An HTTP POST we send to a URL you control when an event fires — e.g. a scan completes or an OCR document finishes processing. Secured with an HMAC signature.
A blood-pressure estimate computed from the face scan. Suitable for wellness and screening contexts, not diagnosis. Customers who need clinical BP use our medical-device OCR on a cuff reading.
An account setting that disables all server-side storage of request and response bodies. We process the request, return the result, and drop it. Audit logs retain only metadata.
Email docs@smartscanpro.ai and we'll add it — credited to you if you like. We update this page monthly.