Can my doctor get my temperature and breathing numbers from my home, without special equipment?

The question of whether a clinician can read a patient's temperature and breathing at home without shipping a kit of dedicated sensors is no longer hypothetical for telehealth operations and EHR integration teams. It is a procurement and architecture decision. As remote patient monitoring (RPM) programs scale, the cost and logistics of distributing peripheral devices become the limiting factor, not the clinical demand for the data. A growing body of work on camera-based and software-defined sensing suggests that two of the four classic vital signs, respiratory rate and body temperature, can be derived from hardware patients already own. For implementation teams, the relevant work is no longer "how do we mail another device" but "how do we ingest a software-generated observation into the chart with the same trust we give a Bluetooth thermometer."

A 2024 systematic review of contactless respiratory rate measurement using RGB cameras reported a mean absolute error of roughly 0.61 to 0.95 breaths per minute across different skin tones and ages, a range the authors describe as clinically relevant.

Capturing temperature and breathing at home without dedicated hardware

Two distinct technical paths make it possible to measure temperature and breathing at home using consumer hardware. The first is remote photoplethysmography (rPPG), which extracts subtle color and motion changes from ordinary RGB video, the kind produced by any smartphone front camera or webcam. Respiratory rate falls out of this signal because breathing modulates both chest motion and the underlying pulse waveform. The second path is smartphone-based thermography, which uses infrared sensing to estimate body surface temperature.

The accuracy picture differs sharply between the two signals, and that distinction matters for how an integration team should treat each observation downstream. Respiratory rate from standard cameras has reached a credible operating range. Body temperature is more constrained: standard RGB cameras cannot measure it directly, and software estimates from skin appearance or low-cost thermal attachments behave as screening tools rather than calibrated clinical readings.

For health IT directors, the practical framing is a sourcing matrix. Each method carries a different hardware dependency, a different accuracy profile, and a different data-governance footprint once it lands in the EHR.

Method	Hardware needed	Vital signs captured	Reported accuracy	Integration considerations
RGB camera rPPG	Existing phone or webcam	Respiratory rate, heart rate	MAE ~0.61-0.95 breaths/min (2024 review)	Tag observations with method + lighting/motion metadata
Smartphone thermography	Phone with IR sensor or clip-on	Skin/surface temperature	~97% vs reference thermal camera in lab tests	Treat as screening; flag ambient-condition dependency
Bluetooth peripheral (thermometer, etc.)	Dedicated device per patient	Core temperature, SpO2, BP	Device-specific clinical grade	Highest per-patient cost and logistics overhead
Manual patient entry	None	Self-reported temperature, breaths	Variable, no provenance	Lowest trust; needs reconciliation workflow

The operational appeal of the first two rows is obvious. They collapse the device supply chain. But they shift the burden onto the data layer, where provenance, method tagging, and quality scoring decide whether a clinician trusts the number.

Why respiratory rate is the easier win

Respiratory rate has long been the most under-measured vital sign in both inpatient and home settings, partly because no convenient consumer device captures it well. Camera-based methods change that economics. Because the signal can be derived from video a patient is already generating during a telehealth visit, respiratory rate can be captured passively, without asking the patient to do anything beyond sitting in frame.

• The measurement is software-defined, so improvements ship as updates rather than new hardware.
• Real-world performance still depends on lighting, motion, and camera quality, which means metadata about capture conditions should travel with the value.
• A 2023 smartphone rPPG study of the WellFie application reported about 84 percent accuracy for respiratory rate against certified reference devices, with a relative mean absolute percentage error near 15.7 percent, a reminder that home conditions widen error bars.

Why temperature needs a screening label

Surface temperature from a phone is not the same measurement as core temperature from an oral or tympanic device. Smartphone thermography exploits infrared radiation emitted by the body and can produce detailed heat maps, but ambient temperature, distance, and skin emissivity all influence the reading. For integration teams, the safest pattern is to ingest these values explicitly labeled as screening observations, never as a substitute for a calibrated thermometer in a clinical decision rule.

Industry applications for telehealth and integration teams

Telehealth visit enrichment

The most immediate application is passive enrichment of a video encounter. A telehealth platform already streams patient video; layering rPPG analysis lets the same session yield a respiratory rate and heart rate observation without adding a workflow step. For telehealth operations, this turns an unstructured video call into a structured data event that can be coded and documented.

RPM program expansion without device fleets

For RPM coordinators managing CMS-reimbursed programs, the device fleet is the dominant operational cost and the leading cause of patient non-adherence. Software-defined capture of temperature and breathing at home lowers the barrier to enrollment, particularly for patients who struggle with multiple gadgets. The trade-off is that program designers must decide which decisions can rest on screening-grade data and which require a calibrated peripheral.

EHR ingestion and FHIR mapping

Once a value exists, the integration question is identical to any other RPM stream: how does it become a trustworthy row in the chart. Respiratory rate maps cleanly to a FHIR Observation resource with LOINC code 9279-1, and body temperature to 8310-5. The discipline that separates a credible program from a noisy one is the use of the Observation.method and Observation.device elements to record that the value came from a camera algorithm, plus extensions or components carrying confidence and capture conditions. That provenance lets clinical decision support suppress or downweight a reading taken in poor lighting rather than alerting on an artifact.

• Map each derived vital to its standard LOINC code so it aggregates with device-sourced data.
• Populate method and device fields so downstream systems can distinguish camera-derived from peripheral-derived values.
• Carry a quality or confidence score as a component to drive alert filtering.

Current research and evidence

The evidence base has matured quickly. The 2024 systematic review of RGB-camera respiratory measurement, summarizing dozens of studies, placed mean absolute error under one breath per minute in controlled conditions while cautioning that environmental variance and the shortage of diverse real-world datasets remain open problems. On the temperature side, researchers at Georgia Tech have argued that thermal imaging could become a simple, highly accurate route to tracking vital signs, and a 2023 arXiv study, "Temperature Detection from Images Using Smartphones," reported roughly 97 percent agreement with a reference FLIR thermal camera using a blackbody-radiation model.

A 2024 scoping review of smartphone-based thermography in MDPI catalogued its appeal for affordability and operational simplicity while flagging legal and implementation challenges, including the absence of standardized calibration and the risk of treating screening output as diagnostic. The consistent theme across this literature is that contactless capture is viable for trend monitoring and triage but demands explicit handling of provenance and uncertainty before it feeds clinical workflows. Interest accelerated during the COVID-19 pandemic, when contactless monitoring moved from research novelty to deployed screening at scale.

The future of temperature and breathing at home

The trajectory points toward software-defined vitals becoming a default layer in telehealth rather than a specialty add-on. As model accuracy improves and capture conditions get auto-detected, the friction of distributing peripherals will look increasingly avoidable for low-acuity and screening use cases. The decisive work for the next several years is not in the algorithms but in standardization: agreed methods for tagging confidence, shared FHIR profiles that encode capture conditions, and reconciliation pipelines that let camera-derived and device-derived readings coexist in one chart without confusing clinicians. Teams that build that data governance now will absorb each accuracy improvement as a configuration change rather than a re-implementation.

Frequently asked questions

Can a doctor really get my temperature and breathing numbers without a special device?

Breathing rate can be estimated from ordinary phone or webcam video using remote photoplethysmography, with reported errors under about one breath per minute in studies. Temperature is harder: standard cameras cannot measure it directly, and phone-based thermal estimates work as screening rather than as a substitute for a calibrated thermometer.

How accurate is camera-based respiratory rate compared to a clinical monitor?

A 2024 systematic review reported a mean absolute error of roughly 0.61 to 0.95 breaths per minute in controlled conditions. Real-world accuracy is lower because lighting, motion, and camera quality affect the signal, which is why capture metadata should be stored alongside each reading.

How does this data get into the EHR?

Each derived vital maps to a standard FHIR Observation with a LOINC code, the same as device-sourced values. The key difference is populating the method and device fields to mark the value as camera-derived and attaching a confidence score so clinical decision support can weight it appropriately.

Should screening-grade temperature drive clinical alerts?

Not on its own. Smartphone thermography is best treated as a screening signal that prompts confirmation with a calibrated device. Encoding it as a screening observation prevents downstream rules from acting on ambient-affected readings as if they were clinical-grade.

Circadify is building toward this exact problem: capturing temperature and breathing at home from consumer hardware and delivering it as HL7 FHIR vital-signs data that drops into existing EHR and telehealth workflows with full method provenance. Teams evaluating how to ingest camera-derived observations without rebuilding their stack can review the integration documentation and EHR guides at circadify.com/solutions/telehealth.