When a Number on a Screen Becomes a Clinical Decision

A vital signs monitor is not a sensor with a display. It is a device whose output causes someone to act: call a nurse, escalate a patient, adjust a drip, or go back to sleep. That is what separates it from a fitness tracker, and it is also what pulls it into FDA jurisdiction. Before any hardware decision, settle whether the product measures for wellness or informs care, because the answer changes the budget by an order of magnitude. The FDA definition test is the right place to start.

Define the Clinical Question Before the Parameter List

Teams tend to start with a list: heart rate, SpO2, respiration rate, temperature, blood pressure. That is backwards. The useful question is what decision the device supports, for which population, in which setting. Continuous SpO2 on a post-surgical ward has a different accuracy requirement, alarm behavior, and duty cycle than spot-check SpO2 in a pharmacy kiosk.

Write the intended use statement first, in one sentence, and let it constrain everything downstream. It determines device class, whether a predicate exists, whether pediatric or neonatal use is claimed (which roughly doubles validation work), and whether the user is a clinician or a patient at home. Home use adds a whole usability layer described in designing for a non-clinical user.

The Signal Chain Behind One Number

Every displayed value is the end of a long chain, and accuracy is lost at every link. A pulse oximeter reading depends on LED wavelength tolerance, photodiode placement, optical isolation between emitter and detector, skin contact pressure, ambient light rejection, perfusion, and the motion artifact rejection in the algorithm. An ECG front end depends on electrode material and impedance, input common-mode rejection, driven-right-leg circuitry, and how well the analog section is isolated from the switching supply.

  • Analog front end. Use a purpose-built AFE rather than discrete op-amps. The integrated parts handle lead-off detection, gain switching, and the impedance requirements that take months to get right discretely.
  • Sampling and resolution. A 24-bit ADC does not make the measurement accurate if the noise floor sits at bit 16. Specify effective number of bits, not marketing bits.
  • Mechanical coupling. Contact pressure, cuff geometry, and probe fit dominate accuracy more often than electronics. A finger clip that shifts 2 mm during motion invalidates the reading.
  • Thermal path. A skin temperature sensor sitting on the same copper pour as a regulator reads the regulator. Isolate it with slots and thermal relief.
  • Patient-contact materials. Anything touching skin needs ISO 10993 biocompatibility testing, and material choices made late force retesting.

Alarms Are Where Products Fail

More monitoring devices stumble on alarm design than on measurement accuracy. IEC 60601-1-8 defines alarm priorities, audible patterns, visual indicators, and the behavior required when an alarm is silenced or the device loses its signal. Reviewers read this section closely.

The engineering problem underneath the standard is alarm fatigue. A monitor that cries wolf on every motion artifact gets muted, and a muted monitor is a device that does nothing. Practical defenses: a delay before annunciation so transients self-clear, a signal quality index that suppresses alarms when the input is untrustworthy, distinct handling of technical alarms (probe off, low battery) versus physiological alarms, and default limits that match published clinical practice rather than something chosen in a meeting. Test alarm behavior against recorded patient data with artifacts, not against a clean simulator.

Connectivity and Data Continuity

Most monitors now push data somewhere. That introduces obligations beyond the radio. Data must survive a dropped link, so the device needs local storage and a store-and-forward buffer with timestamps that stay coherent when the clock resyncs. Decide early whether the device or the gateway holds the record of truth, because that determines what regulators consider part of the device.

Bluetooth Low Energy is the common choice for wearables and home units, with cellular or Wi-Fi at the gateway, and the radio choice drives battery size more than any other decision. Once the device is connected, FDA expects a documented threat model, signed firmware updates, and a software bill of materials, as laid out in the cybersecurity requirements for connected devices.

The Regulatory Path

Most vital signs monitors are Class II and clear through 510(k) against an existing predicate. That is good news for schedule and cost, but only if the predicate genuinely matches the intended use and technology. A novel sensing method with a familiar output often fails the substantial equivalence argument, so run predicate selection before locking the architecture, then work through the 510(k) submission process.

Bench testing is heavier than teams expect. Expect IEC 60601 electrical safety and EMC testing, the relevant particular standard for the parameter measured, software documentation at the appropriate level of concern, and full design controls with traceability from requirement to test. Lab time alone commonly runs $40,000 to $90,000 for a connected multi-parameter monitor, with 10 to 16 weeks of queue and retest.

Proving It Works on People, Not on the Bench

A simulator proves the electronics. It does not prove the claim. Pulse oximeters need a controlled desaturation study on human volunteers, typically 10 to 12 subjects across a range of skin pigmentation, to establish accuracy against arterial blood gas. Blood pressure devices need a clinical accuracy validation against a reference method across a defined distribution of arm sizes and pressures. Budget $60,000 to $150,000 and three to six months for a study of this kind, and design the device so the study does not have to be repeated after a hardware change. The distinction between bench verification and clinical validation is worth understanding cold; see verification vs validation.

Talk Through Your Monitoring Concept

Projects House develops vital signs monitoring devices end to end: sensing architecture, analog front end, firmware, enclosure, alarm strategy, and the documentation package a 510(k) reviewer expects. If you have a parameter you want to measure and a claim you want to make, send the intended use and target setting through our contact form and we will tell you what the path looks like.