Biometric sensors for medical and IoT devices: design and certification

A biometric sensor is an integrated MEMS or optical component that converts a physiological signal into firmware-usable digital data via a high-resolution ADC. Biometric sensors measure heart rate (PPG, MAX30101), the Electrocardiogram (ECG, AD8232 single-lead, ADS1292R 24-bit AFE), SpO2 (around 2 percent accuracy), skin temperature (MLX90614, 0.5 deg C), fingerprint (FPC1020, 508 DPI, FAR below 1e-6), or the EEG, EMG and GSR signals. According to STMicroelectronics and Bosch Sensortec, MEMS biometrics are the fastest-growing sensor segment for the 2025 to 2028 window.

At AESTECHNO in Montpellier, we design the analog front end, the low-power firmware (characterized with the Nordic PPK2) and the conformity dossier against IEC 60601-1, IEC 62304 and ISO 13485 for medical wearables. As Analog Devices and Maxim Integrated underline, the precision of the analog front end (24-bit ADC, CMRR above 110 dB) is what conditions the clinical acceptability of the captured signal (references: st.com, bosch-sensortec.com, analog.com, maximintegrated.com).

Contents

• Concrete challenges in health wearable design
• Comparison of biometric sensor technologies
• PPG vs ECG vs BIA: which technology for which measurement?
• Product-oriented project methodology
• Our technical solutions
• Example wearable projects
• Security and product differentiation
• Bottom line
• FAQ: biometric sensors

The health wearable market is expanding fast: connected watches, smart patches, tracking rings, instrumented textiles. These devices open the way to more personalized, continuous and preventive medical follow-up. The technical challenges are many: sensor reliability, miniaturization, energy budget (3 to 14 days depending on which sensors are active), GDPR article 9 handling, wireless link choice, ergonomics and medical standards.

At AESTECHNO, we help companies turn an idea or a prototype into a production-ready electronic product. Our skill set combines embedded systems, low-power electronics and biometric sensors, from the R&D-stage start-up to the established player improving an existing product. See our design house methodology for the upstream picture.

We have, in particular, designed a medical-grade personalized lighting system, a project that confronted us with the strictest regulatory requirements of the field (galvanic isolation per IEC 60601-1 with a withstand voltage of at least 1500 V, patient leakage currents kept under 10 microamps in normal conditions, component traceability per ISO 13485). That experience feeds our approach to health wearables: safety and certification constraints are anticipated from the feasibility phase, not patched at the end. The same risks-first discipline shows up in our product specification guide.

Concrete challenges we help you solve

Designing a wearable health device is a balancing act that demands measurement accuracy, mechanical robustness, battery life and regulatory compliance in a miniaturized, body-worn form factor. Every biometric wearable has to combine clinical-grade signal quality, comfortable industrial design, multi-day autonomy and a defensible regulatory pathway.

Here are the recurring problems we see at customers, with the concrete solutions we put in place:

• Accuracy and reliability of the biometric measurement (heart rate, temperature, SpO2, GSR) under real conditions (sweat, motion, ambient light). In our practice the anatomical sensor position influences signal quality more than the chosen part itself.
• Battery life: tight power management, integrated charging, firmware-level optimization.
• Sensor selection matched to the use case (medical-grade, sport, wellness, research).
• PCB miniaturization and mechanical integration into worn objects (bracelets, rings, patches).
• Data security and BLE / Wi-Fi / LoRa communication to mobile, cloud or medical gateway.
• Standards compliance (IEC 60601, IEC 62304, medical CE, EMC, IP). Medical standards have become a regulatory marathon and a real entry barrier; certifications have grown harder to clear since the 2018 revision.
• Data flood: sensors, IoT, images and documents all generate a continuous stream that drives analysis costs up. Reducing volume without reducing information is now a key success factor.

Field cases observed in our lab

Here are three situations we have run into in the lab on biometric wearable projects. They illustrate the gap between datasheet theory and the real signal measured on a human body.

• Case 1: PPG placement on a bony wrist. An SpO2 sensor positioned over the radial bone produced a PPG signal saturated with motion artefacts. Contrary to the intuition that calls for firm flat contact, we measured that shifting the sensor towards the soft pulpy zone between the extensor tendons clearly improved the signal-to-noise ratio. We recommend always validating the anatomical placement with a measurement campaign before freezing the mechanical design.
• Case 2: fall detection with the IIS3DWB. On a fall-detection project, the STMicroelectronics IIS3DWB accelerometer let us combine wide bandwidth (short shock signatures) with very low power draw (multi-month battery life). Contrary to a classical IMU where you have to trade bandwidth against current, this part removes the compromise. We recommend it whenever a vibration or impact signature must be captured without sacrificing battery life.
• Case 3: medical lighting and patient safety constraints. Our experience designing a medical-grade personalized lighting system taught us that patient-safety constraints (galvanic isolation, leakage currents, component traceability) impose schematic-level decisions, not late-stage corrections. That medical discipline now informs our wearable designs even when the device is not formally certified medical.

Medical standards and validation tools

A health wearable destined for medical or paramedical use sits inside a dense web of standards. As the ISO highlights, and according to NIST, component traceability and risk management are the two pillars of an admissible medical dossier. According to the FIDO Alliance, biometric authentication wearables now have to target FIDO2 and NIST SP 800-63 AAL3 to remove patient-safety doubts (references: iso.org, nist.gov, fidoalliance.org). We work systematically against the following references:

• IEC 60601-1-2: electromagnetic compatibility for medical devices (stricter than the consumer EN 55032).
• ISO 10993: biocompatibility of materials in skin contact.
• IEC 62304: software life cycle for medical device software (Class A / B / C depending on risk).
• ISO 14971: risk management applied to medical devices.
• ISO/IEC 19794: data interchange formats for biometrics (fingerprints, iris, face).
• ISO/IEC 30107: presentation attack detection (anti-spoofing).
• MDR 2017/745: European Medical Device Regulation.
• NIST SP 800-63: digital identity authentication assurance levels (AAL1 to AAL3).

On the instrumentation side, we use the Nordic PPK2 (Power Profiler Kit II) to characterize the current draw in active and sleep modes and to verify the announced battery life. Combined with our long-run logging benches, this tool detects the leakages (forgotten pull-ups, an I2C bus held low) that quietly kill the battery life of a wearable in production. In our lab the PPK2 is paired with a Tektronix Keithley DMM7510 7.5-digit multimeter to capture quiescent currents down to the picoampere range, where the profiler reaches its precision limit. For high-speed serial links on the same board, our Tektronix oscilloscope runs the TekExpress compliance suite (PCI Express, USB 3.x, MIPI, DDR4, HDMI, Ethernet, LVDS), so we pre-qualify those interfaces in-house before sending the design to the accredited lab. According to our project log, this in-house pre-compliance has shortened DVT iterations by roughly one full pass on the 65 projects we have delivered since 2022.

On the application security side, we instantiate a hardware Root of Trust (RoT) coupled to a Trusted Execution Environment (TEE) to isolate biometric templates from application firmware. According to Infineon (OPTIGA Trust M), NXP (EdgeLock SE050) and Texas Instruments (SimpleLink SIP), a dedicated secure element is the industry reference for biometric-template protection. According to the FIDO Alliance, FIDO2-certified biometric authentication requires a False Acceptance Rate below 0.01 percent and a False Rejection Rate below 3 percent under nominal conditions (references: infineon.com, nxp.com, ti.com, fidoalliance.org).

Contrary to consumer wearables where battery-life marketing dominates, medical biometric sensors require a design where measurement accuracy and component traceability come before battery life. In our practice, a 3-day medical patch with a perfectly traced and calibrated chain has more clinical value than a 7-day wearable whose measurement chain drifts. See our testing and validation playbook for how we instrument that calibration step.

Comparison of biometric sensor technologies

Choosing a biometric sensor is a decision that constrains the electronic architecture, the power budget and the clinical performance of the wearable. The table below summarizes the main technologies used in health wearables, with their typical current draw and their preferred application domains.

Sensor	Measurement	Technology	Power	Applications
PPG (photoplethysmography)	SpO2, heart rate	LED + photodiode	~1 mW (during measurement)	Smart watches, oximeters, fitness bands
ECG (electrocardiogram)	Cardiac electrical activity	Electrodes + AFE	~0.5 mW	Medical patches, cardiac monitoring
Skin temperature	Body temperature	NTC thermistor / IR	~0.1 mW	Wearables, connected thermometers
IMU (accelerometer + gyroscope)	Motion, fall, activity	Capacitive MEMS	~0.5 mW	Fall detection, activity tracking, rehab
BIA (bioelectrical impedance)	Body composition	Current injection + Z measurement	~2 mW	Connected scales, hydration tracking

PPG vs ECG vs BIA: which technology for which measurement?

Choosing between PPG (photoplethysmography), ECG (electrocardiography) and BIA (bioimpedance) is what conditions clinical accuracy, power budget and the physical form factor of a wearable. Each technology answers a different physiological question, and none replaces the others. We routinely combine two of them when the use case demands it, for example PPG plus ECG on a chest patch to confirm an arrhythmia detection.

• PPG (MAX30101, MAX30102): optical heart-rate and SpO2 measurement via red (660 nm) and IR (940 nm) LED absorption. Consumer SpO2 accuracy plus or minus 2 percent, HR plus or minus 5 to 10 bpm depending on motion. Power around 1 mW during measurement, sensitive to anatomical placement and skin pigmentation. Preferred choice for watches and fitness bands where footprint dominates.
• ECG (AD8232 single channel, ADS1292R two channels at 24 bits): direct electrical measurement of cardiac activity via Ag/AgCl electrodes. Useful bandwidth 0.05 to 150 Hz, CMRR above 80 dB, reliable arrhythmia detection. Power around 0.5 mW. Contrary to PPG, ECG is largely insensitive to motion artefacts but demands stable skin contact, which is why it dominates on medical patches (3 to 7 days of battery life).
• BIA (current injection at around 50 kHz): body-impedance measurement to estimate composition (water, fat mass, lean mass). Pulsed power around 2 mW. Clinical accuracy plus or minus 3 to 5 percent under standardized conditions. Use cases: connected scales, hydration tracking.

Capacitive vs optical fingerprint: for wearables with biometric authentication, the FPC1020 capacitive sensor (resolution 508 DPI, FAR below 1e-6) offers stronger immunity to photographic spoofing than optical sensors, with lower power draw (10 to 50 mA during read). Rather than defaulting to a legacy optical sensor, we recommend a capacitive or silicon sensor for new secure designs.

The choice has to factor in the regulatory class targeted: a device measuring SpO2 for diagnostic purposes falls under MDR 2017/745 (Class IIa or IIb) and triggers conformity with IEC 60601-1 (electrical safety) and IEC 62304 (medical software life cycle, Class A / B / C). For US market access, the FDA 510(k) or PMA pathway applies depending on the risk class.

An agile, product-oriented project methodology

Our methodology for a health wearable combines technical agility with the regulatory rigor required for medical devices. An iterative approach centered on the clinical or wellness use case lets each step be validated before committing to certification and industrialization spend.

Our methodology is centered on your real use case. After a technical analysis phase, we propose component and architecture choices that fit the use case, then build functional mock-ups, prototypes, and accompany the transfer to production. We work with your team, your industrial designers and your medical partners to keep the electronics, the use experience and the product strategy fully aligned.

Our objective is the customer's objective: a beautiful, functional, ergonomic and durable product. At AESTECHNO we have observed that the most successful wearable projects are the ones where the hardware, firmware and industrial design teams work in parallel from the first weeks of the project. The same parallel methodology is documented in our design house playbook.

Our technical and methodological solutions

Integrating biometric sensors into a wearable demands mastery of analog electronics, low-power firmware and wireless protocols. Each technical brick has to be optimized in isolation while still fitting into a coherent, manufacturable system architecture.

At AESTECHNO, we do not stop at identifying problems: we build reliable, concrete solutions that fit your industrial constraints. Our approach rests on tested technical choices, deep component knowledge and the ability to prototype quickly. Here is how we address the sector's main pain points:

• Measurement-chain optimization: fine sensor calibration, software filtering, motion-artefact rejection in real conditions.
• Low-power electronic design: matched microcontrollers, deep sleep modes, smart power-rail switching, supercapacitors or solar harvesting where it makes sense.
• Reliable, secure communication: BLE, LoRa or Wi-Fi modules with sturdy protocols, latency control, encryption and secure over-the-air updates (OTA). For deeper coverage see IoT cybersecurity.
• Local data processing: simple on-device algorithms (threshold detection, moving averages, lightweight classification, embedded ML) to cut transmitted volume and extend battery life.
• Field tests and validation: in-house test benches, field measurement, log capture, fast iteration. For motion and activity detection, we integrate the STMicroelectronics low-power IIS3DWB accelerometer; its wide bandwidth lets us cleanly separate walking, running, sleep and fall signatures. The same instrumentation pass is described in our testing and validation guide.
• Battery and BMS for medical wearables: our portfolio covers a wide range of cells and BMS designs, from coin cells for disposable patches to rechargeable lithium-ion packs with multi-level protection. On a body-worn medical device, the BMS must guarantee patient safety (short-circuit, thermal, overvoltage protection) while preserving multi-day battery life in a constrained footprint.
• Industrialization readiness: compact yet repairable routing, components that are durable and available, support on CE certification, EMC tests and IP-sealing recommendations.

We always work as close as possible to the end-user experience. That means adjusting the design not just to technical performance, but also to ergonomics, maintenance and the operating environment (indoor or outdoor, hot or cold, mobile or stationary). The user experience is a key success factor.

Example wearable projects you could build

Health wearables span a very wide application range, from consumer wellness to certified medical devices. Each project type carries its own constraints in terms of sensors, battery life, form factor and the regulatory clearance required to bring the device to market.

• A connected patch for skin tracking of patients on medication.
• A miniaturized cardiac tracker for clinical trials.
• A posture sensor embedded in a t-shirt or back belt to correct posture (accelerometer plus gyroscope).
• Connected socks for gait tracking, fall detection or plantar pressure (pressure-ulcer prevention).
• A fall-detection clip for elderly users, worn on clothing (accelerometer plus on-device ML).
• A sensor embedded in a hard hat for shock or syncope detection on a worksite.

A technical partner at your side

Choosing a design house for a health wearable project is a structural decision that shapes product quality, time-to-market and regulatory compliance. A partner who masters the full chain, from electronic design to certification, reduces risk and shortens the path from prototype to series production.

AESTECHNO is a French design house specialized in embedded electronic systems. We understand your stakes: time-to-market, reliability, cost, standards. We act at every stage:

• Feasibility study.
• Electronic design and firmware.
• Prototyping and tests.
• Certification support (EMC) and DFM industrialization.
• Manufacturing with our partner factories.
• RMA flow.

In our practice we have observed that anticipating certification constraints from the feasibility phase significantly reduces iterations and shortens lead times during product qualification. That observation is also the backbone of our specification guide.

Biometric sensors: security and product differentiation

Biometric sensors are a strategic lever that lets companies differentiate on the connected-health market. Mastering the full technical chain, from sensor through firmware to cloud and certification, builds a real entry barrier that protects your competitive advantage.

For decision-makers, biometric sensors represent both an opportunity for differentiation and a regulatory challenge. A reliable, certified health wearable creates a barrier to entry against competitors. At AESTECHNO we have observed that mastering the full chain, from MEMS sensor to embedded firmware, is essential to keep measurement accuracy under real conditions. Biometric data security (GDPR, encryption) has become a buying criterion for end users. Anticipating EMC certification from the design phase avoids costly delays at market entry. We support you in turning your concept into a manufacturable, compliant product. For the cybersecurity layer alone see industrial IoT cybersecurity.

Bottom line

A reliable biometric wearable is built from a small set of measured, documented decisions taken before the schematic is frozen. The five bullets below summarize what we apply on every project at AESTECHNO Montpellier.

At AESTECHNO, this discipline applies to every wearable project, whether it targets medical certification or stays in the wellness range. Contrary to the assumption that a non-medical product can take shortcuts, we have observed that medical methods (BOM traceability, ISO 14971 risk management, IEC 60601-1-2 EMC tests) systematically yield more reliable products that industrialize faster.

FAQ: biometric sensors and connected health

This FAQ answers the questions we hear most often from wearable project leads. The answers reflect what we have measured on the bench, not just what the datasheets claim.

What are the main biometric sensors used in health wearables?
PPG (photoplethysmography): heart rate, SpO2, LED plus photodiode, low cost. ECG (electrocardiogram): precise cardiac rhythm, arrhythmia detection, skin electrodes. Body temperature: thermistor or IR, fever and ovulation tracking. Accelerometer or gyroscope: physical activity, falls, sleep. Bioelectrical impedance (BIA): body composition. GSR (galvanic skin response): stress markers. Each sensor calls for a specific analog chain (amplification, filtering, ADC).

How do you guarantee PPG accuracy for SpO2 and heart-rate measurement?
Error sources: user motion, skin pigmentation, ambient conditions (light, temperature), sensor placement. Solutions: adaptive filtering (motion-artefact rejection), multi-wavelength calibration (red plus IR LEDs), AI algorithms compensating darker skin, mechanical design that holds constant skin contact. Clinical validation is mandatory for medical devices (comparison vs medical-grade oximeter, ECG). Typical consumer accuracy plus or minus 5 to 10 bpm, medical plus or minus 2 bpm.

Which certifications apply to medical devices with biometric sensors?
EU: MDR 2017/745, Class I (low risk, non-critical measurement) up to Class IIb / III (vital diagnostic). USA: FDA 510(k) or PMA depending on risk. Requirements: clinical trials, ISO 14971 risk management, cybersecurity if connected, IEC 60601 medical EMC. For consumer wearables the generic CE marking is enough. The boundary is fuzzy: an SpO2 oximeter is medical, a fitness tracker is consumer; consult a notified body.

How do you handle GDPR compliance for biometric data?
Biometric data are a special category under GDPR (article 9), with reinforced protection. Obligations: explicit user consent, end-to-end encryption of sensitive data, anonymization or pseudonymization, bounded retention period, right to erasure, DPIA (data protection impact assessment) if risk is high. Recommended architectures: local on-device processing (edge computing), minimal cloud transmission, AES-256 encryption, EU servers certified ISO 27001.

What battery life can I expect for a wearable with continuous biometric sensors?
Depends heavily on measurement frequency and active sensors. Continuous PPG (1 Hz): 3 to 7 days typically with a 200 to 300 mAh battery. ECG plus PPG plus accelerometer: 2 to 4 days. Optimizations: periodic vs continuous measurement (for example, SpO2 every 15 min instead of continuous), ultra low-power MCU (Nordic nRF52, STM32L4), edge algorithms reducing Bluetooth traffic, e-paper screen vs OLED. High-end medical smart watches reach 7 to 14 days.

How do you secure biometric templates on a wearable?
Hardware Root of Trust coupled with a Trusted Execution Environment isolates biometric templates from application firmware. Secure elements such as Infineon OPTIGA Trust M, NXP EdgeLock SE050 or Texas Instruments SimpleLink SIP store the keys and run cryptographic operations. ISO/IEC 30107 covers presentation-attack detection. FIDO2 certification requires FAR below 0.01 percent and FRR below 3 percent under nominal conditions, aligned with NIST SP 800-63 AAL3.

Related articles

To go further on connected health and wearable devices:

• Electronic design house methodology — the AESTECHNO 6-step EVT/DVT/PVT framework, from spec to pre-series.
• Electronic product specification guide — how to write an SRS that survives the schematic and EMC phases.
• Industrial IoT cybersecurity — data protection and secure firmware for connected medical devices.
• Electronic product testing and validation — PPK2, Keithley DMM7510 and HALT/HASS practices for low-power wearables.
• AESTECHNO blog — the full library of technical posts on embedded electronics and certification.