Researchers led by Soon Sil Ko and Jin Young Park of the Center for Digital Health at Yongin Severance Hospital, part of Yonsei University College of Medicine, retrospectively analyzed nine confirmed COVID-19 cases among hospital staff and compared three tracing approaches: closed-circuit television alone, closed-circuit television combined with a real-time location system, and the location system on its own. The results were striking. Tracing that once took more than 12 hours of labor to identify 381 contacts was reduced to roughly 10 minutes for 918 contacts, a transformation the authors measured as a 99.8 percent reduction in the person-minutes required per identified contact.

The technology at the heart of the study is a Bluetooth Low Energy-based real-time location system, or RTLS. Unlike classic Bluetooth, which was designed for streaming audio and file transfers, Bluetooth Low Energy transmits small data packets in short bursts, allowing battery-powered tags to run for months or years on a single coin cell. In the hospital deployment, staff and admitted patients carried or wore small tags that periodically broadcast unique identifiers. Receivers distributed throughout the facility logged which tags passed within range of which access points, along with timestamps and signal-strength information. From those logs, the system could reconstruct who had been near whom, where, and when, without anyone needing to remember their own movements days later.

Signal strength is the crucial physical variable in such systems. Radio waves attenuate as they travel through air and are absorbed by walls, equipment, and human bodies, so the received signal strength indicator at a receiver provides a rough proxy for distance. By combining proximity thresholds with time windows, the RTLS could flag encounters likely to constitute meaningful exposure under epidemiological criteria, such as being within two meters of an infected person for more than a defined duration. This is precisely the kind of judgment that makes manual tracing so laborious: a contact tracer reviewing camera footage must visually estimate distance, track overlapping presence in a corridor or break room, and decide whether a fleeting pass-by counts as an exposure.

The implementation did not happen overnight. The researchers documented a phased rollout beginning in 2019, moving through initial deployment, institutional transition, and system enhancement. Early phases involved technical refinements to the tag network, expanded functionality, and, critically, work on compliance. A location system is only as good as its coverage of the people it is meant to track, and the team steadily improved adherence to BLE tag use among hospital staff until it reached 92.8 percent. Internal hospital documents and operational records were reviewed to identify the key technical and operational changes along the way, from upgrades to the web application server and database infrastructure to improvements in the graphical user interface that infection control staff used to query the system.

The headline numbers deserve unpacking. Before the RTLS was in place, tracing a single case using closed-circuit television alone required 770 minutes of work to yield 381 contacts, which works out to roughly 10.1 person-minutes per contact. After the system matured, the same task performed with RTLS data alone took about 10 minutes to enumerate 918 contacts, or approximately 0.02 person-minutes per contact. In other words, the system did not merely speed up the old process; it changed the economics of the task so dramatically that exhaustive tracing became feasible even when the number of potential contacts ballooned. During a fast-moving outbreak, that difference can determine whether quarantine decisions are made before or after secondary transmission occurs.

The efficiency gain also has a human dimension that the raw figures only hint at. Manual contact tracing in a hospital setting pulls skilled staff, often infection control nurses and epidemiologists, away from patient care for entire shifts. It exposes gaps in memory and gaps in camera coverage alike, and it can leave contacts unidentified until they themselves develop symptoms. A system that produces a contact list in minutes allows quarantine and testing decisions to be implemented the same day, shortens the window during which undetected carriers circulate through wards, and reduces the anxiety of staff who otherwise wait days to learn whether they were exposed. The authors frame this as a substantial improvement in in-hospital infection control capability, with the RTLS supporting rather than replacing clinical judgment.

The study is careful about its limits. It is a single-hospital case study conducted during one phase of one pandemic, and the comparison of CCTV-only, combined, and RTLS-only approaches is retrospective rather than a randomized trial. Accuracy in BLE proximity detection depends on receiver density, tag placement, and the radio environment of a building full of metal beds, imaging equipment, and moving bodies, all of which can scatter signals in ways that produce both false positives and false negatives. The authors also note that wider implementation raises questions of long-term economic sustainability, since maintaining thousands of tags, receivers, and servers carries recurring costs, and of ethics, because continuous location tracking of employees and patients sits uneasily beside privacy expectations even when data are anonymized and consented.

Those ethical questions were addressed explicitly in the study’s governance. Tags were assigned to admitted patients and relevant staff in accordance with institutional procedures, and written informed consent for routine RTLS monitoring included the potential use of anonymized RTLS-derived data for research purposes. The retrospective analysis itself was approved by the Institutional Review Board of Yongin Severance Hospital, which waived the requirement for individual informed consent given the study’s retrospective design, and the work was conducted in accordance with the Declaration of Helsinki. The research was supported by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute, funded by the Ministry of Health and Welfare of the Republic of Korea, with the system’s installation and optimization supported by People and Technology Inc.

The broader significance of the findings extends well beyond one hospital’s pandemic response. Health systems worldwide invested heavily in digital tracing tools during COVID-19, from smartphone exposure-notification apps to wearable badges, with mixed and often debated results. This study offers one of the clearest quantitative demonstrations that, within the controlled environment of a single institution, a BLE-based location infrastructure can collapse the labor cost of contact tracing by three orders of magnitude. The authors point toward future studies evaluating long-term cost-effectiveness and the ethical architecture of wider deployment, questions that will matter as hospitals weigh whether the infrastructure built for pandemic emergencies should become permanent plumbing for infection control, outbreak investigation, and perhaps the everyday operational logistics of knowing where people and equipment are in a modern medical campus.

Subject of Research: Bluetooth Low Energy-based real-time location systems for hospital contact tracing during the COVID-19 pandemic

Article Title: Operational efficiency and practical utility of real-time location systems: a hospital-based case study during the COVID-19 outbreak

Article References: Ko, S. S., Ryu, U. H., Kim, S. J., Lee, J. H., Kim, K. D., Kim, N. Y., & Park, J. Y. (2026). Operational efficiency and practical utility of real-time location systems: a hospital-based case study during the COVID-19 outbreak. BMC Health Services Research. https://doi.org/10.1186/s12913-026-15775-6

Image Credits: AI Generated

DOI: 10.1186/s12913-026-15775-6

Keywords: real-time location systems, Bluetooth Low Energy, contact tracing, COVID-19, infection control, hospital informatics, health services research, Yongin Severance Hospital, digital health, wearable technology, operational efficiency, pandemic response

The team behind TinZr, led by Ludvik Alkhoury of the New Jersey Institute of Technology together with colleagues from Weill Cornell Medicine and collaborating institutions, set out to solve a persistent problem in wearable sensing research. Commercial microcontroller boards tend to force compromises: the Seeed XIAO ESP32-C3 is tiny and wireless but lacks a microSD card reader and a standardized connector for plug-and-play sensors; the Adafruit GEMMA v2 is featherweight but has no charging circuitry or storage; the popular Teensy 4.1 offers generous processing power and an SD card slot but no Wi-Fi or Bluetooth and a bulkier footprint. TinZr was engineered to combine all of these missing pieces in a single module sized for wearables.

At the heart of the platform sits Espressif’s ESP32-C3 microcontroller, a single-core RISC-V processor clocked at up to 160 MHz with native 2.4 GHz Wi-Fi and Bluetooth Low Energy 5.0 built in. The designers acknowledge that the chip offers fewer general-purpose input-output pins than larger development boards, but they argue the pin allocation was deliberately optimized for the interfaces that matter most in physiological sensing: digital, analog, I2C, SPI, and UART buses. An onboard QWIIC connector exposes the I2C bus along with power and ground, letting researchers snap on compatible sensors — optical, environmental, or electrophysiological — without touching a soldering iron.

Motion sensing is handled by an STMicroelectronics LSM6DS3TR-C inertial measurement unit, which pairs a three-axis accelerometer with a three-axis gyroscope in a single low-power package. The accelerometer can be configured for full-scale ranges up to plus or minus 16 g, while the gyroscope measures angular rates up to 2000 degrees per second, with output data rates reaching several kilohertz. In practice, the team envisions this motion data serving double duty: capturing physical activity directly and supporting artifact detection, so that movement contamination in physiological recordings can be identified and flagged during mobile experiments.

Power management, often the Achilles heel of battery-powered wearables, is entrusted to a Texas Instruments BQ24075RGTR charging chip with power-path management. This means the device can run from USB power while simultaneously charging its 150 mAh lithium-polymer battery, and can switch seamlessly to battery operation when unplugged — without interrupting a recording. The charge current is set to roughly 100 mA via an external resistor, matched to the small battery’s 0.67C charge rate, with safety timeouts and input current limits configured to stay comfortably within USB power budgets. The resistors can be swapped to accommodate larger batteries for longer studies.

Battery life figures vary dramatically by operating mode. In deep sleep with everything disabled, TinZr sips just 0.376 mA, enough for an estimated 16.6 days of standby on the standard battery. Active idle draws 14.8 mA, or about ten hours. Streaming inertial or physiological data over Bluetooth pushes consumption into the 41 to 61 mA range, yielding runtimes between roughly 2.5 and 3.7 hours depending on the sensor load — for example, 3.13 hours while streaming dual-wavelength photoplethysmography at 100 Hz, and 2.46 hours when acquiring ECG and PPG simultaneously at the highest tested rates. Wi-Fi transmission is the most power-hungry mode at 93 mA, draining the battery in under two hours. The platform also monitors its own battery voltage, enabling firmware-level low-battery warnings or automatic recording stops.

The validation experiments demonstrate why the platform matters for real research. In a multi-modal demonstration, TinZr simultaneously acquired a single-lead electrocardiogram from a MAX30001-based front-end and red and infrared photoplethysmographic signals from a MAX30102 optical sensor, streaming both over Bluetooth while packet timestamps and sample counters kept the mismatched sampling rates aligned for later analysis. The resulting ten-second recording showed crisp cardiac cycles with clearly distinguishable QRS complexes alongside pulsatile PPG waveforms. Separately, inertial recordings captured repetitive hand motion at metronome-controlled frequencies of 50 and 100 beats per minute, with the oscillation rate on the accelerometer and gyroscope traces tracking the tempo precisely.

Wireless reliability testing yielded striking results. A single TinZr transmitting inertial-style data at 100 Hz over Bluetooth recorded zero packet loss across 180-second trials at every distance tested, from under one foot up to twenty feet, both under direct line-of-sight and with a wooden door blocking the signal path. Signal strength, measured as RSSI, degraded with distance and obstruction as expected, falling to around -89 dBm at the far end, yet the connection held throughout. In a swarm-style test, five TinZr devices streamed simultaneously to a single laptop at 10 Hz each — again with zero packet loss across all conditions — demonstrating that coordinated multi-node deployments, from distributed body sensor networks to collaborative robotics, are practical with the stock hardware.

Perhaps the most consequential design decision is accessibility. TinZr is programmed through the Arduino ecosystem using an accompanying library called TinZrConnect, which ships with example sketches covering everything from basic button and LED control to a full communication framework in which multiple devices connect to a central PC hub over Wi-Fi for broadcast and individual addressing. Companion Python-based graphical interfaces — available both as source code and as installable executables — let researchers control recording start and stop and visualize data without writing firmware. The designers also provide 3D-printable enclosure files that house the board and battery in a wearable unit measuring just 29.5 by 29 by 20 millimeters, printable on an entry-level hobbyist 3D printer in PLA plastic.

The team is careful to note the platform’s limits: TinZr provides no medical-grade electrical isolation, is not certified for clinical use, and users connecting sensors to human subjects are responsible for ensuring the complete measurement chain meets applicable safety standards. Within research settings, however, the board’s combination of local storage, wireless flexibility, standardized sensor interfacing, and a $65 price point positions it as a serious alternative to commercial controllers for prototyping wearables, field data collection, and multi-device experiments. With hardware licensed under the CERN Open Hardware Licence and software under GPLv3, and all production files published on Zenodo with OSHWA certification, the barrier to building a custom multi-sensor wearable has arguably never been lower.

Subject of Research: An open-source ESP32-C3 microcontroller platform for multi-modal wearable physiological sensing and wireless data acquisition

Article Title: TinZr: A compact wireless ESP32-C3 platform for multi-modal physiological sensor integration and data acquisition

Article References: Alkhoury, L., Moore, T., Swissler, P., Hill, N. J., Shah, S. A., & Kam, M. (2026). TinZr: A compact wireless ESP32-C3 platform for multi-modal physiological sensor integration and data acquisition. HardwareX, 28, Article e00833. https://doi.org/10.1016/j.ohx.2026.e00833

Image Credits: AI Generated

DOI: 10.1016/j.ohx.2026.e00833

Keywords: TinZr, ESP32-C3, open-source hardware, wearable sensors, physiological monitoring, ECG, PPG, inertial measurement unit, Bluetooth Low Energy, data acquisition, HardwareX, battery-powered electronics