Determining the performance of a temperature sensor embedded into a mouthguard

Monitoring intra-oral temperature is clinically important for assessing patient status in dentistry and medicine. Sublingual temperature, related to core temperature, can indicate infections, medication reactions, or other disease symptoms. In dentistry, it aids in diagnosing periodontal diseases, tooth erosion, or decay, and can track oral appliance wear time. Accurate temperature readings are essential. Various technologies exist, including electronic sensors and chemical strips. Thermocouples are common due to their wide range, low cost, and quick response; however, resistance temperature detectors and semiconductor-based circuits are also used. Studies on accuracy often focus on reproducibility, pre-encapsulation accuracy, or human subject accuracy against a reference. However, circuit design, sensor technology, placement, temperature range, and encapsulation all affect error. Ideally, tests should explore encapsulated sensor performance across relevant temperatures, considering the time to reach steady-state. This study addresses the lack of "pre-clinical" simulation evaluating both error and time components of wearable temperature monitors embedded in mouthguards. It investigates error against a gold standard and time to equilibrium across three temperatures in a controlled lab experiment, simulating clinically relevant temperatures using a thermostatic fluid bath (standard practice for thermometer calibrations). A human volunteer case study explores the generalizability of lab-bench findings.

Existing literature highlights the clinical significance of intraoral temperature monitoring in various applications. Studies on intraoral temperature sensors frequently focus on individual aspects such as sensor reproducibility, pre-encapsulation accuracy, or accuracy in human subjects compared to a reference thermometer. However, a comprehensive approach considering the combined impact of factors such as sensor technology, encapsulation material, and the oral environment itself is lacking. Previous research on similar devices often does not fully consider the time taken to reach thermal equilibrium, focusing primarily on steady-state accuracy. The current study addresses this gap, aiming to provide a more complete evaluation of embedded temperature sensors in mouthguards, taking into account both accuracy and response time.

Four custom-designed data acquisition systems, each incorporating an ARM Cortex-M4 microprocessor, a Bluetooth 4.2 network processor, flash memory, and a digital temperature sensor (MAX30208, ±0.1 °C accuracy from +30 to +50 °C), were embedded in mouthguards made from ethylene-vinyl acetate (EVA). A maxillary impression was taken and cast in dental stone. A single layer of EVA was thermoformed onto the cast, the sensor board was placed, and a second EVA layer was applied. The sensor was positioned at the buccal side of the upper first molar. In vitro tests, based on BS EN ISO 80601-2-56:2017, used a thermostatic water bath at 34, 38.5, and 43 °C. Readings from the mouthguard sensor and a reference thermometer (RS PRO RS1710 PT1000) were logged every 10 s for 30 min. The water bath temperature was maintained at ±0.1 °C accuracy. In vivo tests used one volunteer wearing a mouthguard. A 3M Tempa DOT Single-Use Clinical Thermometer (±0.1 °C accuracy) provided reference sublingual and buccal temperature measurements. Data were analyzed using Matlab, calculating mean absolute error and root mean squared error. Steady-state was defined as 60 s of continuous data with variation below 0.02 °C. Statistical analysis was performed on data between 20 and 30 min.

In vitro water bath tests showed a median time to reach temperature equilibrium of 380 s (range 130–690 s) across all temperatures. The mean absolute steady-state error was 0.20 ± 0.22 °C against the water bath thermometer and 0.21 ± 0.21 °C against the RS1710 thermometer. The in vivo case study showed a median time to steady-state of 1030 s (950–1110 s). The mean absolute steady-state error compared to the single-use thermometer at the sensor location was 0.23 ± 0.16 °C. Compared to sublingual temperature, the error was higher (0.72 ± 0.11 °C). The study demonstrated that the mouthguard sensor measured temperature with a consistent error across a range of temperatures, suitable for medical applications. However, the time to reach steady-state was significantly longer in the in vivo setting compared to the water bath, suggesting the water bath might not fully represent the oral environment.

The study's comprehensive approach using both in vitro and in vivo testing provides a robust assessment of the temperature sensor's performance. The observed steady-state error is acceptable for clinical applications. The difference in time to reach steady-state between the water bath and in vivo conditions highlights the limitations of using a water bath as a sole testing environment for systems designed for continuous intraoral monitoring. Airflow and temperature heterogeneity in the oral cavity likely contributed to the longer in vivo time. Optimizing sensor placement and encapsulation could potentially reduce this time. The findings raise questions about protocols for testing temperature data loggers used for oral appliance wear-time monitoring, suggesting that a water bath alone might not be sufficient to adequately simulate the oral environment, particularly regarding the time component of the measurement.

This study demonstrated the consistent steady-state accuracy of a temperature sensor embedded in a mouthguard, suitable for clinical use. However, the substantially longer time to reach steady-state temperature in vivo compared to the water bath emphasizes the importance of considering the time component in applications requiring continuous monitoring. Future research should focus on optimizing sensor placement and encapsulation to minimize time to equilibrium and improve the overall system's responsiveness in real-world conditions. A wireless system design could also improve the practical applicability of the technology.

The small sample size (four mouthguards and one volunteer) limits the generalizability of the findings. The accuracy of the reference thermometer was comparable to the mouthguard sensor, limiting detailed error analysis. Ambient factors in the water bath testing could have affected the time to steady-state. The wired system, while providing robust data collection, increased complexity and may not be entirely reflective of a fully wireless system.