As athletic training and the conditions surrounding competitive sports can negatively affect an athlete's menstrual cycle (MC) and as hormonal variations throughout the MC may impact performance and training adaptations (1–4), precise diagnostics of the various phases of the MC are crucial for monitoring cycle health and implementing cycle-oriented and /or cycle-based training periodization approaches, as well as for conducting scientific studies involving women (4–6).

The gold standard for MC diagnostics is serial intravaginal ultrasound examination of the ovaries, a procedure that must be carried out by gynecologists in a clinical setting. This method allows direct observation of follicle development and ovulation (7, 8). In the field of sports science research, the combination of three methods is recommended for the diagnosis of the MC: calendar-based counting method, ovulation prediction using luteinizing hormone (LH) urine sticks and hormone analysis in venous blood (5, 8–11). MC monitoring requires reliable, non-invasive, cost-effective and simple methods that are suitable for everyday use to document ovulation characteristics. While recording the bleeding history is relatively simple, determining ovulation and the integrity of the associated cycle phases is a greater challenge. A frequently used valid and reliable method to detect ovulation is the measurement of the LH rise by application of antibody coated sticks in the morning urine (12). However, urinary LH testing may show variability in duration, intensity, and reliability, particularly in athletes (12–14).

An alternative non-invasive measurement method suitable for detecting ovulation is the basal temperature (BT) measurement (8). BT is defined as the lowest physiological body temperature attained at rest, typically occurring during sleep (15). For practical reasons, temperature is most commonly measured immediately upon waking, at which point it is referred to as the wake-up temperature. It serves as an indirect indicator of ovulation due to the progesterone-induced temperature rise of at least 0.2–0.5 °C during the luteal phase (LP) (8). The detailed metrics of the BT profile along the MC are as follows: During the follicular phase (FP), BT remains in the lower range (36.1–36.7 °C) until approximately one day before ovulation, when it may reach its lowest point (nadir) (12, 16, 17). However, the nadir is not considered a reliable indicator of ovulation (18). During the LP, the ruptured follicle, now referred to as the corpus luteum, begins to synthesize and release progesterone (19). As a result, the body temperature rises and forms a plateau. As the corpus luteum regresses and progesterone levels drop, BT returns to its baseline level within one to two days, coinciding with the onset of menstrual bleeding. This biphasic pattern of BT serves as a reliable retrospective indicator of the occurrence of ovulation (12, 20, 21), even if it does not indicate the exact day of ovulation. The results of BT monitoring and their interpretation can be affected by several factors. These include variations in measurement times, thermometer accuracy and the use of subjective judgment in interpreting questionable temperature fluctuations (8). To use BT for MC diagnostics, sufficient knowledge of the possible confounding factors that influence body temperature measures, such as fever, wake-up time and general measurement errors, is required (9, 18). The greatest challenge here appears to be the intra-individual and inter-individual variation in waking and sleeping times (8). While peripheral body temperature can be measured in the morning using ordinary thermometers (infrared, stick) (22, 23), temperature sensors have recently been developed to overcome the challenges of measurement and to optimize the use of BT in cycle diagnostics. Modern intravaginal temperature sensors can continuously determine the current core body temperature (CBT) value over 24 h, considering diurnal variations in temperature profiles (24). These systems record CBT at short intervals across the entire menstrual cycle and employ algorithm-based analyses to identify the fertile window. Previous work using such vaginal biosensors demonstrated reliable retrospective ovulation detection and prediction of fertile days in the general population (24, 25). However, these studies focused primarily on fertility awareness and did not compare intravaginal CBT with other temperature measurement sites or fixed measurement times, nor did they investigate physically active or athletic women.

The present study therefore extends this literature by directly comparing three different BT measurement methods (external ear canal, sublingual, and rectal methods) and two timings (6 a.m., and the individual wake-up time) against continuous intravaginal CBT in recreational athletes. The goal was to determine which method is most suitable for monitoring menstrual cycle phases in athletic context (recreational). We hypothesized that continuous intravaginal temperature monitoring provides the most accurate and physiologically meaningful temperature profiles, allowing reliable retrospective identification of ovulation and corpus luteum sufficiency. We further assumed that single-point measurement methods, such as sublingual, ear canal or rectal recordings, would show limitations in accuracy and consistency, particularly for women engaging in athletic activity.

This study provides important insights into the application of temperature-based MC diagnostics in sports practice. As hypothesized, 24-h intravaginal CBT monitoring proved sufficiently accurate and reliable for identifying ovulation in recreational athletes during habitual training, whereas other measurement methods and fixed measurement times showed clear limitations. To our knowledge, this is the first study to directly compare multiple temperature measurement sites and standardized measurement times against continuous intravaginal CBT in recreationally trained women, providing a novel perspective on the practical applicability of these methods in a sports context.

A biphasic temperature pattern was automatically detected by the device algorithm in 12 out of 17 cycles. In five cycles, the algorithm did not classify ovulation, not because the temperature sensor failed, but because the algorithm applies strict criteria (e.g., minimum required luteal phase length, suppression of single-day fluctuations). Importantly, in four of these five cases, the biphasic pattern was clearly visible in the raw 24-h CBT data, allowing ovulation to be identified retrospectively. Only one cycle showed no temperature rise at all, indicating an anovulatory cycle (confirmed with missing LH surge).

In this study, the continuous 24-h intravaginal CBT measurement was used as the reference method against which all other temperature measurement methods were compared. The absence of ultrasonography does not affect the validity of the comparison between temperature-based methods, as no absolute ovulation date was required for evaluating agreement between temperature measures (32). Although it is not possible to determine an exact ovulation day using evaluation methods such as the Vollman method and the “three over six” rule, it is at least possible to assess the function of the corpus luteum and thus confirm the occurrence and estimate the time of ovulation (28, 31, 32). The findings of this study therefore underscore the efficacy of nocturnal BT determination as an indirect diagnostic tool for tracking ovulation based on the evidence of a functioning corpus luteum in female athletes.

The rationale for employing body temperature measurement in general cycle diagnostics and specifically for the detection of ovulation, is predicated on the fact that, following ovulation, the corpus luteum produces progesterone in sufficient quantities (32, 33). Progesterone has a thermogenic effect, which Baker et al. attribute to endocrine-induced changes in the thermoregulatory set point. This effect is further explained at the neurophysiological level by Rai et al., who suggest that progesterone acts on the preoptic area of the hypothalamus by inhibiting heat-sensitive and activating cold-sensitive neurons, thereby raising the temperatures at which sweating, cutaneous vasodilation, thermal comfort and chilly feelings begin. As a result, CBT increases (34). In cases of corpus luteum insufficiency, a non-detectable or only marginal rise in progesterone, reflected by a temperature increase of less than the defined threshold of 0.2 °C, would therefore be classified as an anovulatory cycle (31). The temperature rise occurs approximately 1–3 days after the peak LH surge, but with considerable interindividual variation (28, 35). Early studies have even indicated a deviation of up to four days (19). Nevertheless, the BT measurement method advocated by Lui et al. is considered as promising method for continuous, longitudinal monitoring of ovarian function.

The ovulation day was utilized as the diagnostic reference point in this study, based on the expected post-ovulatory body temperature increase ranging from 0.2 °C to 0.5 °C and under the assumption of adequate corpus luteum function (8). Consequently, any measurement method employed for the detection of this rise must demonstrate a variability not exceeding 0.5 °C, which signifies a clinically meaningful threshold.

The variability of temperature measurements depending on the time of recording is attributable to physiological factors. Body temperature exhibits diurnal fluctuations, reaching its nadir during nocturnal sleep (36). The precise timing of this nocturnal nadir varies both intra- and inter-individually and is influenced, among other factors, by sleep patterns (36, 37). The findings of this study demonstrate unequivocally that CBT varies physiologically over a 24-h period and that the timing of the nocturnal minimum differs considerably among athletes. This 24-h temperature pattern underscores the impact of the circadian rhythm and external factors (e.g., wake-up time, physical activity) on body temperature. The continuous intravaginal temperature sensor's capacity to accurately capture the true nocturnal nadir is a significant advantage over conventional methods of temperature monitoring, such as fixed-time recordings (e.g., at 06:00). These methods may fail to detect or accurately record the true minimum temperature.

The utilisation of alternative temperature measurement methods, in conjunction with the variability in measurement timing gave rise to notable deviations from the intravaginal data. None of these methods proved to be reliable. Utilising commercially available clinical thermometers with the Vollman method, only 50% of the biphasic cycles were correctly identified, despite 100% being correctly confirmed as biphasic by the intravaginal methods. Furthermore, estimates of ovulation day deviated by up to 10 days when thermometers were used in comparison to the continuous intravaginal temperature sensor. This discrepancy can be partly attributed to measurement variability exceeding the threshold of 0.5 °C, which is defined as clinically meaningful. The limited measurement accuracy observed in this study is consistent with the findings of previous research on the use of body temperature for the purpose of ovulation and fertility diagnostics. It has been demonstrated by earlier studies that the accuracy of temperature-based methods is influenced by several factors. This includes the type of thermometer used, the measurement location and the time of measurement. In addition, external variables such as sleep duration, alcohol consumption, exposure to heat and cold and fever must be taken into account (12, 18, 35).

A number of studies have previously evaluated the accuracy of thermometers (23). These studies identified significant differences between devices and concluded that measurement data from the Braun infrared thermometer (IRT6520), which was used for temperature assessment in the external auditory canal in the present study, showed the best correlation with a gold standard (the SureTemp Plus medical thermometer from Welch-Allyn). However, in our study on indirect ovulation diagnostics using external ear canal temperature, we were unable to confirm a high level of accuracy in practical application. Theoretically, the configuration and the algorithmic processing of the infrared thermometer should guarantee that the temperature displayed is reflective of the tympanic membrane, as opposed to any other component of the ear canal. This is typically achieved by selecting the highest value from a series of measurements captured at a wide angle around the tip of the sensor (38). However, if the probe makes contact with the canal wall rather than aligning with the tympanic membrane, which is plausible due to anatomical variability, or if the measurement is affected by small hairs in the canal, reproducibility is compromised and the recorded temperature tends to be too low (39). In the present study, the Braun rod thermometer (PRT2000) was found to yield smaller average deviations from the intravaginal reference method when employed for sublingual measurements. It is widely accepted that sublingual measurements are expected to indicate lower temperatures compared to intravaginal measurements owing to the greater distance from the body's core (23). However, the sublingual temperature measured at 6 a.m. and immediately after awakening was, on average, virtually identical to the intravaginal temperature at the same time. This phenomenon is likely attributable to the absence of frequent mouth openings and the observation that, while remaining in bed, peripheral body temperature does not differ substantially from CBT or from intravaginal temperature, which is measured in close proximity to the core.

Continuous intravaginal CBT measurement is widely regarded as one of the most accurate non-invasive approaches for assessing core body temperature, yet several factors can introduce measurement variability. Because the sensor relies on indirect heat transfer from the vaginal epithelium, the recorded temperature may be influenced by local blood flow, tissue contact and thermal conductivity, and small positional shifts of the ring can lead to slight fluctuations in absolute values (25, 40). Although the OvulaRing has demonstrated high technical accuracy under controlled conditions, real-world factors such as physical activity, intermittent removal or changes in vaginal moisture may affect single-point measurements (OVU TD DOC 61403-1 report on measurement accuracy, Alexander, 2022). In addition, ovulation detection depends on the device's internal algorithm. Missed detections typically result from strict classification criteria rather than from shortcomings of the underlying CBT signal, which can still display a clear biphasic pattern (25, 41, 42). Overall, while intravaginal CBT is not a perfect measure of absolute temperature or ovulation timing, its continuous sampling and proximity to the body core make it one of the most robust non-invasive methods currently available for capturing the characteristic biphasic temperature shift.

In our study, we used LH measurements in urine for ovulation detection. In general, ovulation occurs 14–26 h after a rise in LH. The LH surge can vary significantly in both duration and intensity (14). While some women exhibit a rapid surge within a single day, others experience a gradual rise over several days (12). Moreover, an LH surge may be detected even in the absence of actual ovulation (13). Such “false-positive” results are particularly relevant in athletes, who may experience anovulatory cycles due to intense training or low energy availability (13). This variability complicates the accurate determination of the ovulation timing based on urinary LH testing. In our study, the time interval between the urinary LH surge and the subsequent rise in BT was 1.3 ± 3.3 days (max. 5 days; min. 0 days), which falls within the expected wide range of variation (12). The high standard deviation of this time interval is at least partly due to the variability of the configuration, amplitude and duration of the individual LH peaks (14). Therefore, the sole use of urinary LH measurements for ovulation detection seems questionable.

This study demonstrates that commercially available thermometers used at defined peripheral body sites are not sufficiently reliable for identifying biphasic temperature patterns or accurately determining ovulation timing in female athletes. In contrast, continuous intravaginal monitoring of core body temperature proved to be the most accurate and robust method among those evaluated, providing reliable retrospective confirmation of ovulation and luteal function. As a more practical and cost-effective alternative, sublingual temperature measurement at a fixed time point may be considered, though with acknowledged limitations.

These findings underscore the value of accurate temperature-based menstrual cycle diagnostics in sports practice. Reliable identification of menstrual cycle phases can support a more individualized approach to menstrual health monitoring, enabling athletes and coaches to better adapt training and competition schedules to physiological changes across the cycle. Ultimately, this approach may contribute to optimizing both performance and long-term health in female athletes.