Evaluation of the accuracy of wrist skin temperature measured using an infrared temperature sensor for prediction ovulation date (Preprint)

Abstract

BACKGROUND Technological advancements have made it possible to measure vital signs through wearable devices. OBJECTIVE This study aimed to evaluate accuracy of wrist skin temperature measurement using an infrared temperature sensor in a wearable device for predicting ovulation dates. METHODS We recruited 106 women and measured data for a total of 318 cycles. All participants were required to measure their wrist skin temperature while sleeping, their oral temperature every day immediately after waking up, and perform an ovulation test several days before and after the expected ovulation date. We aimed to compare the accuracy of the expected ovulation date and actual ovulation date through wrist skin temperature measurement, accuracy comparison with the next menstrual date, and the correlation between oral temperature and wrist skin temperature. RESULTS In the full analysis set (FAS), data of 225 cycles from 85 women were analyzed after excluding dropouts (21 of 106 women). In the per-protocol set, data of 52 women (156 cycles) were analyzed. When analyzing 225 cycles in FAS and comparing the actual ovulation date with the ovulation date predicted by wrist skin temperature, 64.0% had a difference of less than 2 days (95% CI, 57.7-70.3%). A comparison of differences within 3 days showed a significant value, with an accuracy of 77.8% (95% CI, 72.4-83.2%, P=.005). When analyzing 156 cycles in the PP set, the accuracy of the difference within 2 days was 64.1% (95% CI, 56.6-71.6%) while comparing the actual ovulation date with the ovulation date predicted by the wrist skin temperature. When comparing the differences within 3 days, the accuracy was 78.9% (95% CI, 72.4-85.3%, P=.008), showing significant prediction similar to FAS. The prediction of the next menstrual cycle was also analyzed. In FAS, the difference within 2 days was 68.2% (95% CI, 62.5-74.0%, P=.73), and the difference within 3 days was 78.8% (95% CI, 73.8-83.8%, P=.001). According to per-protocol analysis, 69.9% of cases showed a difference within 2 days (95% CI, 62.7-77.1%, P=.51), and 81.4% of cases showed a difference within 3 days (95% CI, 75.3-87.5%, P=.001). When evaluating the correlation between oral temperature and wrist skin temperature, the FAS showed r = 0.423 (95% CI, 0.403-0.443), and the per-protocol showed r = 0.448 (95% CI, 0.423-0.473). CONCLUSIONS Wrist skin temperature measured using an infrared temperature sensor did not show statistically significant results in predicting the actual ovulation date within 2 days, but showed significant results within 3 days, which is the original approval. Therefore, measuring wrist skin temperature during sleep using a wearable device may be helpful in predicting ovulation date and next expected menstruation date and may be another useful aspect of using wearable devices. CLINICALTRIAL not applicable