Inoue Shota, Dang Van Trong, Liu Hailong, Wada Takahiro
Division of Information Science, Nara Institute of Science and Technology, 8916-5, Takayama-cho, Ikoma, Nara, 630-0192, Japan.
Exp Brain Res. 2025 Apr 5;243(5):108. doi: 10.1007/s00221-025-07052-5.
Computational models predicting motion sickness have advanced, particularly those based on subjective vertical conflict (SVC) theory. While SVC-based models primarily predict motion sickness incidence (MSI), which is defined as the percentage of people who would vomit under a given motion, models predicting milder individual symptoms, which are crucial for daily applications, are still required. Recently, computational models predicting vestibular motion-sickness progression using the SVC theory have been developed by changing the output of a 6DOF-SVC model from MSI to the Misery Scale (MISC), a subjective measure of symptom progression. In practical applications, the ability to predict MISC for unseen motions is crucial. The present study conceived a method for predicting MISC beyond a certain point in the future by identifying parameters from data collected up to that point. Therefore, this study investigates the effect of the number of data points used for parameter identification on the future prediction accuracy. Observed MISC responses from participants exposed to linear lateral motion in darkness were used for model validation. The results indicated that prediction accuracy increased as more data points were included. On average, using more than 5-10 min of data significantly increased the accuracy compared to a model using averaged parameter sets across participants, although the tendency significantly differed based on an individual's MISC history. A trial considering individual MISC histories, in which data points were defined when the observed MISC first reached certain levels, showed a general trend of improved accuracy when data up to MISC Level 3 was used. The findings of this study demonstrate that motion sickness symptom progression can be predicted with reduced error by incorporating individual symptom histories, thereby providing a foundation for the development of personalized motion sickness prediction models applicable to broader applications.
预测晕动病的计算模型已经取得了进展，特别是那些基于主观垂直冲突（SVC）理论的模型。虽然基于SVC的模型主要预测晕动病发病率（MSI），即给定运动下会呕吐的人的百分比，但仍需要能够预测较轻个体症状的模型，这些症状对于日常应用至关重要。最近，通过将6自由度SVC模型的输出从MSI改为痛苦量表（MISC）（一种症状进展的主观测量方法），开发了使用SVC理论预测前庭晕动病进展的计算模型。在实际应用中，预测未见过的运动的MISC的能力至关重要。本研究构思了一种通过从截至该点收集的数据中识别参数来预测未来某一时刻之后的MISC的方法。因此，本研究调查了用于参数识别的数据点数量对未来预测准确性的影响。使用在黑暗中暴露于线性横向运动的参与者的观察到的MISC反应进行模型验证。结果表明，随着纳入的数据点增多，预测准确性提高。平均而言，与使用参与者平均参数集的模型相比，使用超过5 - 10分钟的数据显著提高了准确性，尽管根据个体的MISC历史，这种趋势存在显著差异。一项考虑个体MISC历史的试验，其中当观察到的MISC首次达到某些水平时定义数据点，结果显示当使用达到MISC 3级的数据时，准确性普遍有提高的趋势。本研究结果表明，通过纳入个体症状历史，可以更准确地预测晕动病症状进展，从而为开发适用于更广泛应用的个性化晕动病预测模型奠定基础。
