Kannan N, Raychaudhuri A, Pilmanis A A
Division of Mathematics and Statistics, University of Texas at San Antonio, 78249, USA.
Aviat Space Environ Med. 1998 Oct;69(10):965-70.
Altitude decompression sickness (DCS) is a potential hazard encountered during high altitude flights or during extravehicular activity in space. In this study, the loglogistic distribution was used to model DCS risk and symptom onset time.
The Air Force Research Laboratory, Brooks AFB, TX, has conducted studies on human subjects exposed to simulated altitudes in hypobaric chambers. The dataset from those studies was used to develop the DCS models and consisted of 975 subject-exposures to various altitudes, preoxygenation times, and exercise regimens. Since the risk of DCS is known to increase over time at altitude, and then decrease because of denitrogenation, the loglogistic model was fit to the data. The model assumes that the probability of DCS depends on several risk factors. Maximum likelihood estimates of the parameters were obtained using the statistical software package SAS. Cross validation techniques were provided to examine the goodness of fit of the model.
The fitted model indicated that altitude, ratio of preoxygenation to exposure time, and exercise were the most significant risk factors. The model was used to predict the risk of DCS for a variety of exposure profiles. The predicted probability of DCS agreed very closely with the actual percentages in the database.
The loglogistic distribution was found to be appropriate for modeling the risk of DCS. Based on the cross validation and validation results, we conclude that this model provides good estimates of the probability of DCS over time.
高空减压病（DCS）是在高空飞行或太空舱外活动期间可能遇到的一种潜在危害。在本研究中，采用对数逻辑斯蒂分布对DCS风险和症状发作时间进行建模。
位于德克萨斯州布鲁克斯空军基地的空军研究实验室对在低压舱中暴露于模拟高度的人体受试者进行了研究。这些研究的数据集被用于开发DCS模型，该数据集包含975次受试者暴露于不同高度、预充氧时间和运动方案的情况。由于已知DCS风险在高空会随时间增加，然后因去氮作用而降低，因此将对数逻辑斯蒂模型拟合到数据中。该模型假设DCS的概率取决于几个风险因素。使用统计软件包SAS获得参数的最大似然估计值。提供交叉验证技术以检验模型的拟合优度。
拟合模型表明，高度、预充氧时间与暴露时间的比值以及运动是最显著的风险因素。该模型被用于预测各种暴露情况的DCS风险。DCS的预测概率与数据库中的实际百分比非常接近。
发现对数逻辑斯蒂分布适用于对DCS风险进行建模。基于交叉验证和验证结果，我们得出结论，该模型能很好地估计随时间变化的DCS概率。
