We developed risk prediction scores using established methods in both groups. We used a multivariate logistic regression model to estimate the b coefficients for each risk factor. Variables were retained in the final multivariate model if they had an odds ratio of more than 1.2. We internally validated the scores by using bootstrap resampling with 1000 replications of both groups to estimate b�� shrinkage coefficients. We used the median bootstrapped regression coefficients rounded to the nearest half point as weights, which we combined with the baseline logit function to derive risk prediction scores. We calculated the sensitivity, specificity, predictive values and proportion of correctly classified subjects for each value of the risk prediction score to determine thresholds. This is the first study proposing a risk prediction scoring system for SHAI based on a large cohort of sea-level residents FTY720 inquirer visiting high altitude regions. Ten clinical, environmental and physiological variables were used to compute the risk prediction scoring, with a different weight applied to the 10 items according to presence or absence of previous experience at high altitude. It ranged from 0 to 10 points in subjects without previous experience and 0 to 12 in those with. The NSC-718781 EGFR/HER2 inhibitor obtained scoring systems had very good to excellent discrimination ability and adequate calibration. The two groups showed some different characteristics : as it might be expected, subjects with previous experience at high altitude were older, more trained, more men, had slightly higher blood pressure and went to higher risk locations. These differences justify the stratification made before performing the multivariate logistic regression. We chose to develop two separate multivariate models leading to two prediction scores, in the group of subjects with previous exposure and in the group of subjects without. Indeed, even if the two multivariate models are very close, some predictors differ, precluding the possibility of using one group as development and the other one as validation datasets.