Predicting intra-abdominal hypertension using anthropometric measurements and machine learning.

Almost one in four critically ill patients suffer from intra-abdominal hypertension (IAH). Currently, the gold standard for measuring intra-abdominal pressure (IAP) is via the bladder. Measurement of IAP is important to identify IAH early and thus implement appropriate management in order to avoid complications. It may be possible to use anthropometric parameters to predict IAP and thus identify IAH non-invasively. This retrospective observational study investigated how the most relevant body parameters evolve in relation to IAP, and whether IAP can be predicted based on anthropometric parameters. The IAP and 28 body parameters of 96 critically ill patients were recorded. Following statistical analyses such as Pearson's and mutual information correlation, the collected data were used to train a simulation model to examine reliable relationships between IAP, predict IAP values, and detect IAH. Three metrics were shown to sufficiently predict intra-bladder pressure (IBP) with a Pearson's correlation of 0.75 (R = 0.56). These parameters are the difference between the convex and horizontal xiphoid-to-pubis distance, sagittal abdominal diameter, and abdominal compliance. Subsequently, we found 1 metric that is able to predict the presence of IAH with Pearson correlation of 0.89 (R = 0.79). This metric is the difference between the convex and horizontal xiphoid to pubis distance. Three measured body parameters showed a correlation of more than 50% with IBP and they are sufficient for a reliable prediction of IBP, however, IAH can be most reliably predicted based on the difference between the convex and horizontal xiphoid-pubis distance and sagittal abdominal diameter. Future studies with larger patient populations and diverse body shapes are warranted to confirm these findings.