Predicting risk for trauma patients using static and dynamic information from the MIMIC III database.

Risk quantification algorithms in the ICU can provide (1) an early alert to the clinician that a patient is at extreme risk and (2) help manage limited resources efficiently or remotely. With electronic health records, large data sets allow the training of predictive models to quantify patient risk. A gradient boosting classifier was trained to predict high-risk and low-risk trauma patients, where patients were labeled high-risk if they expired within the next 10 hours or within the last 10% of their ICU stay duration. The MIMIC-III database was filtered to extract 5,400 trauma patient records (526 non-survivors) each of which contained 5 static variables (age, gender, etc.) and 28 dynamic variables (e.g., vital signs and metabolic panel). Training data was also extracted from the dynamic variables using a 3-hour moving time window whereby each window was treated as a unique patient-time fragment. We extracted the mean, standard deviation, and skew from each of these 3-hour fragments and included them as inputs for training. Additionally, a survival metric upon admission was calculated for each patient using a previously developed National Trauma Data Bank (NTDB)-trained gradient booster model. The final model was able to distinguish between high-risk and low-risk patients to an AUROC of 92.9%, defined as the area under the receiver operator characteristic curve. Importantly, the dynamic survival probability plots for patients who die appear considerably different from those who survive, an example of reducing the high dimensionality of the patient record to a single trauma trajectory.

Using the National Trauma Data Bank (NTDB) and machine learning to predict trauma patient mortality at admission.

A 400-estimator gradient boosting classifier was trained to predict survival probabilities of trauma patients. The National Trauma Data Bank (NTDB) provided 799233 complete patient records (778303 survivors and 20930 deaths) each containing 32 features, a number further reduced to only 8 features via the permutation importance method. Importantly, the 8 features can all be readily determined at admission: systolic blood pressure, heart rate, respiratory rate, temperature, oxygen saturation, gender, age and Glasgow coma score. Since death was rare, a rebalanced training set was used to train the model. The model is able to predict a survival probability for any trauma patient and accurately distinguish between a deceased and survived patient in 92.4% of all cases. Partial dependence curves (Psurvival vs. feature value) obtained from the trained model revealed the global importance of Glasgow coma score, age, and systolic blood pressure while pulse rate, respiratory rate, temperature, oxygen saturation, and gender had more subtle single variable influences. Shapley values, which measure the relative contribution of each of the 8 features to individual patient risk, were computed for several patients and were able to quantify patient-specific warning signs. Using the NTDB to sample across numerous patient traumas and hospital protocols, the trained model and Shapley values rapidly provides quantitative insight into which combination of variables in an 8-dimensional space contributed most to each trauma patient's predicted global risk of death upon emergency room admission.

Coagulopathy implications using a multiscale model of traumatic bleeding matching macro- and microcirculation.

Quantifying the relationship between vascular injury and the dynamic bleeding rate requires a multiscale model that accounts for changing and coupled hemodynamics between the global and microvascular levels. A lumped, global hemodynamic model of the human cardiovascular system with baroreflex control was coupled to a local 24-level bifurcating vascular network that spanned diameters from the muscular artery scale (0.1-1.3 mm) to capillaries (5-10 µm) via conservation of momentum and conservation of mass boundary conditions. For defined injuries of severing all vessels at each nth-level, the changing pressures and flowrates were calculated using prescribed shear-dependent hemostatic clot growth rates (normal or coagulopathic). Key results were as follows: 1) the upstream vascular network rapidly depressurizes to reduce blood loss; 2) wall shear rates at the hemorrhaging wound exit are sufficiently high (~10,000 s-1) to drive von Willebrand factor unfolding; 3) full coagulopathy results in >2-liter blood loss in 2 h for severing all vessels of 0.13- to 0.005-mm diameter within the bifurcating network, whereas full hemostasis limits blood loss to <100 ml within 2 min; and 4) hemodilution from transcapillary refill increases blood loss and could be implicated in trauma-induced coagulopathy. A sensitivity analysis on length-to-diameter ratio and branching exponent demonstrated that bleeding was strongly dependent on these tissue-dependent network parameters. This is the first bleeding model that prescribes the geometry of the injury to calculate the rate of pressure-driven blood loss and local wall shear rate in the presence or absence of coagulopathic blood. NEW & NOTEWORTHY We developed a multiscale model that couples a lumped, global hemodynamic model of a patient to resolved, single-vessel wounds ranging from the small artery to capillary scale. The model is able to quantify wall shear rates, seal rates, and blood loss rates in the presence and absence of baroreflex control and hemodilution.