Clinical elements that predict outcome after traumatic brain injury: a prospective multicenter recursive partitioning (decision-tree) analysis.

Department of Physical Medicine and Rehabilitation, Mayo Clinic College of Medicine, Rochester, MN 55905, USA.
Journal of Neurotrauma (Impact Factor: 3.97). 11/2005; 22(10):1040-51. DOI: 10.1089/neu.2005.22.1040
Source: PubMed

ABSTRACT Traumatic brain injury (TBI) often presents clinicians with a complex combination of clinical elements that can confound treatment and make outcome prediction challenging. Predictive models have commonly used acute physiological variables and gross clinical measures to predict mortality and basic outcome endpoints. The primary goal of this study was to consider all clinical elements available concerning a survivor of TBI admitted for inpatient rehabilitation, and identify those factors that predict disability, need for supervision, and productive activity one year after injury. The Traumatic Brain Injury Model Systems (TBIMS) database was used for decision tree analysis using recursive partitioning (n = 3463). Outcome measures included the Functional Independence Measure(), the Disability Rating Scale, the Supervision Rating Scale, and a measure of productive activity. Predictor variables included all physical examination elements, measures of injury severity (initial Glasgow Coma Scale score, duration of post-traumatic amnesia [PTA], length of coma, CT scan pathology), gender, age, and years of education. The duration of PTA, age, and most elements of the physical examination were predictive of early disability. The duration of PTA alone was selected to predict late disability and independent living. The duration of PTA, age, sitting balance, and limb strength were selected to predict productive activity at 1 year. The duration of PTA was the best predictor of outcome selected in this model for all endpoints and elements of the physical examination provided additional predictive value. Valid and reliable measures of PTA and physical impairment after TBI are important for accurate outcome prediction.

  • [Show abstract] [Hide abstract]
    ABSTRACT: Background. Several novel drug- and cell-based potential therapies for spinal cord injury (SCI) have either been applied or will be considered for future clinical trials. Limitations on the number of eligible patients require trials be undertaken in a highly efficient and effective manner. However, this is particularly challenging when people living with incomplete SCI (iSCI) represent a very heterogeneous population in terms of recovery patterns and can improve spontaneously over the first year after injury. Objective. The current study addresses 2 requirements for designing SCI trials: first, enrollment of as many eligible participants as possible; second, refined stratification of participants into homogeneous cohorts from a heterogeneous iSCI population. Methods. This is a retrospective, longitudinal analysis of prospectively collected SCI data from the European Multicenter study about Spinal Cord Injury (EMSCI). We applied conditional inference trees to provide a prediction-based stratification algorithm that could be used to generate decision rules for the appropriate inclusion of iSCI participants to a trial. Results. Based on baseline clinical assessments and a defined subsequent clinical endpoint, conditional inference trees partitioned iSCI participants into more homogeneous groups with regard to the illustrative endpoint, upper extremity motor score. Assuming a continuous endpoint, the conditional inference tree was validated both internally as well as externally, providing stable and generalizable results. Conclusion. The application of conditional inference trees is feasible for iSCI participants and provides easily implementable, prediction-based decision rules for inclusion and stratification. This algorithm could be utilized to model various trial endpoints and outcome thresholds. © The Author(s) 2015.
    Neurorehabilitation and neural repair 02/2015; DOI:10.1177/1545968315570322 · 4.62 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Previous studies have identified some clinical parameters for predicting long-term functional recovery and mortality after traumatic brain injury (TBI). Here, data mining methods were combined with serial Glasgow Coma Scale (GCS) scores and clinical and laboratory parameters to predict 6-month functional outcome and mortality in patients with TBI. Data of consecutive adult patients presenting at a trauma center with moderate-to-severe head injury were retrospectively analyzed. Clinical parameters including serial GCS measurements at emergency department, 7th day, and 14th day and laboratory data were included for analysis (n = 115). We employed artificial neural network (ANN), naïve Bayes (NB), decision tree, and logistic regression to predict mortality and functional outcomes at 6 months after TBI. Favorable functional outcome was achieved by 34.8 % of the patients, and overall 6-month mortality was 25.2 %. For 6-month functional outcome prediction, ANN was the best model, with an area under the receiver operating characteristic curve (AUC) of 96.13 %, sensitivity of 83.50 %, and specificity of 89.73 %. The best predictive model for mortality was NB with AUC of 91.14 %, sensitivity of 81.17 %, and specificity of 90.65 %. Sensitivity analysis demonstrated GCS measurements on the 7th and 14th day and difference between emergency room and 14th day GCS score as the most influential attributes both in mortality and functional outcome prediction models. Analysis of serial GCS measurements using data mining methods provided additional predictive information in relation to 6-month mortality and functional outcome in patients with moderate-to-severe TBI.
    Journal of Medical Systems 02/2015; 39(2):187. DOI:10.1007/s10916-014-0187-x · 1.37 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Predicting the evolution of individuals is a rather new mining task with applications in medicine. Medical researchers are interested in the progression of a disease and/or how do patients evolve or recover when they are subjected to some treatment. In this study, we investi- gate the problem of patients’ evolution on the basis of medical tests before and after treatment after brain trauma: we want to understand to what extend a patient can become similar to a healthy participant. We face two challenges. First, we have less information on healthy participants than on the patients. Second, the values of the medical tests for patients, even after treatment started, remain well-separated from those of healthy people; this is typical for neurodegenerative diseases, but also for further brain impairments. Our approach encompasses methods for mod- elling patient evolution and for predicting the health improvement of different patients’ subpopulations, i.e. prediction of label if they recovered or not. We test our approach on a cohort of patients treated after brain trauma and a corresponding cohort of controls.
    02/2015; 2(1):33-44. DOI:10.1007/s40708-015-0010-6