Prognostic predictors of outcome in an operative series in traumatic brain injury patients

Automated pupillometry and optic nerve sheath diameter ultrasound to define tuberculous meningitis disease severity and prognosis

BACKGROUND

Tuberculous meningitis (TBM) causes high mortality and morbidity, in part due to raised intracranial pressure (ICP). Automated pupillometry (NPi) and optic nerve sheath diameter (ONSD) are both low-cost, easy-to-use and non-invasive techniques that correlate with ICP and neurological status. However, it is uncertain how to apply these techniques in the management of TBM.

METHODS

We conducted a pilot study enrolling 20 adults with TBM in the Hospital for Tropical Diseases, Ho Chi Minh City, Vietnam. Our objective was to investigate the relationships between baseline and serial measurements of NPi and ONSD and disease severity and outcome. Serial NPi and ONSD were performed for 30 days, at discharge, and at 3-months, with measurements correlated with clinical progression and outcomes.

RESULTS

ONSD and NPi measurements had an inverse relationship. Higher ONSD and lower NPi values were associated with lower Glasgow coma score. Baseline NPi was a strong predictor 3-month outcome (median NPi 4.55, interquartile range 4.35-4.65 for good outcomes versus 2.60, IQR 0.65-3.95 for poor outcomes, p = 0.002). Pupil inequality (NPi ≥0.7) was also strongly associated with poor 3-month outcomes (p = 0.006). Individual participants' serial NPi and ONSD were variable during initial treatment and correlated with clinical condition and outcome.

CONCLUSION

Pupillometry and ONSD may be used to predict clinical deterioration and outcome from TBM. Future, larger studies are need explore the optimal timing of measurements and to define how they might be used to optimise treatments and improve outcomes from TBM.

A Computer-Assisted System for Early Mortality Risk Prediction in Patients with Traumatic Brain Injury Using Artificial Intelligence Algorithms in Emergency Room Triage

Traumatic brain injury (TBI) remains a critical public health challenge. Although studies have found several prognostic factors for TBI, a useful early predictive tool for mortality has yet to be developed in the triage of the emergency room. This study aimed to use machine learning algorithms of artificial intelligence (AI) to develop predictive models for TBI patients in the emergency room triage. We retrospectively enrolled 18,249 adult TBI patients in the electronic medical records of three hospitals of Chi Mei Medical Group from January 2010 to December 2019, and undertook the 12 potentially predictive feature variables for predicting mortality during hospitalization. Six machine learning algorithms including logistical regression (LR) random forest (RF), support vector machines (SVM), LightGBM, XGBoost, and multilayer perceptron (MLP) were used to build the predictive model. The results showed that all six predictive models had high AUC from 0.851 to 0.925. Among these models, the LR-based model was the best model for mortality risk prediction with the highest AUC of 0.925; thus, we integrated the best model into the existed hospital information system for assisting clinical decision-making. These results revealed that the LR-based model was the best model to predict the mortality risk in patients with TBI in the emergency room. Since the developed prediction system can easily obtain the 12 feature variables during the initial triage, it can provide quick and early mortality prediction to clinicians for guiding deciding further treatment as well as helping explain the patient’s condition to family members.

Multiple Machine Learning Approaches Based on Postoperative Prediction of Pulmonary Complications in Patients With Emergency Cerebral Hemorrhage Surgery

Objective

This study aimed to create a prediction model of postoperative pulmonary complications for the patients with emergency cerebral hemorrhage surgery.

Methods

Patients with hemorrhage surgery who underwent cerebral hemorrhage surgery were included and divided into two groups: patients with or without pulmonary complications. Patient characteristics, previous history, laboratory tests, and interventions were collected. Univariate and multivariate logistic regressions were used to predict postoperative pulmonary infection. Multiple machine learning approaches have been used to compare their importance in predicting factors, namely K-nearest neighbor (KNN), stochastic gradient descent (SGD), support vector classification (SVC), random forest (RF), and logistics regression (LR), as they are the most successful and widely used models for clinical data.

Results

Three hundred and fifty four patients with emergency cerebral hemorrhage surgery between January 1, 2017 and December 31, 2020 were included in the study. 53.7% (190/354) of the patients developed postoperative pulmonary complications (PPC). Stepwise logistic regression analysis revealed four independent predictive factors associated with pulmonary complications, including current smoker, lymphocyte count, clotting time, and ASA score. In addition, the RF model had an ideal predictive performance.

Conclusions

According to our result, current smoker, lymphocyte count, clotting time, and ASA score were independent risks of pulmonary complications. Machine learning approaches can also provide more evidence in the prediction of pulmonary complications.

Random Forest-Based Prediction of Outcome and Mortality in Patients with Traumatic Brain Injury Undergoing Primary Decompressive Craniectomy

BACKGROUND

Various prognostic models are used to predict mortality and functional outcome in patients after traumatic brain injury with a trend to incorporate machine learning protocols. None of these models is focused exactly on the subgroup of patients indicated for decompressive craniectomy. Evidence regarding efficiency of this surgery is still incomplete, especially in patients undergoing primary decompressive craniectomy with evacuation of traumatic mass lesions.

METHODS

In a prospective study with a 6-month follow-up period, we assessed postoperative outcome and mortality of 40 patients who underwent primary decompressive craniectomy for traumatic brain injuries during 2018-2019. The results were analyzed in relation to a wide spectrum of preoperatively available demographic, clinical, radiographic, and laboratory data. Random forest algorithms were trained for prediction of both mortality and unfavorable outcome, with their accuracy quantified by area under the receiver operating curves (AUCs) for out-of-bag samples.

RESULTS

At the end of the follow-up period, we observed mortality of 57.5%. Favorable outcome (Glasgow Outcome Scale [GOS] score 4-5) was achieved by 30% of our patients. Random forest-based prediction models constructed for 6-month mortality and outcome reached a moderate predictive ability, with AUC = 0.811 and AUC = 0.873, respectively. Random forest models trained on handpicked variables showed slightly decreased AUC = 0.787 for 6-month mortality and AUC = 0.846 for 6-month outcome and increased out-of-bag error rates.

CONCLUSIONS

Random forest algorithms show promising results in prediction of postoperative outcome and mortality in patients undergoing primary decompressive craniectomy. The best performance was achieved by Classification Random forest for 6-month outcome.

Effects of Primary Decompressive Craniectomy on the Outcomes of Serious Traumatic Brain Injury with Mass Lesions, and Independent Predictors of Operation Decision

BACKGROUND

Although operative indications for traumatic brain injury (TBI) are known, neurosurgeons are unsure whether to remove the bone flap after mass lesion extraction, and an efficient scoring system for predicting which patients should undergo decompressive craniectomy (DC) does not exist.

METHODS

Nine parameters were assessed. In total, 245 patients with severe TBI were retrospectively assessed from June 2015 to May 2019, who underwent DC or craniotomy to remove mass lesions. The 6-month mortality and Extended Glasgow Outcome Scale scores were compared between the DC and craniotomy groups. Using univariable and multivariable logistic regression equations, receiver operating characteristic curves were obtained for predicting the decision for DC.

RESULTS

The overall 6-month mortality in the entire cohort was 11.43% (28/245). Patients undergoing DC had lower mean preoperative Glasgow Coma Scale scores (P = 0.01), and higher amounts of individuals with a Glasgow Coma Scale score of 6 (P = 0.007), unresponsive pupillary light reflex (P < 0.001), closed basal cisterns (P < 0.001), and diffuse injury (P = 0.025), compared with the craniotomy group. Because of high disease severity, individuals administered primary DC showed increased 6-month mortality compared with the craniotomy group. However, in surviving patients, favorable Extended Glasgow Outcome Scale rates were similar in both groups. Pupillary light reflex and basal cisterns were independent predictors of the DC decision. Based on receiver operating characteristic curves, the model had sensitivity and specificity of 81.6% and 84.9%, respectively, in predicting the probability of DC.

CONCLUSIONS

These preliminary data showed that primary DC may benefit some patients with severe TBI with mass lesions. In addition, unresponsive preoperative pupil reaction and closed basal cistern could predict the DC decision.

Automated Pupillary Measurements Inversely Correlate With Increased Intracranial Pressure in Pediatric Patients With Acute Brain Injury or Encephalopathy

OBJECTIVES

The purpose of this study was to determine correlation and temporal association between automated pupillary measurements and intracranial pressure in pediatric patients with brain injury or encephalopathy requiring intracranial pressure monitoring. We hypothesized that abnormal pupillary measurements would precede increases in intracranial pressure.

DESIGN

A prospective cohort study was performed. Automated pupillometry measurements were obtained at the same frequency as the patients' neurologic assessments with concurrent measurement of intracranial pressure, for up to 72 hours. Pupillary measurements and the Neurologic Pupil index, an algorithmic score that combines measures of pupillary reactivity, were assessed for correlation with concurrent and future intracranial pressure measurements.

SETTING

Single-center pediatric quaternary ICU, from July 2017 to October 2018.

PATIENTS

Pediatric patients 18 years or younger with a diagnosis of acute brain injury or encephalopathy requiring an intracranial pressure monitor.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Twenty-eight patients were analyzed with a total of 1,171 intracranial pressure measurements. When intracranial pressure was elevated, the Neurologic Pupil index, percent change in pupillary size, constriction velocity, and dilation velocity were significantly lower than when intracranial pressure was within normal range (p < 0.001 for all). There were mild to moderate negative correlations between concurrent intracranial pressure and pupillary measurements. However, there was an inconsistent pattern of abnormal pupillary measurements preceding increases in intracranial pressure; some patients had a negative association, while others had a positive relationship or no relationship between Neurologic Pupil index and intracranial pressure.

CONCLUSIONS

Our data indicate automated assessments of pupillary reactivity inversely correlate with intracranial pressure, demonstrating that pupillary reactivity decreases as intracranial pressure increases. However, a temporal association in which abnormal pupillary measurements precede increases in intracranial pressure was not consistently observed. This work contributes to limited data available regarding automated pupillometry in neurocritically ill patients, and the even more restricted subset available in pediatrics.

A Critical Review for Developing Accurate and Dynamic Predictive Models Using Machine Learning Methods in Medicine and Health Care

Recently, Artificial Intelligence (AI) has been used widely in medicine and health care sector. In machine learning, the classification or prediction is a major field of AI. Today, the study of existing predictive models based on machine learning methods is extremely active. Doctors need accurate predictions for the outcomes of their patients' diseases. In addition, for accurate predictions, timing is another significant factor that influences treatment decisions. In this paper, existing predictive models in medicine and health care have critically reviewed. Furthermore, the most famous machine learning methods have explained, and the confusion between a statistical approach and machine learning has clarified. A review of related literature reveals that the predictions of existing predictive models differ even when the same dataset is used. Therefore, existing predictive models are essential, and current methods must be improved.

Changes in Pupil Size Following Panretinal and Focal/Grid Retinal Photocoagulation: Automatic Infrared Pupillometry Study

PURPOSE

To evaluate possible changes in pupil size subsequent to panretinal and focal/grid laser photocoagulation.

METHODS

Sixty-four eyes of 64 participants were included. Thirty-two eyes with planned panretinal photocoagulation formed Group 1, and 32 eyes with planned focal retinal photocoagulation formed Group 2. The participants underwent full ophthalmologic examination at baseline. Automated infrared pupillometry was performed at baseline and month 1.

RESULTS

The mean pupillary measurements (in millimeters) for Group 1 (in order photopic, mesopic, scotopic) were 3.09 ± 0.69 mm, 3.66 ± 0.85 mm, and 3.87 ± 1.01 mm and changed to 3.34 ± 0.74 mm, 3.82 ± 0.92 mm, and 4.03 ± 1.02 mm. There was a significant increase in pupil size at month 1 (P = 0.001, P = 0.001, P = 0.003). For Group 2, they were 2.65 ± 0.87 mm, 3.08 ± 1.08 mm, and 3.18 ± 1.19 mm and changed to 2.92 ± 0.72 mm, 3.45 ± 0.76 mm, and 3.57 ± 0.88 mm. There was no significant difference in pupil size at month 1 (P = 0.151, P = 0.106, P = 0.095).

CONCLUSION

We have demonstrated through automated infrared pupillary measurements that panretinal laser photocoagulation may significantly influence pupil size and focal/grid laser photocoagulation may not.

Infrared pupillometry, the Neurological Pupil index and unilateral pupillary dilation after traumatic brain injury: implications for treatment paradigms

Pupillary dysfunction, a concerning finding in the neurologic examination of the patient with an acute traumatic brain injury often dictates the subsequent treatment paradigm. Patients were monitored closely with an infrared pupillometer, with NPi technology, for acute changes in pupillary function. NPi technology applies a scalar value to pupillary function. A retrospective chart review was performed of traumatic brain injury patients with acute unilateral pupillary dilation, admitted to Legacy Emanuel Medical Center’s NeuroTrauma Unit, Portland, OR, and followed as outpatients, between January 2012 and December 2013. Clinical exam findings of pupillary size, NPi scores, and brain Magnetic Resonance Imaging and Computed Tomography images were analyzed. Five traumatic brain injury patients were identified with unilateral pupillary dysfunction with long-term follow-up after the initial injury. Each patient was monitored closely in the trauma bay for neurological deterioration with a pupillometer and the clinical exam. Two patients underwent subsequent intracranial pressure monitoring based on a deteriorating clinical scenario, including consistent abnormal unilateral NPi scores. One patient with consistent abnormal NPi scores and an improved clinical exam did not undergo invasive interventions. Two patients showed early improvement in NPi scores correlating with the normalization of their pupillary reactivity. Anisocoria improved in all patients despite concurrent abnormal NPi scores. Magnetic Resonance Imaging and Computed Tomography imaging studies, with a focus on the third nerve, revealed focal abnormalities consistent with the clinical findings. A unilateral blown pupil and abnormal NPi score in a traumatic brain injury patient are not necessarily indicative of intracranial pressure issues, and must be correlated with the entire clinical scenario, to determine the etiology of the third nerve injury and direct potential therapeutic interventions. Early NPi score normalization suggests pupillary function may improve. We found that NPi scores, as a component of the clinical exam, provide a sensitive, noninvasive and quantitative means of following pupillary function acutely and chronically after a traumatic brain injury.

Pupillary motility: bringing neuroscience to the psychiatry clinic of the future

Modern pupillometry has expanded the study and utility of pupil responses in many new domains, including psychiatry, particularly for understanding aspects of cognitive and emotional information processing. Here, we review the applications of pupillometry in psychiatry for understanding patients' information processing styles, predicting treatment, and augmenting function. In the past year pupillometry has been shown to be useful in specifying cognitive/affective occurrences during experimental tasks and informing clinical diagnoses. Such studies demonstrate the potential of pupillary motility to be used in clinical psychiatry much as it has been in neurology for the past century.

What is New and Innovative in Emergency Neurosurgery? Emerging Diagnostic Technologies Provide Better Care and Influence Outcome: A Specialist Review

The development of emergency medical services and especially neurosurgical emergencies during recent decades has necessitated the development of novel tools. Although the gadgets that the neurosurgeon uses today in emergencies give him important help in diagnosis and treatment, we still need new technology, which has rapidly developed. This review presents the latest diagnostic tools, which offer precious help in everyday emergency neurosurgery practice. New ultrasound devices make the diagnosis of haematomas easier. In stroke, the introduction of noninvasive new gadgets aims to provide better treatment to the patient. Finally, the entire development of computed tomography and progress in radiology have resulted in innovative CT scans and angiographic devices that advance the diagnosis, treatment, and outcome of the patent. The pressure on physicians to be quick and effective and to avoid any misjudgement of the patient has been transferred to the technology, with the emphasis on developing new systems that will provide our patients with a better outcome and quality of life.

Trajectories of sleep changes during the acute phase of traumatic brain injury: a 7-day actigraphy study

BACKGROUND/PURPOSE

To examine trajectories of change in sleep during the acute phase of traumatic brain injury (TBI), and whether specific demographic and disease characteristics predicted the initial levels of sleep and the trajectories of change in sleep parameters.

METHODS

This was a prospective observational study. Fifty-two patients with first-ever TBI were enrolled. Sleep parameters were measured using actigraphy for 7 consecutive days after admission. Hierarchical linear modeling was used for data analyses in 52 TBI patients and in a subgroup of mild TBI patients (n = 31).

RESULTS

Participants had significant lower sleep efficiency, longer wake time after sleep onset, and longer 24-hour total sleep time (TST) than the normative data (all p < 0.05). Seventy-two percent of participants experienced prolonged 24-hour TST. Both daytime and 24-hour TST showed a significant downward trend across the study period. An initial Glasgow Coma Scale score < 11 significantly predicted the slope of change of daytime TST over time. Without initial loss of consciousness and age < 40 years were independent predictors of the change pattern of 24-hour TST over time. In the mild TBI subgroup, 24-hour TST significantly and gradually declined over time. Gender significantly predicted the trajectory of 24-hour sleep duration.

CONCLUSION

Poor sleep efficiency and longer sleep duration are common symptoms in acute TBI patients. Both head injury severity and age predicted the trajectories of daytime and 24-hour sleep duration during the acute phase of TBI, whereas gender predicted the trajectories of 24-hour sleep duration in the mild TBI subgroup.

[Prognostic factors in severe traumatic brain injury]

OBJECTIVE

To describe the factors associated to mortality of patients with severe traumatic brain injury (TBI).

DESIGN

A retrospective observational study was made of patients with severe TBI covering the period between 1 January 2007 and 31 December 2010.

SETTING

Virgen de la Vega Hospital, Salamanca (Spain).

PATIENTS

All patients hospitalized during the consecutive study period.

MAIN VARIABLES OF INTEREST

The dependent variable has been the death rate, while the independent variables were divided into two groups: epidemiological and clinical.

RESULTS

The mean patient age was the 50.84 years; 75.5% were males. The average score on the Glasgow Coma Scale (GCS) was 5.09, and the average Injury Severity Score (ISS) was 30.8. Higher mortality was observed in older patients with a higher ISS score. A total of 68.1% of the patients in which intracranial pressure (ICP) was monitored showed intracranial hypertension (ICH). The mortality rate in patients with severe TBI was 36.8%, and was associated mainly to a lower GCS score, the existence of hyperglycemia, coagulopathy, hypoxemia, the presence of mydriasis and shock. The independent mortality indicators in patients with severe TBI were the existence of mydriasis (OR: 32.75), the GCS score (OR: 2.65) and hyperglycemia (OR: 6.08).

CONCLUSIONS

The existence of mydriasis, the GCS score and hyperglycemia should be taken into account as prognostic factors in TBI patients.

Operative and nonoperative linguistic outcomes in brain injury patients

OBJECTIVE AND BACKGROUND

Linguistic function is one of vulnerable aspects of traumatic brain injury (TBI) which may have destructive effects on patients' communicative activities and daily life, years following trauma. This paper attempts to answer the controversy whether surgery affects increase and decrease of linguistic impairment or not.

MATERIALS AND METHODS

Two hundred forty-one TBI patients aged 18-65 with abnormal CT findings and at least 20 minute post-trauma amnesia (PTA), who were conscious at discharge, participated in this study. Based on operative intervention, the samples were divided into two groups: operative and nonoperative. Cognitive and aphasic deficits were inspected formally and pragmatic disorder was informally appraised at discharge.

RESULTS

The groups had no significant differences in aphasia incidence and language pragmatic impairment, though they were significantly distinctive in aphasia subcategories and cognitive deficit after trauma. Fluent aphasia was more common in both groups alike. In aphasia subcategories, however, transcortical sensory aphasia (TSA) in operative and anomia in nonoperative group were the most prevalent. Several variables appeared strikingly related to higher aphasia in operative groups as follows: moderate to severe injury, 18-35 and over 50 years of age, more than 1 week PTA, intracranial surgery of multiple lesions in left or bilateral hemisphere fronto-temporal cortex plus post-trauma cognitive and pragmatic impairments, and diffuse axonal injuries.

DISCUSSION

Almost certainly, meaningful drop of cognitive function post surgery roots back in significant loss of initial consciousness level. Related factors to postoperative aphasia suggest taking policies through surgery intervention. Discerning the indispensable contributions of neurosurgeons, neurolinguists, and neuroscientists, results inspire more clinical future studies.

Effective factors on linguistic disorder during acute phase following traumatic brain injury in adults

Traumatic brain injury (TBI) has been known to be the leading cause of breakdown and long-term disability in people under 45 years of age. This study highlights the effective factors on post-traumatic (PT) linguistic disorder and relations between linguistic and cognitive function after trauma in adults with acute TBI. A cross-sectional design was employed to study 60 post-TBI hospitalized adults aged 18-65 years. Post-traumatic (PT) linguistic disorder and cognitive deficit after TBI were respectively diagnosed using the Persian Aphasia Test (PAT) and Persian version of Mini-Mental State Examination (MMSE) at discharge. Primary post-resuscitation consciousness level was determined using the Glasgow Coma Scale (GCS). Paracilinical data was obtained by CT scan technique. Multiple logistic regression analysis illustrated that brain injury severity was the first powerful significant predictor of PT linguistic disorder after TBI and frontotemporal lesion was the second. It was also revealed that cognitive function score was significantly correlated with score of each language skill except repetition. Subsequences of TBI are more commonly language dysfunctions that demand cognitive flexibility. Moderate, severe and fronto-temporal lesion can increase the risk of processing deficit in linguistic macrostructure production and comprehension. The dissociation risk of cortical and subcortical pathways related to cognitive-linguistic processing due to intracranial lesions can augment possibility of lexical-semantic processing deficit in acute phase which probably contributes to later cognitive-communication disorder.