Comparison of intracranial injury predictability between machine learning algorithms and the nomogram in pediatric traumatic brain injury

OBJECTIVE The overuse of head CT examinations has been much discussed, especially those for minor traumatic brain injury (TBI). In the disruptive era, machine learning (ML) is one of the prediction tools that has been used and applied in various fields of neurosurgery. The objective of this study was to compare the predictive performance between ML and a nomogram, which is the other prediction tool for intracranial injury following cranial CT in children with TBI. METHODS Data from 964 pediatric patients with TBI were randomly divided into a training data set (75%) for hyperparameter tuning and supervised learning from 14 clinical parameters, while the remaining data (25%) were used for validation purposes. Moreover, a nomogram was developed from the training data set with similar parameters. Therefore, models from various ML algorithms and the nomogram were built and deployed via web-based application. RESULTS A random forest classifier (RFC) algorithm established the best performance for predicting intracranial injury following cranial CT of the brain. The area under the receiver operating characteristic curve for the performance of RFC algorithms was 0.80, with 0.34 sensitivity, 0.95 specificity, 0.73 positive predictive value, 0.80 negative predictive value, and 0.79 accuracy. CONCLUSIONS The ML algorithms, particularly the RFC, indicated relatively excellent predictive performance that would have the ability to support physicians in balancing the overuse of head CT scans and reducing the treatment costs of pediatric TBI in general practice.

Download Full-text

Nowcasting heavy precipitation over the Netherlands using a 13-year radar archive: a machine learning approach

10.5194/egusphere-egu21-12814 ◽

2021 ◽

Author(s):

Eva van der Kooij ◽

Marc Schleiss ◽

Riccardo Taormina ◽

Francesco Fioranelli ◽

Dorien Lugt ◽

...

Keyword(s):

Machine Learning ◽

The Netherlands ◽

Heavy Rainfall ◽

Predictive Performance ◽

Heavy Precipitation ◽

Early Warning Systems ◽

Training Data ◽

Short Term ◽

Data Set ◽

Radar Images

Accurate short-term forecasts, also known as nowcasts, of heavy precipitation are desirable for creating early warning systems for extreme weather and its consequences, e.g. urban flooding. In this research, we explore the use of machine learning for short-term prediction of heavy rainfall showers in the Netherlands.We assess the performance of a recurrent, convolutional neural network (TrajGRU) with lead times of 0 to 2 hours. The network is trained on a 13-year archive of radar images with 5-min temporal and 1-km spatial resolution from the precipitation radars of the Royal Netherlands Meteorological Institute (KNMI). We aim to train the model to predict the formation and dissipation of dynamic, heavy, localized rain events, a task for which traditional Lagrangian nowcasting methods still come up short.We report on different ways to optimize predictive performance for heavy rainfall intensities through several experiments. The large dataset available provides many possible configurations for training. To focus on heavy rainfall intensities, we use different subsets of this dataset through using different conditions for event selection and varying the ratio of light and heavy precipitation events present in the training data set and change the loss function used to train the model.To assess the performance of the model, we compare our method to current state-of-the-art Lagrangian nowcasting system from the pySTEPS library, like S-PROG, a deterministic approximation of an ensemble mean forecast. The results of the experiments are used to discuss the pros and cons of machine-learning based methods for precipitation nowcasting and possible ways to further increase performance.

Download Full-text

Subspace Clustering of Physiological Data From Acute Traumatic Brain Injury Patients: Retrospective Analysis Based on the PROTECT III Trial

JMIR Biomedical Engineering ◽

10.2196/24698 ◽

2021 ◽

Vol 6 (1) ◽

pp. e24698

Author(s):

Sina Ehsani ◽

Chandan K Reddy ◽

Brandon Foreman ◽

Jonathan Ratcliff ◽

Vignesh Subbian

Keyword(s):

Machine Learning ◽

Traumatic Brain Injury ◽

Brain Injury ◽

Subspace Clustering ◽

Machine Learning Techniques ◽

Phase Iii ◽

Physiological Data ◽

Clustering Methods ◽

Data Set ◽

Health Technologies

Background With advances in digital health technologies and proliferation of biomedical data in recent years, applications of machine learning in health care and medicine have gained considerable attention. While inpatient settings are equipped to generate rich clinical data from patients, there is a dearth of actionable information that can be used for pursuing secondary research for specific clinical conditions. Objective This study focused on applying unsupervised machine learning techniques for traumatic brain injury (TBI), which is the leading cause of death and disability among children and adults aged less than 44 years. Specifically, we present a case study to demonstrate the feasibility and applicability of subspace clustering techniques for extracting patterns from data collected from TBI patients. Methods Data for this study were obtained from the Progesterone for Traumatic Brain Injury, Experimental Clinical Treatment–Phase III (PROTECT III) trial, which included a cohort of 882 TBI patients. We applied subspace-clustering methods (density-based, cell-based, and clustering-oriented methods) to this data set and compared the performance of the different clustering methods. Results The analyses showed the following three clusters of laboratory physiological data: (1) international normalized ratio (INR), (2) INR, chloride, and creatinine, and (3) hemoglobin and hematocrit. While all subclustering algorithms had a reasonable accuracy in classifying patients by mortality status, the density-based algorithm had a higher F1 score and coverage. Conclusions Clustering approaches serve as an important step for phenotype definition and validation in clinical domains such as TBI, where patient and injury heterogeneity are among the major reasons for failure of clinical trials. The results from this study provide a foundation to develop scalable clustering algorithms for further research and validation.

Download Full-text

Subspace Clustering of Physiological Data From Acute Traumatic Brain Injury Patients: Retrospective Analysis Based on the PROTECT III Trial (Preprint)

10.2196/preprints.24698 ◽

2020 ◽

Author(s):

Sina Ehsani ◽

Chandan K Reddy ◽

Brandon Foreman ◽

Jonathan Ratcliff ◽

Vignesh Subbian

Keyword(s):

Machine Learning ◽

Traumatic Brain Injury ◽

Brain Injury ◽

Subspace Clustering ◽

Machine Learning Techniques ◽

Phase Iii ◽

Physiological Data ◽

Clustering Methods ◽

Data Set ◽

Health Technologies

BACKGROUND With advances in digital health technologies and proliferation of biomedical data in recent years, applications of machine learning in health care and medicine have gained considerable attention. While inpatient settings are equipped to generate rich clinical data from patients, there is a dearth of actionable information that can be used for pursuing secondary research for specific clinical conditions. OBJECTIVE This study focused on applying unsupervised machine learning techniques for traumatic brain injury (TBI), which is the leading cause of death and disability among children and adults aged less than 44 years. Specifically, we present a case study to demonstrate the feasibility and applicability of subspace clustering techniques for extracting patterns from data collected from TBI patients. METHODS Data for this study were obtained from the Progesterone for Traumatic Brain Injury, Experimental Clinical Treatment–Phase III (PROTECT III) trial, which included a cohort of 882 TBI patients. We applied subspace-clustering methods (density-based, cell-based, and clustering-oriented methods) to this data set and compared the performance of the different clustering methods. RESULTS The analyses showed the following three clusters of laboratory physiological data: (1) international normalized ratio (INR), (2) INR, chloride, and creatinine, and (3) hemoglobin and hematocrit. While all subclustering algorithms had a reasonable accuracy in classifying patients by mortality status, the density-based algorithm had a higher F1 score and coverage. CONCLUSIONS Clustering approaches serve as an important step for phenotype definition and validation in clinical domains such as TBI, where patient and injury heterogeneity are among the major reasons for failure of clinical trials. The results from this study provide a foundation to develop scalable clustering algorithms for further research and validation.

Download Full-text

Acute traumatic brain injury presentations and the use of head CT scans in the Emergency Department

10.21203/rs.3.rs-51284/v1 ◽

2020 ◽

Author(s):

Alice Rogan ◽

Vimal Patel ◽

Jane Birdling ◽

Harnah Simmonds ◽

Jessica Lockett ◽

...

Keyword(s):

Traumatic Brain Injury ◽

Brain Injury ◽

Intracranial Injury ◽

Nice Guidelines ◽

Head Ct ◽

Ct Head ◽

Acute Traumatic Brain Injury ◽

Observational Design ◽

Clinically Significant ◽

Significant Injury

Abstract Objective: The use of CT head scanning for traumatic brain injury (TBI) is a vital diagnostic tool, guided by risk stratification tools. This study aims to review the use of CT head scans for TBI in two Australasian Emergency Departments (ED) in New Zealand.Methods: Retrospective observational design of patients referred for head CT from ED to exclude a significant intracranial injury between 1st September 2018 and 31st August 2019. Clinical data were collected regarding presenting patterns, identification of injuries on CT scan and adherence to CT guidelines.Results: Out of 425 cases reviewed, a clinically significant injury was identified in 41 (10%) patients. Patients who reported loss (32% vs 20% p < 0.05) or possible loss of consciousness (34% vs 22% p < 0.05) and had GCS < 13 (17% vs 8%, p < 0.05) or focal neurology (10% vs 3%, p < 0.05) were more likely to have a significantly intracranial injury on CT. Interestingly, 17 (41%) patients with significant injury were GCS 15 with no focal neurology. NICE guidelines were adhered to in 364 (86%) patients. In the 14% of cases that did not meet guideline criteria, all CT head scans were negative.Conclusion: CT head scans are a valuable tool in TBI and guidelines successfully identify those with significant intracranial injuries. However, the rate of significant injury for the total population requiring head CT remains low, with over 90% of head CTs in the population normal, despite high guideline compliance perhaps identifying a role for novel objective tests in ED guidelines internationally.

Download Full-text

S100B outperforms clinical decision rules for the identification of intracranial injury on head CT scan after mild traumatic brain injury

Brain Injury ◽

10.1080/02699052.2020.1725123 ◽

2020 ◽

Vol 34 (3) ◽

pp. 407-414 ◽

Cited By ~ 1

Author(s):

Courtney Marie Cora Jones ◽

Christopher Harmon ◽

Molly McCann ◽

Holly Gunyan ◽

Jeffrey J. Bazarian

Keyword(s):

Traumatic Brain Injury ◽

Brain Injury ◽

Ct Scan ◽

Mild Traumatic Brain Injury ◽

Decision Rules ◽

Clinical Decision ◽

Intracranial Injury ◽

Clinical Decision Rules ◽

Head Ct

Download Full-text

Predictive Performance of Traumatic Brain Injury Biomarkers in High-Risk Elderly Patients

The Journal of Applied Laboratory Medicine ◽

10.1093/jalm.2019.031393 ◽

2019 ◽

Vol 5 (1) ◽

pp. 91-100 ◽

Cited By ~ 2

Author(s):

Matthew D Ward ◽

Art Weber ◽

VeRonika D Merrill ◽

Robert D Welch ◽

Jeffrey J Bazarian ◽

...

Keyword(s):

Traumatic Brain Injury ◽

Brain Injury ◽

High Risk ◽

Elderly Patients ◽

Predictive Performance ◽

Study Data ◽

Intracranial Injury ◽

Younger Patients ◽

Significant Difference ◽

Patient Variables

Abstract Background Serum glial fibrillary acidic protein (GFAP) and ubiquitin carboxyl-terminal esterase L1 (UCH-L1) have recently received US Food and Drug Administration approval for prediction of abnormal computed tomography (CT) in mild traumatic brain injury patients (mTBI). However, their performance in elderly patients has not been characterized. Methods We performed a posthoc analysis using the A Prospective Clinical Evaluation of Biomarkers of Traumatic Brain Injury (ALERT-TBI) study data. Previously recorded patient variables and serum values of GFAP and UCH-L1 from mTBI patients were partitioned at 65 years of age (herein referred to as ≥65, high-risk; <65, low-risk). We sought to assess the influence of age on predictive performance, sensitivity, and negative predictive value (NPV) of serum UCH-L1 and GFAP to predict intracranial injury by CT. Results Elderly mTBI patients constituted 25.7% of the patient cohort (n = 504/1959). Sensitivity and NPV of GFAP/UCH-L1 were 100%, with no significant difference from younger patients (P = 0.5525 and P > 0.9999, respectively). Specificity was significantly lower in elderly patients (0.131 vs 0.442; P < 0.0001) and decreased stepwise with older age. Compared to younger patients, elderly mTBI patients without abnormal (i.e., normal) CT findings also had a significantly higher GFAP (38.6 vs 16.2 pg/mL; P < 0.0001) and UCH-L1 (347.4 vs 232.1 pg/mL; P < 0.0001). Conclusions Sensitivity and NPV to predict intracranial injury by CT was nearly identical between younger and elderly mTBI patients. Decrements in specificity and increased serum values suggest that special deference may be warranted for elderly patients.

Download Full-text

Predicting need for hospital admission in patients with traumatic brain injury or skull fractures identified on CT imaging: a machine learning approach

Emergency Medicine Journal ◽

10.1136/emermed-2020-210776 ◽

2021 ◽

pp. emermed-2020-210776

Author(s):

Carl Marincowitz ◽

Lewis Paton ◽

Fiona Lecky ◽

Paul Tiffin

Keyword(s):

Machine Learning ◽

Traumatic Brain Injury ◽

Brain Injury ◽

Hospital Admission ◽

Decision Trees ◽

Decision Rule ◽

Predictive Value ◽

Learning Approach ◽

Skull Fractures ◽

Machine Learning Approach

BackgroundPatients with mild traumatic brain injury on CT scan are routinely admitted for inpatient observation. Only a small proportion of patients require clinical intervention. We recently developed a decision rule using traditional statistical techniques that found neurologically intact patients with isolated simple skull fractures or single bleeds <5 mm with no preinjury antiplatelet or anticoagulant use may be safely discharged from the emergency department. The decision rule achieved a sensitivity of 99.5% (95% CI 98.1% to 99.9%) and specificity of 7.4% (95% CI 6.0% to 9.1%) to clinical deterioration. We aimed to transparently report a machine learning approach to assess if predictive accuracy could be improved.MethodsWe used data from the same retrospective cohort of 1699 initial Glasgow Coma Scale (GCS) 13–15 patients with injuries identified by CT who presented to three English Major Trauma Centres between 2010 and 2017 as in our original study. We assessed the ability of machine learning to predict the same composite outcome measure of deterioration (indicating need for hospital admission). Predictive models were built using gradient boosted decision trees which consisted of an ensemble of decision trees to optimise model performance.ResultsThe final algorithm reported a mean positive predictive value of 29%, mean negative predictive value of 94%, mean area under the curve (C-statistic) of 0.75, mean sensitivity of 99% and mean specificity of 7%. As with logistic regression, GCS, severity and number of brain injuries were found to be important predictors of deterioration.ConclusionWe found no clear advantages over the traditional prediction methods, although the models were, effectively, developed using a smaller data set, due to the need to divide it into training, calibration and validation sets. Future research should focus on developing models that provide clear advantages over existing classical techniques in predicting outcomes in this population.

Download Full-text

Computed Tomography of the Head in Children with Mild Traumatic Brain Injury

The American Surgeon ◽

10.1177/000313481408000912 ◽

2014 ◽

Vol 80 (9) ◽

pp. 841-843 ◽

Cited By ~ 7

Author(s):

Esther Mihindu ◽

Indermeet Bhullar ◽

Joseph Tepas ◽

Andrew Kerwin

Keyword(s):

Computed Tomography ◽

Traumatic Brain Injury ◽

Brain Injury ◽

Negative Predictive Value ◽

Predictive Value ◽

Pediatric Emergency ◽

Research Network ◽

Head Ct ◽

Pediatric Emergency Care ◽

Head Computed Tomography

Pediatric Emergency Care Applied Research Network (PECARN) guidelines have a near 100 per cent negative predictive value for clinically important traumatic brain injury (ciTBI) in children with mild head injury (Glasgow Coma Score [GCS] 14 or 15). Our goal was to retrospectively apply their criteria to our database to determine the potential impact on the rates of unnecessary head computed tomography (CT) and ciTBI detection. The records of pediatric patients with GCS 14 to 15 that had a head CT for suspected TBI after blunt trauma from 2008 to 2010 were reviewed. Of 493 children, CT was negative in 447 (91%), but findings were present in 46 (9%). Applying PECARN recommendations, 178 (36%) met all six criteria but still underwent head CT; all were negative. The remaining 315 (64%) missed one or more PECARN criteria and underwent CT; only 46 (15%) had findings, and two (0.6%) required surgery. There were no false-negatives. The negative predictive value for ciTBI was 100 per cent. Observance of PECARN guidelines identifies children who do not require CT, increasing the yield of finding a ciTBI among those who cannot satisfy all six criteria.

Download Full-text

Acute traumatic brain injury and the use of head computed tomography scans in the emergency department

Trauma ◽

10.1177/14604086211023646 ◽

2021 ◽

pp. 146040862110236

Author(s):

Alice Rogan ◽

Vimal Patel ◽

Jane Birdling ◽

Jessica Lockett ◽

Harnah Simmonds ◽

...

Keyword(s):

Traumatic Brain Injury ◽

Brain Injury ◽

Clinical Excellence ◽

Intracranial Injury ◽

Adherence To Guidelines ◽

Nice Guidelines ◽

Head Ct ◽

Ct Head ◽

Acute Traumatic Brain Injury ◽

Head Computed Tomography

Introduction The use of CT head scanning for traumatic brain injury (TBI) is a vital diagnostic tool, guided by risk stratification tools. This study aims to review the use of CT head scans and adherence to guidelines for TBI in two New Zealand emergency departments (EDs). Methods Retrospective observational study of patients referred for head CT from EDs to exclude a significant intracranial injury between 1st September 2018 and 31st August 2019. Clinical data were collected regarding presenting patterns, identification of injuries on CT scan and adherence to National Institute of Clinical Excellence (NICE) CT head guidelines. Results Out of 425 included cases, 41 (10%) patients had an intracranial injury seen on their CT head scan. Patients who reported loss (32% vs 20%, p < 0.05) or possible loss of consciousness (34% vs 22%, p < 0.05) and had a Glasgow Coma Score (GCS) <13 (17% vs 8%, p < 0.05) or focal neurology (10% vs 3%, p < 0.05) were more likely to have an intracranial injury on CT. Interestingly, 17 (41%) patients with CT diagnosed injuries had a GCS 15 and no focal neurology. NICE guidelines were adhered to in 364 (86%) of CT requests. In the 14% of cases that did not meet guideline criteria, all CT head scans were negative. Conclusion CT head scans are a valuable tool in TBI, and guidelines successfully identify those with significant intracranial injuries. However, the rate of significant injury for the total population requiring head CT remains low, with over 90% of head CTs in the population normal, despite high guideline compliance, perhaps identifying a role for novel objective tests in ED guidelines internationally.

Download Full-text