Estimate Injury Risk Research Articles

Baseline vulnerable road user injury risk in multiple U.S. dense urban driving environments

ABSTRACT Objective Understanding and modeling baseline driving safety risk in dense urban areas represents a crucial starting point for automated driving system (ADS) safety impact analysis. The purpose of this study was to leverage naturalistic vulnerable road user (VRU) collision data to quantify collision rates, crash severity, and injury risk distributions in the absence of objective injury outcome data. Methods From over 500 million vehicle miles traveled, a total of 335 collision events involving VRUs were video verified and reconstructed (126 pedestrians, 144 cyclists, and 65 motorcyclists). Data consisted of anonymized video and sensor data (Global Positioning System and accelerometer) from vehicles equipped with Nexar dash cameras. Each event was qualitatively evaluated to assess the collision geometries and other extrinsic collision factors (e.g., time of day, roadway location, presence of occlusions). Previously published injury risk models were utilized to provide estimates of Maximum Abbreviated Injury Scale MAIS2+ and MAIS3+ injury potential. Collision severity distributions and aggregated injury risk estimates were compared. Results Vehicle speeds at the time of impact ranged from 0 to 39 mph, and VRU speeds ranged from 0 to 27 mph. In general, vehicles were traveling at lower speeds at the time of impact when turning in comparison to straight travel. Nearly all events (95%) were associated with MAIS3+ injury risk estimates below 10%. Collisions involving a potential occlusion or the vehicle responding to surprising behavior by the VRU were associated with higher estimates of injury risk than those without occlusion or with the vehicle initiating the conflict. Based on accumulated risk for each event, it can be estimated that approximately 55 persons could be moderately injured (MAIS2+) and approximately 6 persons could be seriously injured (MAIS3+). Conclusion These data indicate that responding to surprising VRU behavior, having visibility be potentially occluded, and vehicle travel behavior were associated with differences in collision speed and injury risk estimation. The specific comparisons made in this study are not intended to be comprehensive but serve as a starting point for considering baseline driving risk associated with VRU collisions in dense urban areas.

Injury Risk Estimation in Mandible-Related Surgical Procedures: A CBCT Study Based on Vital Interforaminal Anatomical Structures.

The mandibular interforaminal region has been considered safe for surgical procedures; nevertheless, the risk of injury to neurovascular structures, such as the mental foramen (MF) and its related structures (anterior loop [AL] and lingual foramina [LF]) should not be overlooked. The study aimed to evaluate the relative risk of injury to these structures during surgical procedures in the anterior region of the mandible based on cone-beam computed tomography (CBCT) scans. A retrospective cross-sectional and observational study was performed on 250 CBCTs from adults (18-69 years) with dentate jaws. Linear measurements of the MF, AL, and LF were evaluated to estimate the risk of injury to these structures during chin-related surgical procedures. The most frequent distance between the base of the mandible (BM) and MF was 8 mm (30.2%). In addition, 20.4% of the CTs had 6 mm from the vestibular cortical bone to the LF. The commonly found measurement from LF to the apex of the nearest tooth was 7 mm (24.0%); 64.2% of the CTs showed a 2-mm distance between the most distal point of the dental implant site to the most anterior point of the AL. Safety distances for genioplasty techniques (MF to mandible base > 6 mm, 96.6% [CI 95%, 95.0%-98.2%]) were observed. Considering the 5-mm cut-off point between the lower limit of a hypothetical bone graft and the chin, 65.4% (CI 95%, 58.9%-71.9%) of CTs were within this distance. Regarding the safety margin of 8 mm, 85.6% (CI 95%, 80.8%-90.4%) were up to this value. This study found safety margins for genioplasty and chin bone grafting surgical techniques that adopt a 5-mm cut-off point. Further similar studies assessing other surgical methods and employing larger samples from different geographical origins may contribute to this field of investigation.

Integrating Machine Learning in Clinical Decision Support Systems

This review article examines the role of machine learning (ML) in enhancing Clinical Decision Support Systems (CDSSs) within the modern healthcare landscape. Focusing on the integration of various ML algorithms, such as regression, random forest, and neural networks, the review aims to showcase their potential in advancing patient care. A rapid review methodology was utilized, involving a survey of recent articles from PubMed and Google Scholar on ML applications in healthcare. Key findings include the demonstration of ML's predictive power in patient outcomes, its ability to augment clinician knowledge, and the effectiveness of ensemble algorithmic approaches. The review highlights specific applications of diverse ML models, including moment kernel machines in predicting surgical outcomes, k-means clustering in simplifying disease phenotypes, and extreme gradient boosting in estimating injury risk. Emphasizing the potential of ML to tackle current healthcare challenges, the article highlights the critical role of ML in evolving CDSSs for improved clinical decision-making and patient care. This comprehensive review also addresses the challenges and limitations of integrating ML into healthcare systems, advocating for a collaborative approach to refine these systems for safety, efficacy, and equity.

Adolescent track and field injuries presenting to US emergency departments

ABSTRACT Objectives Track and field (T&F) is a highly popular sport for adolescents. The diversity of running, jumping, and throwing events within the sport can result in unique injury patterns for adolescent track and field participants. The purpose of this study was to estimate injury risk in adolescent T&F and describe the types of injuries resulting in ED visits, classified by T&F events. Methods Emergency department (ED) data from the National Electronic Injury Surveillance System were obtained for a 20-year period from 2000 through 2019. Cases involving 14- to 18-year-olds participating in T&F were classified by sex, case severity, involved body region, and the T&F event patients were engaged in at the time of their injury. National estimates and Injury rates were calculated using national high school T&F participation data. Longitudinal trends in ED visits were measured using linear regression. Rate ratios (RRs) were used to compare the risk and severity of ED visits by sex. Results 8,060 track and field related ED encounters were observed, representing an estimated 272,227 encounters nationally over the 20-year study period. The rate of ED encounters increased significantly over the study period (p < 0.001). Adolescent females exhibited a higher rate of ED encounters (RR: 1.23, 95% CI: 1.22–1.24), but a lower rate of hospital admissions (RR: 0.68, 95% CI: 0.64–0.73) compared to males. The lower extremity was the most commonly injured body region for most T&F events, but this differed for sprinting, high jumping, shot put, and javelin events. Most T&F events resulted in unique injury patterns characteristic of their athletic demands. Conclusions There was an increasing trend of ED visits from adolescent T&F participants throughout the 20-year study period with different injury patterns observed by sex and T&F event discipline.

Assessment of finite element human body and ATD models in estimating injury risk in far-side impacts using field-based injury risk

The objective of this study was to assess the ability of finite element human body models (FEHBMs) and Anthropometric Test Device (ATD) models to estimate occupant injury risk by comparing it with field-based injury risk in far-side impacts. The study used the Global Human Body Models Consortium midsize male (M50-OS+B) and small female (F05-OS+B) simplified occupant models with a modular detailed brain, and the ES-2Re and SID-IIs ATD models in the simulated far-side crashes. A design of experiments (DOE) with a total of 252 simulations was conducted by varying lateral ΔV (10-50kph; 5kph increments), the principal direction of force (PDOF 50°, 60°, 65°, 70°, 75°, 80°, 90°), and occupant models. Models were gravity-settled and belted into a simplified vehicle model (SVM) modified for far-side impact simulations. Acceleration pulses and vehicle intrusion profiles used for the DOE were generated by impacting a 2012 Camry vehicle model with a mobile deformable barrier model across the 7 PDOFs and 9 lateral ΔV’s in the DOE for a total of 63 additional simulations. Injury risks were estimated for the head, chest, lower extremity, pelvis (AIS 2+; AIS 3+), and abdomen (AIS 3+) using logistic regression models. Combined AIS 3+ injury risk for each occupant was calculated using AIS 3+ injury risk estimations for the head, chest, abdomen, and lower extremities. The injury risk calculated using computational models was compared with field-based injury risk derived from NASS-CDS by calculating their correlation coefficient. The field-based injury risk was calculated using risk curves that were created based on real-world crash data in a previous study (Hostetler et al., 2020). Occupant age (40 years), seatbelt use (belted occupant), collision deformation classification, lateral ΔV, and PDOF of the crash event were used in these curves to estimate field injury risk. Large differences in the kinematics were observed between HBM and ATD models. ATD models tended to overestimate risk in almost every case whereas HBMs yielded better risk estimates overall. Chest and lower extremity risks were the least correlated with field injury risk estimates. The overall risk of AIS 3+ injury risk was the strongest comparison to the field data-based risk curves. The HBMs were still not able to capture all the variance but future studies can be carried out that are focused on investigating their shortfalls and improving them to estimate injury risk closer to field injury risk in far-side crashes.

An estimation of the endoscopist's musculoskeletal injury risk for right and left lateral decubitus positions during colonoscopy: a field-based ergonomic study

BackgroundColonoscopy exposes endoscopists to awkward postures and prolonged forces, which increases their risk of musculoskeletal injury. Patient positioning has a significant impact on the ergonomics of colonoscopy. Recent trials have found the right lateral decubitus position is associated with quicker insertion, higher adenoma detection rates, and greater patient comfort compared to the left lateral decubitus position. However, this patient position is perceived as more strenuous by endoscopists.MethodsNineteen endoscopists were observed performing colonoscopies during a series of four-hour endoscopy clinics. Durations of each patient position (right lateral decubitus, left lateral decubitus, prone, and supine) were recorded for all observed procedures (n = 64). Endoscopist injury risk was estimated by a trained researcher for the first and last colonoscopies of the shifts (n = 34) using Rapid Upper Limb Assessment (RULA), an observational ergonomic tool that estimates risk of musculoskeletal injury by scoring postures of the upper body and factors such as muscle use, force, and load. The total RULA scores were compared with a Wilcoxon Signed-Rank test for patient position (right and left lateral decubitus) and time (first and last procedures) with significance taken at p < 0.05. Endoscopist preferences were also surveyed.ResultsThe right lateral decubitus position was associated with significantly higher RULA scores than the left lateral decubitus position (median 5 vs. 3, p < 0.001). RULA scores were not significantly different between the first and last procedures of the shifts (median 5 vs. 5, p = 0.816). 89% of endoscopists preferred the left lateral decubitus position, primarily due to superior ergonomics and comfort.ConclusionRULA scores indicate an increased risk of musculoskeletal injury in both patient positions, with greater risk in the right lateral decubitus position.

Changes in knee flexion-extension performance and hamstring-to-quadriceps ratio during a fatiguing isokinetic protocol in male professional futsal players

ABSTRACT Knee flexion (KF) and extension (KE) performance and hamstring-to-quadriceps (H:Q) ratio are used to estimate injury risk and readiness to return to sport. However, the traditional measures (i.e. non-fatigued state and by peak torque [PT]) may not reflect the thigh and knee injury conditions (i.e. high-speed actions and fatigue presence). We verified the effects of a high-speed isokinetic fatigue protocol (30 consecutive KF and KE concentric contractions at 300°/s) on KF and KE performance and H:Q ratios assessed through different parameters (i.e. PT, work, average and peak power) in male professional futsal players (n = 17). A significant (p < 0.01) time × muscle group interaction was verified. There were reductions (p < 0.01) in KF by PT (84.79 ± 15.05 to 50.29 ± 15.29 Nm), work (87.28 ± 14.46 to 39.00 ± 13.83 Nm), average power (227.81 ± 39.31 to 104.75 ± 31.24 W) and peak power (383.40 ± 59.90 to 191.39 ± 48.35 W). Similarly, KE performance decreased (p < 0.01) for all parameters (PT: 131.36 ± 16.05 to 81.61 ± 9.15 Nm; work: 148.83 ± 20.88 to 90.72 ± 11.89 Nm; average power: 148.83 ± 20.88 to 90.72 ± 11.89 W; peak power: 664.74 ± 85.32 to 398.38 ± 42.42 W). KF (−55.18 ± 13.91% to −40.94 ± 12.66%) showed a higher percentage decrease than KE (−40.46 ± 7.38% to −37.45 ± 6.61%). Finally, reductions (p < 0.01) in H:Q ratio by work, average and peak power but not by PT (p = 0.25) were verified. The tested methods can provide additional information about the athlete’s condition.

Relationship between a daily injury risk estimation feedback (I-REF) based on machine learning techniques and actual injury risk in athletics (track and field): protocol for a prospective cohort study over an athletics season

IntroductionTwo-thirds of athletes (65%) have at least one injury complaint leading to participation restriction (ICPR) in athletics (track and field) during one season. The emerging practice of medicine and public...

Machine-learning-based head impact subtyping based on the spectral densities of the measurable head kinematics

BackgroundTraumatic brain injury can be caused by head impacts, but many brain injury risk estimation models are not equally accurate across the variety of impacts that patients may undergo, and the characteristics of different types of impacts are not well studied. We investigated the spectral characteristics of different head impact types with kinematics classification. MethodsData were analyzed from 3262 head impacts from lab reconstruction, American football, mixed martial arts, and publicly available car crash data. A random forest classifier with spectral densities of linear acceleration and angular velocity was built to classify head impact types (e.g., football, car crash, mixed martial arts). To test the classifier robustness, another 271 lab-reconstructed impacts were obtained from 5 other instrumented mouthguards. Finally, with the classifier, type-specific, nearest-neighbor regression models were built for brain strain. ResultsThe classifier reached a median accuracy of 96% over 1000 random partitions of training and test sets. The most important features in the classification included both low- and high-frequency features, both linear acceleration features and angular velocity features. Different head impact types had different distributions of spectral densities in low- and high-frequency ranges (e.g., the spectral densities of mixed martial arts impacts were higher in the high-frequency range than in the low-frequency range). The type-specific regression showed a generally higher R2 value than baseline models without classification. ConclusionThe machine-learning-based classifier enables a better understanding of the impact kinematics spectral density in different sports, and it can be applied to evaluate the quality of impact-simulation systems and on-field data augmentation.

How Accurately Can Wearable Sensors Assess Low Back Disorder Risks during Material Handling? Exploring the Fundamental Capabilities and Limitations of Different Sensor Signals.

Low back disorders (LBDs) are a leading occupational health issue. Wearable sensors, such as inertial measurement units (IMUs) and/or pressure insoles, could automate and enhance the ergonomic assessment of LBD risks during material handling. However, much remains unknown about which sensor signals to use and how accurately sensors can estimate injury risk. The objective of this study was to address two open questions: (1) How accurately can we estimate LBD risk when combining trunk motion and under-the-foot force data (simulating a trunk IMU and pressure insoles used together)? (2) How much greater is this risk assessment accuracy than using only trunk motion (simulating a trunk IMU alone)? We developed a data-driven simulation using randomized lifting tasks, machine learning algorithms, and a validated ergonomic assessment tool. We found that trunk motion-based estimates of LBD risk were not strongly correlated (r range: 0.20-0.56) with ground truth LBD risk, but adding under-the-foot force data yielded strongly correlated LBD risk estimates (r range: 0.93-0.98). These results raise questions about the adequacy of a single IMU for LBD risk assessment during material handling but suggest that combining an IMU on the trunk and pressure insoles with trained algorithms may be able to accurately assess risks.

Personalized Injury Reduction Strategies in Sports Medicine: Lessons Learned from Advances in Breast Cancer Treatment: A Clinical Commentary.

Injury rates across sport have risen over the past twenty years, despite increased efforts in training and injury prevention. The rise in injury rates suggest that current approaches to estimating injury risk and risk management are not effective. One factor limiting progress is the inconsistency in screening, risk assessment, and risk management strategies to guide injury mitigation approaches. How can sports physical therapists identify and apply lessons learned from other healthcare fields to improve athlete injury risk and risk management strategies? Breast cancer mortality has consistently decreased over the last 30 years, largely attributed to advances in personalizing the prevention and treatment strategies which include modifiable and non-modifiable factors when assessing risk, the transition to personalized medicine, and the systematic approach used to investigate individual risk factors. Three critical phases have facilitated the identification and importance of individual risk factors and developing targeted, personalized strategies for breast cancer risk including: 1) Establishing the potential relationship between factors and outcomes; 2) Prospectively investigate the strength and direction of the relationship; 3) Investigating if intervening on identified factors alters prognosis. Applying lessons learned from other healthcare fields could improve shared decision making between the clinician and athlete concerning risk assessment and management. Examples include calculating only non-modifiable risk, creating individualized screening schedules based on risk assessment, or calculating the influence of each intervention on the athlete's injury risk. A systematic approach to identify and intervene on risk is needed to improve athlete outcomes.

Does changing the Functional Movement Screen composite score threshold influence injury risk estimation in junior Australian football players?

ABSTRACT This study aimed to identify whether a revised lower Functional Movement Screen (FMS) composite score threshold would be associated with a greater injury risk for junior athletes than the common threshold of≤14. This prospective cohort study included tracking of 809 elite junior male Australian football players for injuries that resulted in a missed game. All athletes completed pre-season FMS testing and a 12-month self-reported retrospective injury questionnaire. Analyses examined the relationship between composite score thresholds of≤14, ≤13, and≤12 and the risk of injury. The relationship between prospective injury and the common composite threshold score of ≤ 14 was dependent on the presence of a recent injury history (relative risk [RR] = 1.45, p = 0.004) in comparison to no recent injury history (RR = 0.98, p = 0.887). Scoring≤12 in the presence of a recent injury history had the greatest diagnostic accuracy but only a trivial increase in injury risk (RR = 1.59, p = 0.001, sensitivity = 0.35, specificity = 0.80, negative and positive likelihood ratios = 0.81 and 1.75). Whilst some small statistical relationships existed between prospective injury and the FMS composite score thresholds, all three thresholds were not associated with a clinically meaningful relationship with prospective injury and were no more effective than retrospective injury for determining athletes at risk of injury.

Assessment of Finite Element Human Body and ATD Models in Estimating Injury Risk in Far-Side Impacts Using Field-Based Injury Risk

Assessment of Finite Element Human Body and ATD Models in Estimating Injury Risk in Far-Side Impacts Using Field-Based Injury Risk

Principal-stretch-based constitutive neural networks autonomously discover a subclass of Ogden models for human brain tissue

The soft tissue of the brain deforms in response to external stimuli, which can lead to traumatic brain injury. Constitutive models relate the stress in the brain to its deformation and accurate constitutive modeling is critical in finite element simulations to estimate injury risk. Traditionally, researchers first choose a constitutive model and then fit the model parameters to tension, compression, or shear experiments. In contrast, constitutive artificial neural networks enable automated model discovery without having to choose a specific model before learning the model parameters. Here we reverse engineer a constitutive artificial neural network that uses the principal stretches, raised to a wide range of exponential powers, as activation functions. Upon training, the network autonomously discovers a subclass of models with multiple Ogden terms that outperform popular constitutive models including the neo Hooke, Blatz Ko, and Mooney Rivlin models. While invariant-based networks fail to capture the pronounced tension–compression asymmetry of brain tissue, our principal-stretch-based network can simultaneously explain tension, compression, and shear data for the cortex, basal ganglia, corona radiata, and corpus callosum. Without fixing the number of terms a priori, our model self-selects the best subset of terms out of more than a million possible combinations, while simultaneously discovering the best model parameters and best experiment to train itself. Eliminating user-guided model selection has the potential to induce a paradigm shift in soft tissue modeling and democratize brain injury simulations. Our source code, data, and examples are available at https://github.com/LivingMatterLab/CANN.Statement of Significance: Understanding the constitutive response of the brain is critical to estimate brain injury risk, design protective devices, and predict surgical intervention. The current gold standard in constitutive modeling, first choosing a constitutive model and then fitting its parameters to data, is largely biased by user experience and personal preference. Constitutive artificial neural networks eliminate the need for user-guided model selection and enable automated model discovery. Here we reverse-engineer a constitutive artificial neural network with custom-designed activation functions from principal stretches raised to a wide range of exponential powers. When trained with data from human gray and white matter tissue, our network autonomously discovers a subclass of models with multiple Ogden terms that outperform popular constitutive models including the neo Hooke, Blatz Ko, and Mooney Rivlin models. While these classical invariant-based networks fail to capture the pronounced tension-compression asymmetry of brain tissue, our discovered principal-stretch-based models can simultaneously explain tension, compression, and shear data from the human cortex, basal ganglia, corona radiata, and corpus callosum. Without fixing the number of model terms a priori, our network self-selects the best subset of terms out of more than a million possible combinations, while simultaneously discovering the best model parameters, for example, shear moduli of 1.47kPa, 0.68kPa, 0.69kPa, and 0.29kPa for these four brain regions. Our findings are significant in that they eliminate user-guided model selection and have the potential to make brain modeling more accessible to a wide group of scientists with diverse training and backgrounds towards the ultimate goal to democratize human brain simulations.

Feasibility and validity of a single camera CNN driven musculoskeletal model for muscle force estimation during upper extremity strength exercises: Proof-of-concept.

Muscle force analysis can be essential for injury risk estimation and performance enhancement in sports like strength training. However, current methods to record muscle forces including electromyography or marker-based measurements combined with a musculoskeletal model are time-consuming and restrict the athlete's natural movement due to equipment attachment. Therefore, the feasibility and validity of a more applicable method, requiring only a single standard camera for the recordings, combined with a deep-learning model and musculoskeletal model is evaluated in the present study during upper-body strength exercises performed by five athletes. Comparison of muscle forces obtained by the single camera driven model against those obtained from a state-of-the art marker-based driven musculoskeletal model revealed strong to excellent correlations and reasonable RMSD's of 0.4–2.1% of the maximum force (Fmax) for prime movers, and weak to strong correlations with RMSD's of 0.4–0.7% Fmax for stabilizing and secondary muscles. In conclusion, a single camera deep-learning driven model is a feasible method for muscle force analysis in a strength training environment, and first validity results show reasonable accuracies, especially for prime mover muscle forces. However, it is evident that future research should investigate this method for a larger sample size and for multiple exercises.

Motorcycle curve assist: A novel approach based on active speed control for crash injury reduction

Objective Safely negotiating curves with a powered-two-wheeler (PTW) requires a high level of skill, and a significant proportion of PTW crashes have a curve involvement. This study aimed to estimate the applicability, potential benefits and feasibility of novel Motorcycle Curve Assist (MCA). The system is designed to operate an emergency control of the speed of a motorcycle approaching a bend at an inappropriate speed. Methods First, the MCA system intervention was defined. Second, the applicability of the system and an estimate of its potential benefits was performed based on a PTW crash database. Motorcyclists’ injury risk estimates, MCA working parameters and timing of intervention were employed to estimate the potential injury reduction of applicable crash types. Third, a field test campaign involving 29 common riders as participants was conducted to investigate the real-world applicability and acceptability among end-users of the system deployment in one relevant riding condition adopting a range of parameters of intervention. Results In the crash database, 23% of cases had curve involvement and after detailed analysis, 14% resulted to be suitable for MCA (60% of cases with curve involvement). The potential relative injury risk reduction considering only the benefits due to crash speed reduction ranged from 3–9% for MAIS2+ to 9–27% for MAIS3+ injuries. Field tests were performed in corners approached at an average speed of 28.7 km/h and an average lean angle of 20°. The system provided a mean deceleration of 0.33 g reached with a fade-in jerk of 1.73 g/s, for an average total duration of 0.59 s. For the field test component, participants reported good controllability of the system, with no incipient loss of control recorded nor reported by participants. Conclusions The proposed approach for MCA implementation showed considerable potential benefits in terms of injury reduction. The intervention with the defined working parameters was considered feasible by a sample of end-users. When integrated with an intervention logic capable of predicting emergency situations while approaching curves, MCA will be a technology capable of assisting PTW riders in conditions where other available active safety systems do not.

Development and Validation of a Personalized Model With Transfer Learning for Acute Kidney Injury Risk Estimation Using Electronic Health Records

Acute kidney injury (AKI) is a heterogeneous syndrome prevalent among hospitalized patients. Personalized risk estimation and risk factor identification may allow effective intervention and improved outcomes. To develop and validate personalized AKI risk estimation models using electronic health records (EHRs), examine whether personalized models were beneficial in comparison with global and subgroup models, and assess the heterogeneity of risk factors and their outcomes in different subpopulations. This diagnostic study analyzed EHR data from 1 tertiary care hospital and used machine learning and logistic regression to develop and validate global, subgroup, and personalized risk estimation models. Transfer learning was implemented to enhance the personalized model. Predictor outcomes across subpopulations were analyzed, and metaregression was used to explore predictor interactions. Adults who were hospitalized for 2 or more days from November 1, 2007, to December 31, 2016, were included in the analysis. Patients with moderate or severe kidney dysfunction at admission were excluded. Data were analyzed between August 28, 2019, and May 8, 2022. Clinical and laboratory variables in the EHR. The main outcome was AKI of any severity, and AKI was defined using the Kidney Disease: Improving Global Outcomes serum creatinine criteria. Performance of the models was measured with area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve, and calibration. The study cohort comprised 76 957 inpatient encounters. Patients had a mean (SD) age of 55.5 (17.4) years and included 42 159 men (54.8%). The personalized model with transfer learning outperformed the global model for AKI estimation in terms of AUROC among general inpatients (0.78 [95% CI, 0.77-0.79] vs 0.76 [95% CI, 0.75-0.76]; P < .001) and across the high-risk subgroups (0.79 [95% CI, 0.78-0.80] vs 0.75 [95% CI, 0.74-0.77]; P < .001) and low-risk subgroups (0.74 [95% CI, 0.73-0.75] vs 0.71 [95% CI, 0.70-0.72]; P < .001). The AUROC improvement reached 0.13 for the high-risk subgroups, such as those undergoing liver transplant and cardiac surgery. Moreover, the personalized model with transfer learning performed better than or comparably with the best published models in well-studied AKI subgroups. Predictor outcomes varied significantly between patients, and interaction analysis uncovered modifiers of the predictor outcomes. Results of this study demonstrated that a personalized modeling with transfer learning is an improved AKI risk estimation approach that can be used across diverse patient subgroups. Risk factor heterogeneity and interactions at the individual level highlighted the need for agile, personalized care.

Modeling Injury Risk From Multiple-Impulse, Area-Distributed Flash-bangs Using an Uncertainty Bounding Approach to Dose Accumulation.

Modeling of injury risk from nonlethal weapons including flash-bangs is a critical step in the design, acquisition, and application of such devices for military purposes. One flash-bang design concept currently being developed involves multiple, area-distributed flash-bangs. It is particularly difficult to model the variation inherent in operational settings employing such devices due to the randomness of flash-bang detonation positioning relative to targets. The problem is exacerbated by uncertainty related to changes in the mechanical properties of auditory system tissues and contraction of muscles in the middle ear (the acoustic reflex), which can both immediately follow impulse-noise exposure. In this article, we demonstrate a methodology to quantify uncertainty in injury risk estimation related to exposure to multiple area-distributed flash-bang impulses in short periods of time and analyze the effects of factors such as the number of impulses, their spatial distribution, and the uncertainties in their parameters on estimated injury risk. We conducted Monte Carlo simulations of dispersion and timing of a mortar-and-submunition flash-bang device that distributes submunitions over an area, using the Auditory 4.5 model developed by L3 Applied Technologies to estimate the risk of hearing loss (permanent threshold shift) in an exposure area. We bound injury risk estimates by applying limiting assumptions for dose accumulation rules applied to short inter-pulse intervals and varied impulse-noise-intensity exposure characteristic of multi-impulse flash-bangs. The upper bound of risk assumes no trading of risk between the number of impulses and intensity of individual impulses, while the lower bound assumes a perfectly protective acoustic reflex. In general, the risk to individuals standing in the most hazardous zone of the simulation is quite sensitive to the pattern of submunitions, relative to the sensitivity for those standing farther from that zone. Larger mortar burst radii (distributing submunitions over a wider area) reduce expected peak risk, while increasing the number of submunitions, the intensity of individual impulses, or the uncertainty in impulse intensity increases expected risk. We find that injury risk calculations must factor in device output variation because the injury risk curve in the flash-bang dose regime is asymmetric. We also find that increased numbers of submunitions increase the peak risk in an area more rapidly than scene-averaged risk and that the uncertainty related to dose accumulation in the acoustic reflex regime can be substantial for large numbers of submunitions and should not be ignored. This work provides a methodology for exploring both the role of device parameters and the choice of dose accumulation rule in estimating the risk of significant injury and associated uncertainty for multi-impulse, area-distributed flash-bang exposures. This analysis can inform decisions about the design of flash-bangs and training for their operational usage. The methodology can be extended to other device designs or deployment concepts to generate risk maps and injury risk uncertainty ranges. This work does not account for additional injury types beyond permanent threshold shift that may occur as a result of flash-bang exposure. A useful extension of this work would be similar work connecting design and operational parameters to human effectiveness.

Crash characteristics and injury risk of adult car occupants in near-side impacts

Objective In the US, 27% of car occupant fatalities occur in side-impacts. Near-side impacts cause more serious injuries than far-side impacts. Car occupant safety overall has improved, but rear-seat occupant protection lags behind front-seat protection. The current study had two aims: first, to describe crash characteristics and injury outcomes for near-side rear-seated restrained occupants before and after side-impact regulations (FMVSS-214) were mandated; and secondly, to estimate injury risks in near-side impact depending on seating position, delta-v, and other predictors. Methods The National Automotive Sampling System – Crashworthiness Data System (NASS – CDS) database for 1995 to 2015 was analyzed. The descriptive statistics focused on rear-seated restrained adults, comparing old (<1997) and new (≥1997) vehicles exposed to near-side impacts. Weighted logistic regression models including all front and rear-seat occupants were used to predict the occurrence of at least moderate injury in near-side impacts. Results Rear-seated restrained adults in new vehicles were exposed to higher delta-v and intrusion levels compared to old vehicles. Moreover, injuries to rear-seat occupants in new vehicles were more distributed across body regions, compared to old vehicles (mainly thorax and head). The risk of sustaining injuries increased significantly with delta-v and occupant age, when not using a seatbelt, when seated in a passenger car and when hit in the passenger compartment. The risk increased by trend, but not significantly, when seated in a newer vehicle, when seated in the rear-seat and when the crash opponent was an SUV or van. Conclusions Trends and risk factors for at least moderate injuries in near-side impacts were identified to guide future directions for occupant protection in side-impacts and its assessments.

The agreement and repeatability of measurements of ankle joint dorsiflexion and poplietal angle in healthy adolescents

BackgroundThe intra-rater repeatability and inter-rater agreement of orthopaedics measurements are important for estimating injury risk and appropriate treatment. In clinical practice, it is often unavoidable to trust the measurements of other health professionals.MethodsThis study tested the agreement and repeatability of measurements of the dorsiflexion of the foot, dorsiflexion with 90-degrees knee flexion, and popliteal angle test in healthy adolescents performed twice by three raters differing in clinical experience. Three raters, i.e., an orthopaedics specialist (16 years of experience), a resident medical doctor in orthopaedics (4 years of experience), and a physiotherapy student (1 year of experience) measured the ankle joint dorsiflexion and the popliteal angle in 142 healthy adolescent subjects.ResultsThe student outperformed more experienced raters by displaying good repeatability for all the evaluated parameters. The orthopaedics specialist failed to replicate the measurements of the left ankle joint passive dorsiflexion and the left popliteal angle. The medical resident in orthopaedics displayed a lack of repeatability in evaluating the right ankle joint dorsiflexion with the knee joint bent. Kendall’s W value for all parameters ranged 0.66–0.78, indicating a good inter-rater agreement.ConclusionsThe study highlights that measurements of the ankle joint dorsiflexion and popliteal angle test by different health professionals can generally be trusted. It indicates that novice health professionals could potentially evaluate such parameters in healthy subjects without a quality loss.