Objectives: To develop a predictive model of time from rehabilitation admission to emergence from PTA.

Background: The hallmark of msTBI is a prolonged period of retrograde and anterograde amnesia often coupled with behavioral disturbances termed Post-Traumatic Amnesia [PTA]. Patients in PTA have difficulty with orientation, day to day memory, safety awareness and carry over of treatment paradigms. Patients with ongoing PTA are very difficult to discharge back into their community safely given their supervision needs. Additionally, it is known that duration of PTA following injury is the best indicator of injury severity and long- term resource utilization.


Methods: Of 439 TBI patients admitted to the rehabilitation center 201 were still in PTA upon admission. The primary outcome was time from rehabilitation admission to emergence from PTA in the rehabilitation center as determined by an Orientation-Log [OLOG] score ≥ 25 on two consecutive days administered by unit-based Speech and Language Pathologist. Demographic variables, injury type, duration of PTA at rehabilitation center admission, and intake ‘Self Care’ and ‘Motor’ scores on the Centers for Medicare and Medicaid Services [CMS] Care Tool were considered in relation to time from rehabilitation admission to emergence from PTA. Multivariable Cox regression modeling of time to clear PTA included a complex interaction term of the Care Tool scores and PTA duration measured at admission generated from recursive partitioning investigation using classification and regression trees [CART] methodology.


Results: CART analysis generated the categorization of Care Tool scores and duration of PTA into four groups that most significantly discriminate the probability of emergence from PTA. In patients with a ‘Self Care’ score > 17, 50% will clear their PTA in the first seven days of rehabilitation. For patients with ‘Self Care’ score < 17 and who had more than 45 days in PTA upon admission to rehabilitation, 58% clear PTA after 42 days.


Conclusions: Care Tool ‘Self Care’ Scores and PTA duration upon admission to a rehabilitation unit have a strong prognostic value for estimating time until PTA clearance during rehabilitation. Total duration of PTA is a central aspect of moderate to severe TBI both in terms of assessing injury severity and describing prognosis but also in planning a rehabilitation approach to the patient and their injury. Accurately estimating duration of PTA can help the clinician establish an acute rehabilitation plan and allow them to counsel the patient and family about likely outcomes.

Introduction: Accurate prognosis in TBI is essential for effective treatment and rehabilitation planning, anticipating potential complications, and setting realistic expectations for both patients and their families. A precise prognosis also helps optimize the allocation of medical resources and improve long-term outcomes. A substantial amount of research has focused on predicting patient survival in the first hours to weeks following TBI. However, few studies have examined prognosis during the chronic stages of the condition, specifically in a neurorehabilitation setting where patients are stable, and the available information and expectations differ. The aim of this study was to predict the functional independence of TBI patients one year after admission to a neurorehabilitation center.

Methods: The demographic and clinical data of 209 TBI patients, admitted to a long-term neurorehabilitation center and not in a post-traumatic confusional state (PTCS), were retrospectively analyzed. Their motor and cognitive functions, communication abilities, emotional state, behavior, and functional independence -assessed with the Functional Independence Measure + Functional Assessment Measure (FIM+FAM)- were assessed at admission, and again at 6 and 12 months post-admission. Six different machine learning models aiming to predict the FIM+FAM scores one year post-admission, were trained incrementally, where each model built upon the information from the previous one, adding new, specific data. The additional information incorporated by the models included: (i) demographic data; (ii) injury-related data; (iii) global independence; (iv) clinical condition at admission to the neurorehabilitation program; (v) clinical condition at 6 months post-admission; and (vi) progression from admission to the 6-month follow-up.

Results: Random forest models achieved the highest performance. The models had increasing accuracy at predicting the FIM+FAM scores, with an explained variability (R-squared) and prediction error (RMSE) of: (i) 14.3% and 23.3; (ii) 23.8% and 22.0; (iii) 47.7% and 18.2; (iv) 62.1% and 15.5; (v) 90.6 and 7.7; and (vi) 91.1% and 7.5. After hyperparameter optimization, the final model had a R-squared of 93.1% and RMSE of 6.6. Consequently, the model was able to predict the FIM+FAM scores with an average prediction error of 6.6 over 210 points. The FIM+FAM score at 6 months post-admission and the progression in the FIM+FAM from admission to 6 months follow-up were the most relevant predictors.

Conclusions: The data used in predictive studies during the acute stage has very limited value for predicting outcomes in the neurorehabilitation setting. In contrast, functional independence at 6 months post-admission, and the progress made from admission to this point (among other variables) can accurately predict functional independence one year after admission.

Acknowledgments: This work was supported by Conselleria d’Innovació, Universitats, Ciència i Societat Digital of Generalitat Valenciana (CIDEXG/2022/15), Ministerio de Ciencia e Innovación (PID2022-141498OA-I00), Ministerio de Economía, Comercio y Empresa (CPP2022-009580) and Fundació la Marató de la TV3 (60/2023).

Introduction: Autonomic-nervous-system (ANS) dysfunction is frequently reported after a concussion. It has been proposed as a contributing factor to post-concussion-symptoms (PCS) and complaints. There is a wide overlap between symptoms associated with ANS dysfunction and PCS, e.g. dizziness, nausea, fatigue, and exercise intolerance. But, ANS dysfunction can be manifested in both ANS-related symptoms and ANS impaired function. The relationship between PCS and ANS dysfunction is not well understood. The aim of this preliminary study was to examine the relationship between PCS, ANS-related symptoms and ANS function in youth post-concussion.

Methods: Fourteen youth with concussion (mean age 14.2±2.1 years; 10 females; median 23 [13-64] days post-injury) seen at the Montreal Children's Hospital Trauma Centre Concussion Clinic, were included. PCS were assessed using the Post-Concussion-Symptoms-Inventory for Children (PCSI-C). ANS-related symptoms were assessed using the Composite-Autonomic-Symptom-Score-31 (COMPASS-31). ANS function was evaluated using the Polar H10 device under three conditions: 2-minutes at rest in sitting, 2-minutes grip test and 2-minutes paced breathing test. ANS function was assessed using Heart-Rate (HR) and Heart-Rate-Variability (HRV), including the Standard-Deviation-of-the-N-N-interval (SDNN), and the Root-Mean-Square-of-Successive-Differences-between-N-N-intervals (RMSSD).

Results: The median pre-injury PCSI-C score was 13 (range: 0-34), and post-injury PCSI-C score was 48 (5-66); the median PCSI-C-delta score was 29 (3-60). The median COMPASS-31 score was 22.0 (1.3-38.1). The median HR, SDNN and RMSSD values at rest were 73.6 bpm (61.9-106.6), 50.5 ms (28.9-145.9) and 44.6 ms (18.6-188.7), respectively. Moderate significant positive correlations were found between the delta-PCSI-C and COMPASS-31 scores (rs=0.59, p=0.027), and between the SDNN during rest to the COMPASS-31 scores (rs=0.65, p=0.012). Significantly moderate negative correlations were found between the COMPASS-31 scores and the changes from rest in SDNN and RMSSD during the grip test (rs=-0.56, p=0.036 and rs=-0.59, p=0.025). In addition, a significant moderate negative correlation was found between the COMPASS-31 scores and the changes from rest in RMSSD values during the paced breathing test (rs=-0.55, p=0.041). No significant correlations were found between the delta-PCSI-C score and the HR, SDNN or RMSSD values at rest nor during the tests.

Conclusions: This preliminary study suggests that PCS and ANS symptoms severity are moderately related in youth post-concussion. However, ANS function appears to be related only to the ANS symptoms severity and not the PCS severity. These results emphasize that PCS and ANS-related symptoms might represent different aspects of the post-concussion injury. Furthermore, it highlights the importance of assessing ANS symptoms and function in addition to the PCS to evaluate ANS dysfunction post-concussion. Further research with a larger sample size and longer follow-up periods is required to examine the influence of other known concussion recovery factors, such as age, sex, and premorbid health conditions, on the relationship between PCS and ANS dysfunction.

The environment and context of mild traumatic brain injury (mTBI) are critical factors that can determine timely treatment, risk of secondary injury, and long-term health outcomes. However, classifying the context and circumstances of an mTBI is a complex and labor-intensive task, typically requiring comprehensive documentation and description of events, as well as expert interpretation and an agreed set of standard context classifications. Here, we report results from the Long-term Impact of Military-relevant Brain Injury Consortium (LIMBIC) Prospective Longitudinal Study (PLS) that assessed the performance of large language models (LLMs) tasked to classify the context of historical mTBIs from text descriptions of events reported by deployed Service Members and Veterans (SM/V). Text descriptions of 15,539 events were summarized from structured interviews of 2,750 SM/V, with a median age 24 years at time of injury. An expert panel was convened that identified four context categories that encompassed all possible environments in which injury had accrued: 1) Civilian, 2) Military duty outside of deployment, 3) Deployment outside of combat action, and 4) Combat action. Seven expert raters provided gold standard labels for 750 event descriptions, achieving 98.3% agreement among raters. A range of pretrained and fine-tuned LLMs of varying architecture and parameter size were tasked to classify deidentified text descriptions, training on 250 rater-labeled descriptions with performance assessed on 500 held-out multinomial labeled events. Overall, an optimal fine-tuned model reached 4.3% train loss after 3 epochs. Models were then retrained on a subset of discordant edge case examples. The final model achieved 3.1% error under cross-validation. The most frequent contexts reported were Combat action (40.0%), followed by Civilian (29.4%), Military duty outside of deployment (20.4%) and Deployment outside of combat action (10.2%), indicating high rates of mTBI during combat action environments among deployed SM/V. In post-hoc analysis, raters identified that the remaining errors tended to be ambiguous cases that lacked clear context. Categorical context of injury was also found to be significantly associated (p<0.05) with several tested mTBI related measures in ways that suggest further research into etiological considerations may offer personalized approaches and improved quality of life for SM/V with history of mTBI.

One of the common tools used in both research and clinical practice for traumatic brain injury (TBI) is symptom inventories. There are a number of symptom inventories that are routinely collected, including several that are included as Common Data Elements (CDEs) for TBI, but without a way to convert between scales, results drawn from different settings and studies are not comparable. The ENIGMA Clinical Endpoints Working Group set out to determine how to harmonize data from different scales to enable well-powered mega-analyses of large datasets drawn from a variety of sources. The four scales included are the Neurobehavioral Symptom Inventory (NSI), Rivermead Postconcussive Symptom Questionnaire (RPQ), Brief Symptom Inventory (BSI), and the Symptom Checklist-90 (SCL-90). NSI and RPQ are CDEs for TBI, but the BSI and SCL-90 are also routinely collected. Here, we present an approach using semantic textual similarity (STS) to link symptoms and scores across previously incongruous symptom inventories by ranking item text similarities according to their conceptual likeness. We tested the ability of four pre-trained deep learning models to screen thousands of symptom description pairs for related content - a challenging task typically requiring expert panels. Models were tasked to predict symptom severity across four different inventories for 6,607 participants drawn from 16 international data sources. The STS approach achieved 74.8% accuracy across five tasks, outperforming other models tested. The resulting analysis pipeline is available as a free online tool [https://enigma-tools.shinyapps.io/symptom-inventories-calculator/] that can convert scores across previously incompatible symptom inventories. This work suggests that incorporating contextual, semantic information can assist expert decision-making processes, yielding broad gains for the harmonization of TBI assessment.

Background: Acquired brain injury (ABI) may result in a variety of cognitive deficits. The Montreal Cognitive Assessment (MoCA) is one of many screening tools used to assess cognition; however, it does not adequately capture the executive dysfunction and behaviours commonly seen in this population, as it was not created specifically for ABI. The MoCA Observation Tool (MOT) was developed as an add-on to the MoCA, in attempt to capture some of the valuable dysexecutive and behavioural observations made by examiners while administering the MoCA.

Objective: This study aimed to assess the inter-rater reliability of the MOT in its current form, and to assess the opinions of interdisciplinary health care practitioners with respect to its usability in practice.

Methods: Following preliminary testing of the tool, multiple focus groups with an interdisciplinary health care team were conducted to further develop the MOT. The MOT included eight constructs, several of which had subcomponents, detailed as follows: processing speed (mini trails and total test time), attentional regulation (distractibility), perseveration (verbal and motor), behavioural regulation/disinhibition (impulsivity and expletive language), self-monitoring (overlooks mistakes, unaware of mistakes and estimation of performance), cooperation (needs encouragement and refuses task), response to task (not engaged, anxiety, anger/frustration, and other), and memory (category cueing and multiple choice cueing). To assess inter-rater reliability, 30 ABI patients and 11 interdisciplinary health care practitioners were recruited. Videos of the 30 patients completing the MoCA were obtained. The MOT was then applied to each video twice, by two different health care practitioners. Results were analyzed using percentages of agreement and kappa coefficients for each item. Afterwards, a Systems Usability Scale (SUS) was sent to the health care practitioners to assess their overall impression of the tool.

Results: The overall rate of agreement was 82% with the individual percentages of agreement for each item ranging from 53% to 97%. The overall kappa coefficient was 0.54 with individual kappa coefficients ranging from 0.07 to 0.87. Five of the items were scored too infrequently to calculate kappa coefficients. The mean SUS score was 77.08 (SD 12.56).

Conclusion: Results from this study helped further develop the MOT as a tool with acceptable inter-rater reliability. Furthermore, the SUS results indicate an overall positive impression of the MOT with respect to its usability in practice. The MOT offers an objective way to capture valuable information about ABI patients’ executive dysfunction and behaviours using a common cognitive assessment measure; information which can then help inform rehabilitation efforts and approaches to optimize function. Further research is required to assess validity.

Introduction: Practical judgment and problem-solving abilities, critical for independent daily living, are often impaired following a traumatic brain injury (TBI), particularly due to damage to the frontal lobe. However, few comprehensive and valid measures have been designed to assess these abilities. Existing measures either rely on subscales from broader cognitive batteries not specifically designed for this purpose or have been developed for an elderly population.

Objectives: This study aims to validate the Test de Gestion des problèmes de la Vie quotidienne (TGV), a novel tool designed to assess the three key phases of practical judgment: 1) Problem detection, 2) Generation of solutions, 3) Problem severity assessment.

Methods: To date, 42 participants were recruited. The preliminary sample included 21 healthy control participants (M= 46.24 years old; SD = 18.63), and 21 patients who sustained a moderate-to-severe TBI more than a year ago (M = 48.86 years old; SD = 13.24). All participants completed 5 questionnaires: the TGV, the third edition of the Adaptive Behavior Assessment System (ABAS-III), which assesses daily living autonomy, the Quality of life After a Brain Injury (QOLIBRI), the Cognitive Competency Test (CCT), which assesses general judgment, and the Brief Cognitive Exam in Traumatology (EXACT), which assesses global cognitive functioning.

Preliminary Results: Patients with chronic TBI experienced greater challenges in generating effective solutions to everyday problems and accurately assessing their severity compared to healthy control participants. The overall TGV score of patients with chronic TBI was also significantly correlated with the level of autonomy in daily living (r = .61), quality of life (r = .52), and general judgment (r = .64). The score from the second phase of the TGV (generation of solutions) was significantly correlated with the global cognitive functioning (r = .67). These results support the construct and concurrent validities of this new tool. In addition, the internal consistency of the TGV was deemed acceptable (α = .64 to .75).

Conclusions: The TGV is a promising tool for professionals seeking to assess various aspects of practical judgment in patients who sustained a TBI, in a valid and precise manner. A future study will compare patients with chronic TBI who live independently with those hospitalized for moderate-to-severe TBI and deemed unable for independent living by the medical team.