A hierarchical expert-guided machine learning framework for clinical decision support systems: an application to traumatic brain injury prognostication

Traumatic Brain Injury and Artificial Intelligence: Shaping the Future of Neurorehabilitation-A Review

Traumatic brain injury (TBI) is a leading cause of disability and death globally, presenting significant challenges for diagnosis, prognosis, and treatment. As healthcare technology advances, artificial intelligence (AI) has emerged as a promising tool in enhancing TBI rehabilitation outcomes. This literature review explores the current and potential applications of AI in TBI management, focusing on AI's role in diagnostic tools, neuroimaging, prognostic modeling, and rehabilitation programs. AI-driven algorithms have demonstrated high accuracy in predicting mortality, functional outcomes, and personalized rehabilitation strategies based on patient data. AI models have been developed to predict in-hospital mortality of TBI patients up to an accuracy of 95.6%. Furthermore, AI enhances neuroimaging by detecting subtle abnormalities that may be missed by human radiologists, expediting diagnosis and treatment decisions. Despite these advances, ethical considerations, including biases in AI algorithms and data generalizability, pose challenges that must be addressed to optimize AI's implementation in clinical settings. This review highlights key clinical trials and future research directions, emphasizing AI's transformative potential in improving patient care, rehabilitation, and long-term outcomes for TBI patients.

Machine Learning Approaches to Prognostication in Traumatic Brain Injury

PURPOSE OF REVIEW

This review investigates the use of machine learning (ML) in prognosticating outcomes for traumatic brain injury (TBI). It underscores the benefits of ML models in processing and integrating complex, multimodal data-including clinical, imaging, and physiological inputs-to identify intricate non-linear relationships that traditional methods might overlook.

RECENT FINDINGS

ML algorithms of clinical features, neuroimaging, and metrics from the autonomic nervous system enhance the early detection of clinical deterioration and improve outcome prediction. Challenges persist, including issues of data variability, model interpretability, and overfitting. However, advancements in model standardization and validation are key to enhancing their clinical applicability. ML-based, multimodal approaches offer transformative potential for personalized treatment planning and patient management. Future directions include integrating digital twins and real-time continuous data analysis, reinforcing the idea that comprehensive data amalgamation is essential for precise, adaptive prognostication and decision-making in neurocritical care, ultimately leading to better patient outcomes.

Predicting clinical pathways of traumatic brain injuries (TBIs) through process mining

The quality of healthcare services is influenced by a multitude of unpredictable events. Changes in patient clinical conditions and challenges in service organization are only some of the vivid examples that can make the management in healthcare difficult. Estimating patient journeys, known as clinical pathways (CPs), can support care providers in resource planning and enhancing service efficiency. This study presents a decision support system to assist clinicians in predicting CPs and outcomes for patients with traumatic brain injuries (TBIs). This machine learning framework employs an optimal decision tree next to a Markov-based trace clustering as predictive model components. A Shapely value approach extract knowledge of features contribution at both individual and population levels. The proposed approach is validated through a real-life event data, demonstrating high accuracy and providing insights into the rationale behind specific CP predictions which facilitate the adoption of machine learning models in clinical settings.

Proxy endpoints - bridging clinical trials and real world data

OBJECTIVE

Disease severity scores, or endpoints, are routinely measured during Randomized Controlled Trials (RCTs) to closely monitor the effect of treatment. In real-world clinical practice, although a larger set of patients is observed, the specific RCT endpoints are often not captured, which makes it hard to utilize real-world data (RWD) to evaluate drug efficacy in larger populations.

METHODS

To overcome this challenge, we developed an ensemble technique which learns proxy models of disease endpoints in RWD. Using a multi-stage learning framework applied to RCT data, we first identify features considered significant drivers of disease available within RWD. To create endpoint proxy models, we use Explainable Boosting Machines (EBMs) which allow for both end-user interpretability and modeling of non-linear relationships.

RESULTS

We demonstrate our approach on two diseases, rheumatoid arthritis (RA) and atopic dermatitis (AD). As we show, our combined feature selection and prediction method achieves good results for both disease areas, improving upon prior methods proposed for predictive disease severity scoring.

CONCLUSION

Having disease severity over time for a patient is important to further disease understanding and management. Our results open the door to more use cases in the space of RA and AD such as treatment effect estimates or prognostic scoring on RWD. Our framework may be extended beyond RA and AD to other diseases where the severity score is not well measured in electronic health records.

Artificial Intelligence-Enhanced Neurocritical Care for Traumatic Brain Injury : Past, Present and Future

In neurointensive care units (NICUs), particularly in cases involving traumatic brain injury (TBI), swift and accurate decision-making is critical because of rapidly changing patient conditions and the risk of secondary brain injury. The use of artificial intelligence (AI) in NICU can enhance clinical decision support and provide valuable assistance in these complex scenarios. This article aims to provide a comprehensive review of the current status and future prospects of AI utilization in the NICU, along with the challenges that must be overcome to realize this. Presently, the primary application of AI in NICU is outcome prediction through the analysis of preadmission and high-resolution data during admission. Recent applications include augmented neuromonitoring via signal quality control and real-time event prediction. In addition, AI can integrate data gathered from various measures and support minimally invasive neuromonitoring to increase patient safety. However, despite the recent surge in AI adoption within the NICU, the majority of AI applications have been limited to simple classification tasks, thus leaving the true potential of AI largely untapped. Emerging AI technologies, such as generalist medical AI and digital twins, harbor immense potential for enhancing advanced neurocritical care through broader AI applications. If challenges such as acquiring high-quality data and ethical issues are overcome, these new AI technologies can be clinically utilized in the actual NICU environment. Emphasizing the need for continuous research and development to maximize the potential of AI in the NICU, we anticipate that this will further enhance the efficiency and accuracy of TBI treatment within the NICU.

Discovery of generalizable TBI phenotypes using multivariate time-series clustering

Traumatic Brain Injury (TBI) presents a broad spectrum of clinical presentations and outcomes due to its inherent heterogeneity, leading to diverse recovery trajectories and varied therapeutic responses. While many studies have delved into TBI phenotyping for distinct patient populations, identifying TBI phenotypes that consistently generalize across various settings and populations remains a critical research gap. Our research addresses this by employing multivariate time-series clustering to unveil TBI's dynamic intricates. Utilizing a self-supervised learning-based approach to clustering multivariate time-Series data with missing values (SLAC-Time), we analyzed both the research-centric TRACK-TBI and the real-world MIMIC-IV datasets. Remarkably, the optimal hyperparameters of SLAC-Time and the ideal number of clusters remained consistent across these datasets, underscoring SLAC-Time's stability across heterogeneous datasets. Our analysis revealed three generalizable TBI phenotypes (α, β, and γ), each exhibiting distinct non-temporal features during emergency department visits, and temporal feature profiles throughout ICU stays. Specifically, phenotype α represents mild TBI with a remarkably consistent clinical presentation. In contrast, phenotype β signifies severe TBI with diverse clinical manifestations, and phenotype γ represents a moderate TBI profile in terms of severity and clinical diversity. Age is a significant determinant of TBI outcomes, with older cohorts recording higher mortality rates. Importantly, while certain features varied by age, the core characteristics of TBI manifestations tied to each phenotype remain consistent across diverse populations.

Discovery of Generalizable TBI Phenotypes Using Multivariate Time-Series Clustering

Traumatic Brain Injury (TBI) presents a broad spectrum of clinical presentations and outcomes due to its inherent heterogeneity, leading to diverse recovery trajectories and varied therapeutic responses. While many studies have delved into TBI phenotyping for distinct patient populations, identifying TBI phenotypes that consistently generalize across various settings and populations remains a critical research gap. Our research addresses this by employing multivariate time-series clustering to unveil TBI's dynamic intricates. Utilizing a self-supervised learning-based approach to clustering multivariate time-Series data with missing values (SLAC-Time), we analyzed both the research-centric TRACK-TBI and the real-world MIMIC-IV datasets. Remarkably, the optimal hyperparameters of SLAC-Time and the ideal number of clusters remained consistent across these datasets, underscoring SLAC-Time's stability across heterogeneous datasets. Our analysis revealed three generalizable TBI phenotypes (α, β, and γ), each exhibiting distinct non-temporal features during emergency department visits, and temporal feature profiles throughout ICU stays. Specifically, phenotype α represents mild TBI with a remarkably consistent clinical presentation. In contrast, phenotype β signifies severe TBI with diverse clinical manifestations, and phenotype γ represents a moderate TBI profile in terms of severity and clinical diversity. Age is a significant determinant of TBI outcomes, with older cohorts recording higher mortality rates. Importantly, while certain features varied by age, the core characteristics of TBI manifestations tied to each phenotype remain consistent across diverse populations.

Early identification of autism spectrum disorder based on machine learning with eye-tracking data

BACKGROUND

Early identification of autism spectrum disorder (ASD) improves long-term outcomes, yet significant diagnostic delays persist.

METHODS

A retrospective cohort of 449 children (ASD: 246, typically developing [TD]: 203) was used for model development. Eye-movement data were collected from the participants watching videos that featured eye-tracking paradigms for assessing social and non-social cognition. Five machine learning algorithms, namely random forest, support vector machine, logistic regression, artificial neural network, and extreme gradient boosting, were trained to classify children with ASD and TD. The best-performing algorithm was selected to build the final model which was further evaluated in a prospective cohort of 80 children. The Shapley values interpreted important eye-tracking features.

RESULTS

Random forest outperformed other algorithms during model development and achieved an area under the curve of 0.849 (< 3 years: 0.832, ≥ 3 years: 0.868) on the external validation set. Of the ten most important eye-tracking features, three measured social cognition, and the rest were related to non-social cognition. A deterioration in model performance was observed using only the social or non-social cognition-related eye-tracking features.

LIMITATIONS

The sample size of this study, although larger than that of existing studies of ASD based on eye-tracking data, was still relatively small compared to the number of features.

CONCLUSIONS

Machine learning models based on eye-tracking data have the potential to be cost- and time-efficient digital tools for the early identification of ASD. Eye-tracking phenotypes related to social and non-social cognition play an important role in distinguishing children with ASD from TD children.

Development of a Multimodal Machine Learning-Based Prognostication Model for Traumatic Brain Injury Using Clinical Data and Computed Tomography Scans: A CENTER-TBI and CINTER-TBI Study

Computed tomography (CT) is an important imaging modality for guiding prognostication in patients with traumatic brain injury (TBI). However, because of the specialized expertise necessary, timely and dependable TBI prognostication based on CT imaging remains challenging. This study aimed to enhance the efficiency and reliability of TBI prognostication by employing machine learning (ML) techniques on CT images. A retrospective analysis was conducted on the Collaborative European NeuroTrauma Effectiveness Research in TBI (CENTER-TBI) data set (n = 1016). An ML-driven binary classifier was developed to predict favorable or unfavorable outcomes at 6 months post-injury. The prognostic performance was assessed using the area under the curve (AUC) over fivefold cross-validation and compared with conventional models that depend on clinical variables and CT scoring systems. An external validation was performed using the Comparative Indian Neurotrauma Effectiveness Research in Traumatic Brain Injury (CINTER-TBI) data set (n = 348). The developed model achieved superior performance without the necessity for manual CT assessments (AUC = 0.846 [95% CI: 0.843-0.849]) compared with the model based on the clinical and laboratory variables (AUC = 0.817 [95% CI: 0.814-0.820]) and established CT scoring systems requiring manual interpretations (AUC = 0.829 [95% CI: 0.826-0.832] for Marshall and 0.838 [95% CI: 0.835-0.841] for International Mission for Prognosis and Analysis of Clinical Trials in TBI [IMPACT]). The external validation demonstrated the prognostic capacity of the developed model to be significantly better (AUC = 0.859 [95% CI: 0.857-0.862]) than the model using clinical variables (AUC = 0.809 [95% CI: 0.798-0.820]). This study established an ML-based model that provides efficient and reliable TBI prognosis based on CT scans, with potential implications for earlier intervention and improved patient outcomes.

Predicting autism spectrum disorder using maternal risk factors: A multi-center machine learning study

Autism spectrum disorder (ASD) is a neurodevelopmental disorder with a complex environmental etiology involving maternal risk factors, which have been combined with machine learning to predict ASD. However, limited studies have considered the factors throughout preconception, perinatal, and postnatal periods, and even fewer have been conducted in multi-center. In this study, five predictive models were developed using 57 maternal risk factors from a cohort across ten cities (ASD:1232, typically developing[TD]: 1090). The extreme gradient boosting model performed best, achieving an accuracy of 66.2 % on the external cohort from three cities (ASD:266, TD:353). The most important risk factors were identified as unstable emotions and lack of multivitamin supplementation using Shapley values. ASD risk scores were calculated based on predicted probabilities from the optimal model and divided into low, medium, and high-risk groups. The logistic analysis indicated that the high-risk group had a significantly increased risk of ASD compared to the low-risk group. Our study demonstrated the potential of machine learning models in predicting the risk for ASD based on maternal factors. The developed model provided insights into the maternal emotion and nutrition factors associated with ASD and highlighted the potential clinical applicability of the developed model in identifying high-risk populations.

Enhancing hospital course and outcome prediction in patients with traumatic brain injury: A machine learning study

PURPOSE

We aimed to use machine learning (ML) algorithms with clinical, lab, and imaging data as input to predict various outcomes in traumatic brain injury (TBI) patients.

METHODS

In this retrospective study, blood samples were analyzed for glial fibrillary acidic protein (GFAP) and ubiquitin C-terminal hydrolase L1 (UCH-L1). The non-contrast head CTs were reviewed by two neuroradiologists for TBI common data elements (CDE). Three outcomes were designed to predict: discharged or admitted for further management (prediction 1), deceased or not deceased (prediction 2), and admission only, prolonged stay, or neurosurgery performed (prediction 3). Five ML models were trained. SHapley Additive exPlanations (SHAP) analyses were used to assess the relative significance of variables.

RESULTS

Four hundred forty patients were used to predict predictions 1 and 2, while 271 patients were used in prediction 3. Due to Prediction 3's hospitalization requirement, deceased and discharged patients could not be utilized. The Random Forest model achieved an average accuracy of 1.00 for prediction 1 and an accuracy of 0.99 for prediction 2. The Random Forest model achieved a mean accuracy of 0.93 for prediction 3. Key features were extracranial injury, hemorrhage, UCH-L1 for prediction 1; The Glasgow Coma Scale, age, GFAP for prediction 2; and GFAP, subdural hemorrhage volume, and pneumocephalus for prediction 3, per SHAP analysis.

CONCLUSION

Combining clinical and laboratory parameters with non-contrast CT CDEs allowed our ML models to accurately predict the designed outcomes of TBI patients. GFAP and UCH-L1 were among the significant predictor variables, demonstrating the importance of these biomarkers.

Predicting return to work after traumatic brain injury using machine learning and administrative data

BACKGROUND

Accurate patient-specific predictions on return-to-work after traumatic brain injury (TBI) can support both clinical practice and policymaking. The use of machine learning on large administrative data provides interesting opportunities to create such prognostic models.

AIM

The current study assesses whether return-to-work one year after TBI can be predicted accurately from administrative data. Additionally, this study explores how model performance and feature importance change depending on whether a distinction is made between mild and moderate-to-severe TBI.

METHODS

This study used a population-based dataset that combined discharge, claims and social security data of patients hospitalized with a TBI in Belgium during the year 2016. The prediction of TBI was attempted with three algorithms, elastic net logistic regression, random forest and gradient boosting and compared in their performance by their accuracy, sensitivity, specificity and area under the receiver operator curve (ROC AUC).

RESULTS

The distinct modelling algorithms resulted in similar results, with 83% accuracy (ROC AUC 85%) for a binary classification of employed vs. not employed and up to 76% (ROC AUC 82%) for a multiclass operationalization of employment outcome. Modelling mild and moderate-to-severe TBI separately did not result in considerable differences in model performance and feature importance. The features of main importance for return-to-work prediction were related to pre-injury employment.

DISCUSSION

While clearly offering some information beneficial for predicting return-to-work, administrative data needs to be supplemented with additional information to allow further improvement of patient-specific prognose.

Machine Learning Approach for the Prediction of In-Hospital Mortality in Traumatic Brain Injury Using Bio-Clinical Markers at Presentation to the Emergency Department

BACKGROUND

Accurate prediction of in-hospital mortality is essential for better management of patients with traumatic brain injury (TBI). Machine learning (ML) algorithms have been shown to be effective in predicting clinical outcomes. This study aimed to identify predictors of in-hospital mortality in TBI patients using ML algorithms.

MATERIALS AND METHOD

A retrospective study was performed using data from both the trauma registry and electronic medical records among TBI patients admitted to the Hamad Trauma Center in Qatar between June 2016 and May 2021. Thirteen features were selected for four ML models including a Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), and Extreme Gradient Boosting (XgBoost), to predict the in-hospital mortality.

RESULTS

A dataset of 922 patients was analyzed, of which 78% survived and 22% died. The AUC scores for SVM, LR, XgBoost, and RF models were 0.86, 0.84, 0.85, and 0.86, respectively. XgBoost and RF had good AUC scores but exhibited significant differences in log loss between the training and testing sets (% difference in logloss of 79.5 and 41.8, respectively), indicating overfitting compared to the other models. The feature importance trend across all models indicates that aPTT, INR, ISS, prothrombin time, and lactic acid are the most important features in prediction. Magnesium also displayed significant importance in the prediction of mortality among serum electrolytes.

CONCLUSIONS

SVM was found to be the best-performing ML model in predicting the mortality of TBI patients. It had the highest AUC score and did not show overfitting, making it a more reliable model compared to LR, XgBoost, and RF.

A self-supervised learning-based approach to clustering multivariate time-series data with missing values (SLAC-Time): An application to TBI phenotyping

Self-supervised learning approaches provide a promising direction for clustering multivariate time-series data. However, real-world time-series data often include missing values, and the existing approaches require imputing missing values before clustering, which may cause extensive computations and noise and result in invalid interpretations. To address these challenges, we present a Self-supervised Learning-based Approach to Clustering multivariate Time-series data with missing values (SLAC-Time). SLAC-Time is a Transformer-based clustering method that uses time-series forecasting as a proxy task for leveraging unlabeled data and learning more robust time-series representations. This method jointly learns the neural network parameters and the cluster assignments of the learned representations. It iteratively clusters the learned representations with the K-means method and then utilizes the subsequent cluster assignments as pseudo-labels to update the model parameters. To evaluate our proposed approach, we applied it to clustering and phenotyping Traumatic Brain Injury (TBI) patients in the Transforming Research and Clinical Knowledge in Traumatic Brain Injury (TRACK-TBI) study. Clinical data associated with TBI patients are often measured over time and represented as time-series variables characterized by missing values and irregular time intervals. Our experiments demonstrate that SLAC-Time outperforms the baseline K-means clustering algorithm in terms of silhouette coefficient, Calinski Harabasz index, Dunn index, and Davies Bouldin index. We identified three TBI phenotypes that are distinct from one another in terms of clinically significant variables as well as clinical outcomes, including the Extended Glasgow Outcome Scale (GOSE) score, Intensive Care Unit (ICU) length of stay, and mortality rate. The experiments show that the TBI phenotypes identified by SLAC-Time can be potentially used for developing targeted clinical trials and therapeutic strategies.

Uncovering expression signatures of synergistic drug responses via ensembles of explainable machine-learning models

Machine learning may aid the choice of optimal combinations of anticancer drugs by explaining the molecular basis of their synergy. By combining accurate models with interpretable insights, explainable machine learning promises to accelerate data-driven cancer pharmacology. However, owing to the highly correlated and high-dimensional nature of transcriptomic data, naively applying current explainable machine-learning strategies to large transcriptomic datasets leads to suboptimal outcomes. Here by using feature attribution methods, we show that the quality of the explanations can be increased by leveraging ensembles of explainable machine-learning models. We applied the approach to a dataset of 133 combinations of 46 anticancer drugs tested in ex vivo tumour samples from 285 patients with acute myeloid leukaemia and uncovered a haematopoietic-differentiation signature underlying drug combinations with therapeutic synergy. Ensembles of machine-learning models trained to predict drug combination synergies on the basis of gene-expression data may improve the feature attribution quality of complex machine-learning models.

AI-Based Decision Support System for Traumatic Brain Injury: A Survey

Traumatic brain injury (TBI) is one of the major causes of disability and mortality worldwide. Rapid and precise clinical assessment and decision-making are essential to improve the outcome and the resulting complications. Due to the size and complexity of the data analyzed in TBI cases, computer-aided data processing, analysis, and decision support systems could play an important role. However, developing such systems is challenging due to the heterogeneity of symptoms, varying data quality caused by different spatio-temporal resolutions, and the inherent noise associated with image and signal acquisition. The purpose of this article is to review current advances in developing artificial intelligence-based decision support systems for the diagnosis, severity assessment, and long-term prognosis of TBI complications.

Predicting the need for intubation within 3 h in the neonatal intensive care unit using a multimodal deep neural network

Respiratory distress is a common chief complaint in neonates admitted to the neonatal intensive care unit. Despite the increasing use of non-invasive ventilation in neonates with respiratory difficulty, some of them require advanced airway support. Delayed intubation is associated with increased morbidity, particularly in urgent unplanned cases. Early and accurate prediction of the need for intubation may provide more time for preparation and increase safety margins by avoiding the late intubation at high-risk infants. This study aimed to predict the need for intubation within 3 h in neonates initially managed with non-invasive ventilation for respiratory distress during the first 48 h of life using a multimodal deep neural network. We developed a multimodal deep neural network model to simultaneously analyze four time-series data collected at 1-h intervals and 19 variables including demographic, physiological and laboratory parameters. Evaluating the dataset of 128 neonates with respiratory distress who underwent non-invasive ventilation, our model achieved an area under the curve of 0.917, sensitivity of 85.2%, and specificity of 89.2%. These findings demonstrate promising results for the multimodal model in predicting neonatal intubation within 3 h.

Developing AI enabled sensors and decision support for military operators in the field

Wearable sensors enable down range data collection of physiological and cognitive performance of the warfighter. However, autonomous teams may find the sensor data impractical to interpret and hence influence real-time decisions without the support of subject matter experts. Decision support tools can reduce the burden of interpreting physiological data in the field and incorporate a systems perspective where noisy field data can contain useful additional signals. We present a methodology of how artificial intelligence can be used for modeling human performance with decision-making to achieve actionable decision support. We provide a framework for systems design and advancing from the laboratory to real world environments. The result is a validated measure of down-range human performance with a low burden of operation.

An end-end deep learning framework for lesion segmentation on multi-contrast MR images-an exploratory study in a rat model of traumatic brain injury

Traumatic brain injury (TBI) engenders traumatic necrosis and penumbra-areas of secondary neural injury which are crucial targets for therapeutic interventions. Segmenting manually areas of ongoing changes like necrosis, edema, hematoma, and inflammation is tedious, error-prone, and biased. Using the multi-parametric MR data from a rodent model study, we demonstrate the effectiveness of an end-end deep learning global-attention-based UNet (GA-UNet) framework for automatic segmentation and quantification of TBI lesions. Longitudinal MR scans (2 h, 1, 3, 7, 14, 30, and 60 days) were performed on eight Sprague-Dawley rats after controlled cortical injury was performed. TBI lesion and sub-regions segmentation was performed using 3D-UNet and GA-UNet. Dice statistics (DSI) and Hausdorff distance were calculated to assess the performance. MR scan variations-based (bias, noise, blur, ghosting) data augmentation was performed to develop a robust model.Training/validation median DSI for U-Net was 0.9368 with T2w and MPRAGE inputs, whereas GA-UNet had 0.9537 for the same. Testing accuracies were higher for GA-UNet than U-Net with a DSI of 0.8232 for the T2w-MPRAGE inputs.Longitudinally, necrosis remained constant while oligemia and penumbra decreased, and edema appearing around day 3 which increased with time. GA-UNet shows promise for multi-contrast MR image-based segmentation/quantification of TBI in large cohort studies.

Machine learning and network analysis for diagnosis and prediction in disorders of consciousness

BACKGROUND

Prolonged Disorders of Consciousness (PDOC) resulting from severe acquired brain injury can lead to complex disabilities that make diagnosis challenging. The role of machine learning (ML) in diagnosing PDOC states and identifying intervention strategies is relatively under-explored, having focused on predicting mortality and poor outcome. This study aims to: (a) apply ML techniques to predict PDOC diagnostic states from variables obtained from two non-invasive neurobehavior assessment tools; and (b) apply network analysis for guiding possible intervention strategies.

METHODS

The Coma Recovery Scale-Revised (CRS-R) is a well-established tool for assessing patients with PDOC. More recently, music has been found to be a useful medium for assessment of coma patients, leading to the standardization of a music-based assessment of awareness: Music Therapy Assessment Tool for Awareness in Disorders of Consciousness (MATADOC). CRS-R and MATADOC data were collected from 74 PDOC patients aged 16-70 years at three specialist centers in the USA, UK and Ireland. The data were analyzed by three ML techniques (neural networks, decision trees and cluster analysis) as well as modelled through system-level network analysis.

RESULTS

PDOC diagnostic state can be predicted to a relatively high level of accuracy that sets a benchmark for future ML analysis using neurobehavioral data only. The outcomes of this study may also have implications for understanding the role of music therapy in interdisciplinary rehabilitation to help patients move from one coma state to another.

CONCLUSIONS

This study has shown how ML can derive rules for diagnosis of PDOC with data from two neurobehavioral tools without the need to harvest large clinical and imaging datasets. Network analysis using the measures obtained from these two non-invasive tools provides novel, system-level ways of interpreting possible transitions between PDOC states, leading to possible use in novel, next-generation decision-support systems for PDOC.

Application of explainable artificial intelligence for healthcare: A systematic review of the last decade (2011-2022)

BACKGROUND AND OBJECTIVES

Artificial intelligence (AI) has branched out to various applications in healthcare, such as health services management, predictive medicine, clinical decision-making, and patient data and diagnostics. Although AI models have achieved human-like performance, their use is still limited because they are seen as a black box. This lack of trust remains the main reason for their low use in practice, especially in healthcare. Hence, explainable artificial intelligence (XAI) has been introduced as a technique that can provide confidence in the model's prediction by explaining how the prediction is derived, thereby encouraging the use of AI systems in healthcare. The primary goal of this review is to provide areas of healthcare that require more attention from the XAI research community.

METHODS

Multiple journal databases were thoroughly searched using PRISMA guidelines 2020. Studies that do not appear in Q1 journals, which are highly credible, were excluded.

RESULTS

In this review, we surveyed 99 Q1 articles covering the following XAI techniques: SHAP, LIME, GradCAM, LRP, Fuzzy classifier, EBM, CBR, rule-based systems, and others.

CONCLUSION

We discovered that detecting abnormalities in 1D biosignals and identifying key text in clinical notes are areas that require more attention from the XAI research community. We hope this is review will encourage the development of a holistic cloud system for a smart city.

Predicting outcome of patients with prolonged disorders of consciousness using machine learning models based on medical complexity

Patients with severe acquired brain injury and prolonged disorders of consciousness (pDoC) are characterized by high clinical complexity and high risk to develop medical complications. The present multi-center longitudinal study aimed at investigating the impact of medical complications on the prediction of clinical outcome by means of machine learning models. Patients with pDoC were consecutively enrolled at admission in 23 intensive neurorehabilitation units (IRU) and followed-up at 6 months from onset via the Glasgow Outcome Scale-Extended (GOSE). Demographic and clinical data at study entry and medical complications developed within 3 months from admission were collected. Machine learning models were developed, targeting neurological outcomes at 6 months from brain injury using data collected at admission. Then, after concatenating predictions of such models to the medical complications collected within 3 months, a cascade model was developed. One hundred seventy six patients with pDoC (M: 123, median age 60.2 years) were included in the analysis. At admission, the best performing solution (k-Nearest Neighbors regression, KNN) resulted in a median validation error of 0.59 points [IQR 0.14] and a classification accuracy of dichotomized GOS-E of 88.6%. Coherently, at 3 months, the best model resulted in a median validation error of 0.49 points [IQR 0.11] and a classification accuracy of 92.6%. Interpreting the admission KNN showed how the negative effect of older age is strengthened when patients' communication levels are high and ameliorated when no communication is present. The model trained at 3 months showed appropriate adaptation of the admission prediction according to the severity of the developed medical complexity in the first 3 months. In this work, we developed and cross-validated an interpretable decision support tool capable of distinguishing patients which will reach sufficient independence levels at 6 months (GOS-E > 4). Furthermore, we provide an updated prediction at 3 months, keeping in consideration the rehabilitative path and the risen medical complexity.

An interpretable neural network for outcome prediction in traumatic brain injury

BACKGROUND

Traumatic Brain Injury (TBI) is a common condition with potentially severe long-term complications, the prediction of which remains challenging. Machine learning (ML) methods have been used previously to help physicians predict long-term outcomes of TBI so that appropriate treatment plans can be adopted. However, many ML techniques are "black box": it is difficult for humans to understand the decisions made by the model, with post-hoc explanations only identifying isolated relevant factors rather than combinations of factors. Moreover, such models often rely on many variables, some of which might not be available at the time of hospitalization.

METHODS

In this study, we apply an interpretable neural network model based on tropical geometry to predict unfavorable outcomes at six months from hospitalization in TBI patients, based on information available at the time of admission.

RESULTS

The proposed method is compared to established machine learning methods-XGBoost, Random Forest, and SVM-achieving comparable performance in terms of area under the receiver operating characteristic curve (AUC)-0.799 for the proposed method vs. 0.810 for the best black box model. Moreover, the proposed method allows for the extraction of simple, human-understandable rules that explain the model's predictions and can be used as general guidelines by clinicians to inform treatment decisions.

CONCLUSIONS

The classification results for the proposed model are comparable with those of traditional ML methods. However, our model is interpretable, and it allows the extraction of intelligible rules. These rules can be used to determine relevant factors in assessing TBI outcomes and can be used in situations when not all necessary factors are known to inform the full model's decision.

The leap to ordinal: Detailed functional prognosis after traumatic brain injury with a flexible modelling approach

When a patient is admitted to the intensive care unit (ICU) after a traumatic brain injury (TBI), an early prognosis is essential for baseline risk adjustment and shared decision making. TBI outcomes are commonly categorised by the Glasgow Outcome Scale–Extended (GOSE) into eight, ordered levels of functional recovery at 6 months after injury. Existing ICU prognostic models predict binary outcomes at a certain threshold of GOSE (e.g., prediction of survival [GOSE > 1]). We aimed to develop ordinal prediction models that concurrently predict probabilities of each GOSE score. From a prospective cohort (n = 1,550, 65 centres) in the ICU stratum of the Collaborative European NeuroTrauma Effectiveness Research in TBI (CENTER-TBI) patient dataset, we extracted all clinical information within 24 hours of ICU admission (1,151 predictors) and 6-month GOSE scores. We analysed the effect of two design elements on ordinal model performance: (1) the baseline predictor set, ranging from a concise set of ten validated predictors to a token-embedded representation of all possible predictors, and (2) the modelling strategy, from ordinal logistic regression to multinomial deep learning. With repeated k-fold cross-validation, we found that expanding the baseline predictor set significantly improved ordinal prediction performance while increasing analytical complexity did not. Half of these gains could be achieved with the addition of eight high-impact predictors to the concise set. At best, ordinal models achieved 0.76 (95% CI: 0.74–0.77) ordinal discrimination ability (ordinal c-index) and 57% (95% CI: 54%– 60%) explanation of ordinal variation in 6-month GOSE (Somers’ D_xy). Model performance and the effect of expanding the predictor set decreased at higher GOSE thresholds, indicating the difficulty of predicting better functional outcomes shortly after ICU admission. Our results motivate the search for informative predictors that improve confidence in prognosis of higher GOSE and the development of ordinal dynamic prediction models.

Demystifying the Black Box: The Importance of Interpretability of Predictive Models in Neurocritical Care

Neurocritical care patients are a complex patient population, and to aid clinical decision-making, many models and scoring systems have previously been developed. More recently, techniques from the field of machine learning have been applied to neurocritical care patient data to develop models with high levels of predictive accuracy. However, although these recent models appear clinically promising, their interpretability has often not been considered and they tend to be black box models, making it extremely difficult to understand how the model came to its conclusion. Interpretable machine learning methods have the potential to provide the means to overcome some of these issues but are largely unexplored within the neurocritical care domain. This article examines existing models used in neurocritical care from the perspective of interpretability. Further, the use of interpretable machine learning will be explored, in particular the potential benefits and drawbacks that the techniques may have when applied to neurocritical care data. Finding a solution to the lack of model explanation, transparency, and accountability is important because these issues have the potential to contribute to model trust and clinical acceptance, and, increasingly, regulation is stipulating a right to explanation for decisions made by models and algorithms. To ensure that the prospective gains from sophisticated predictive models to neurocritical care provision can be realized, it is imperative that interpretability of these models is fully considered.

Restoring and attributing ancient texts using deep neural networks

Ancient history relies on disciplines such as epigraphy-the study of inscribed texts known as inscriptions-for evidence of the thought, language, society and history of past civilizations1. However, over the centuries, many inscriptions have been damaged to the point of illegibility, transported far from their original location and their date of writing is steeped in uncertainty. Here we present Ithaca, a deep neural network for the textual restoration, geographical attribution and chronological attribution of ancient Greek inscriptions. Ithaca is designed to assist and expand the historian's workflow. The architecture of Ithaca focuses on collaboration, decision support and interpretability. While Ithaca alone achieves 62% accuracy when restoring damaged texts, the use of Ithaca by historians improved their accuracy from 25% to 72%, confirming the synergistic effect of this research tool. Ithaca can attribute inscriptions to their original location with an accuracy of 71% and can date them to less than 30 years of their ground-truth ranges, redating key texts of Classical Athens and contributing to topical debates in ancient history. This research shows how models such as Ithaca can unlock the cooperative potential between artificial intelligence and historians, transformationally impacting the way that we study and write about one of the most important periods in human history.

Combined Radiomics Model for Prediction of Hematoma Progression and Clinical Outcome of Cerebral Contusions in Traumatic Brain Injury

BACKGROUND

Traumatic brain injury is a common and devastating injury that is the leading cause of neurological disability and death worldwide. Patients with cerebral lobe contusion received conservative treatment because of their mild manifestations, but delayed intracranial hematoma may increase and even become life-threatening. We explored the noninvasive method to predict the prognosis of progression and Glasgow Outcome Scale (GOS) by using a quantitative radiomics approach and statistical analysis.

METHODS

Eighty-eight patients who were pathologically diagnosed were retrospectively studied. The radiomics method developed in this work included image segmentation, feature extraction, and feature selection. The nomograms were established based on statistical analysis and a radiomics method. We conducted a comparative study of hematoma progression and GOS between the clinical factor alone and fusion radiomics features.

RESULTS

Nineteen clinical factors, 513 radiomics features, and 116 locational features were considered. Among clinical factors, international normalized ratio, prothrombin time, and fibrinogen were enrolled for hematoma progression. As for GOS, treatment strategy, age, Glasgow Coma Scale score, and blood platelet were associated factors. Eight features for GOS and five features for hematoma progression were filtered by using sparse representation and locality preserving projection-combined method. Four nomograms were constructed. After fusion radiomics features, area under the curve of hematoma progression prediction increased from 0.832 to 0.899, whereas GOS prediction went from 0.794 to 0.844.

CONCLUSIONS

A radiomic-based model that merges radiomics and clinical features is a noninvasive approach to predict hematoma progression and clinical outcomes of cerebral contusions in traumatic brain injury.

A Deep Neural Network Architecture to Model Reference Evapotranspiration Using a Single Input Meteorological Parameter

Hydro-agrological research considers the reference evapotranspiration (ETo), driven by meteorological variables, crucial for achieving precise irrigation in precision agriculture. ETo modelling based on a single meteorological parameter would be beneficial in places where the collection of climatic parameters is challenging. The aim of this research is to develop a deep neural network (DNN) architecture that predicts daily ETo with a single input parameter selected based on the feature importance (FI) score generated by the machine learning techniques, random forest (RF), and extreme gradient boosting (XGBoost). This study also investigated the potential of SHapley Additive exPlanations to interpret and validate the outcomes of the feature selection methods by assessing the contributions of each feature to the ETo prediction. These methods recommended solar radiation as a significant parameter in the datasets of three California Irrigation Management System (CIMIS) weather stations located in distinct ETo zones. Three ETo models (DNN-Ret, XGB-Ret, and RF-Ret) were built using solar radiation as the sole input, and CIMIS ETo as the output. The performance evaluation of the developed models proved that DNN-Ret outperformed XGB-Ret and RF-Ret regardless of the dataset, with coefficients of determination (R²) ranging from 0.914 to 0.954 in the local scenario, with an average decrease of 8–9.5% in mean absolute error and root mean squared error, and an improvement of 2.6–2.9% in Nash–Sutcliffe efficiency and 1.7–2% increase in R². The overall result analysis highlighted the efficiency of DNN-Ret in the single input parameter based ETo modelling in diverse climatic zones.