Clinical risk prediction with random forests for survival, longitudinal, and multivariate (RF-SLAM) data analysis

Neuroception of psychological safety scale (NPSS): validation with a UK based adult community sample

Background: Psychological safety plays a vital role in rest, recovery, and fostering social connections. However, a history of trauma can predispose individuals to perceive heightened levels of threat and danger. Research suggests that a lack of psychological safety may be a defining biopsychosocial characteristic of posttraumatic stress disorder (PTSD). Persistent feelings of threat and danger are associated with a lack of psychological safety and may be predictive of PTSD. Our pioneering work reported on the development of the neuroception of psychological safety (NPSS), rooted in polyvagal theory, and consists of social engagement, compassion, and body sensations dimensions. Understanding more about the dimensionality of the NPSS and further establishing its psychometric properties was our priority.Objective: Our current research aimed to validate and test the reliability and dimensionality of the NPSS, using a large community sample (n = 2035) of adult residents in the UKMethod: We examined the internal and test-retest reliability, convergent, discriminant, and concurrent validity as well as dimensionality of the NPSS.Results: The 3-factor structure of the NPSS was replicated with regard to the absolute fit indices. Internal consistencies ranged from acceptable to excellent across the NPSS's subscales. Providing support for the validity of the NPSS, scores were predictably related to team psychological safety, wellbeing, post-traumatic stress, burnout, body awareness, and personality, with effect sizes typically in the high to medium range. Scores on the NPSS were found to show good test-retest reliability.Conclusions: This study demonstrates the validity, reliability and dimensionality of the NPSS with an adult sample. Further work is underway to support and enhance understandings of psychological safety with diverse clinical populations impacted by trauma. The NPSS has applicability across a range of health and social care contexts, such as shaping new approaches to evaluating trauma treatments and enhancing trauma informed care.

Artificial intelligence-driven clinical decision support systems for early detection and precision therapy in oral cancer: a mini review

Oral cancer (OC) is a significant global health burden, with life-saving improvements in survival and outcomes being dependent on early diagnosis and precise treatment planning. However, diagnosis and treatment planning are predicated on the synthesis of complicated information derived from clinical assessment, imaging, histopathology and patient histories. Artificial intelligence-based clinical decision support systems (AI-CDSS) provides a viable solution that can be implemented via advanced methodologies for data analysis, and synthesis for better diagnostic and prognostic evaluation. This review presents AI-CDSS as a promising solution through advanced methodologies for comprehensive data analysis. In addition, it examines current implementations of AI-CDSS that facilitate early OC detection, precise staging, and personalized treatment planning by processing multimodal patient information through machine learning, computer vision, and natural language processing. These systems effectively interpret clinical results, identify critical disease patterns (including clinical stage, site, tumor dimensions, histopathologic grading, and molecular profiles), and construct comprehensive patient profiles. This comprehensive AI-CDSS approach allows for early cancer detection, a reduction in diagnostic delays and improved intervention outcomes. Moreover, the AI-CDSS also optimizes treatment plans on the basis of unique patient parameters, tumor stages and risk factors, providing personalized therapy.

Evaluation of Machine Learning and Traditional Statistical Models to Assess the Value of Stroke Genetic Liability for Prediction of Risk of Stroke Within the UK Biobank

Background and Objective: Stroke is one of the leading causes of mortality and long-term disability in adults over 18 years of age globally, and its increasing incidence has become a global public health concern. Accurate stroke prediction is highly valuable for early intervention and treatment. There is a scarcity of studies evaluating the prediction value of genetic liability in the prediction of the risk of stroke. Materials and Methods: Our study involved 243,339 participants of European ancestry from the UK Biobank. We created stroke genetic liability using data from MEGASTROKE genome-wide association studies (GWASs). In our study, we built four predictive models with and without stroke genetic liability in the training set, namely a Cox proportional hazard (Coxph) model, gradient boosting model (GBM), decision tree (DT), and random forest (RF), to estimate time-to-event risk for stroke. We then assessed their performances in the testing set. Results: Each unit (standard deviation) increase in genetic liability increases the risk of incident stroke by 7% (HR = 1.07, 95% CI = 1.02, 1.12, p-value = 0.0030). The risk of stroke was greater in the higher genetic liability group, demonstrated by a 14% increased risk (HR = 1.14, 95% CI = 1.02, 1.27, p-value = 0.02) compared with the low genetic liability group. The Coxph model including genetic liability was the best-performing model for stroke prediction achieving an AUC of 69.54 (95% CI = 67.40, 71.68), NRI of 0.202 (95% CI = 0.12, 0.28; p-value = 0.000) and IDI of 1.0 × 10-4 (95% CI = 0.000, 3.0 × 10-4; p-value = 0.13) compared with the Cox model without genetic liability. Conclusions: Incorporating genetic liability in prediction models slightly improved prediction models of stroke beyond conventional risk factors.

Predicting survival in small cell lung cancer patients undergoing various treatments: a machine learning approach

Background

Small cell lung cancer (SCLC) is highly metastatic, accounting for 1.796 million global cancer-related deaths in 2020, with no established standard care. This study aimed to assess treatment effects on SCLC patient survival across stages and develop a machine learning-based survival prediction tool for accurate overall survival (OS) estimation.

Methods

We developed four prediction models: Cox proportional hazard (Cox PH) regression, survival tree (ST), random survival forest (RSF), and gradient boosting survival analysis (GBSA). Patients were randomly split 7:3 into training and test datasets, with 10-fold cross-validation and 50 iterations on the training dataset. Cox PH used hazard ratios, while the other models employed importance values to assess feature predictiveness. Harrell's C-index (C-index) and Brier score (BS) measured model performance, with internal validations using R version 4.2.0.

Results

Cox PH outperformed others based on mean C-index and BS. Multivariate analysis across models highlighted distant metastases (M category), tumor stage, and treatment modalities (radiotherapy, chemotherapy, surgery) as key survival predictors. Stratified Cox PH analysis revealed surgery's efficacy in early-stage SCLC (stage II) and radiotherapy's advantage in stage III. Homogeneity was observed in chemotherapy benefits across cancer stages.

Conclusions

Surgery, chemotherapy, and radiotherapy are integral in SCLC treatment, contingent on cancer stage and characteristics. Surgery offers promise for early-stage cases, while advanced-stage strategies require further exploration.

Penalized landmark supermodels (penLM) for dynamic prediction for time-to-event outcomes in high-dimensional data

BACKGROUND

To effectively monitor long-term outcomes among cancer patients, it is critical to accurately assess patients' dynamic prognosis, which often involves utilizing multiple data sources (e.g., tumor registries, treatment histories, and patient-reported outcomes). However, challenges arise in selecting features to predict patient outcomes from high-dimensional data, aligning longitudinal measurements from multiple sources, and evaluating dynamic model performance.

METHODS

We provide a framework for dynamic risk prediction using the penalized landmark supermodel (penLM) and develop novel metrics ([Formula: see text] and [Formula: see text]) to evaluate and summarize model performance across different timepoints. Through simulations, we assess the coverage of the proposed metrics' confidence intervals under various scenarios. We applied penLM to predict the updated 5-year risk of lung cancer mortality at diagnosis and for subsequent years by combining data from SEER registries (2007-2018), Medicare claims (2007-2018), Medicare Health Outcome Survey (2006-2018), and U.S. Census (1990-2010).

RESULTS

The simulations confirmed valid coverage (~ 95%) of the confidence intervals of the proposed summary metrics. Of 4,670 lung cancer patients, 41.5% died from lung cancer. Using penLM, the key features to predict lung cancer mortality included long-term lung cancer treatments, minority races, regions with low education attainment or racial segregation, and various patient-reported outcomes beyond cancer staging and tumor characteristics. When evaluated using the proposed metrics, the penLM model developed using multi-source data ([Formula: see text]of 0.77 [95% confidence interval: 0.74-0.79]) outperformed those developed using single-source data ([Formula: see text]range: 0.50-0.74).

CONCLUSIONS

The proposed penLM framework with novel evaluation metrics offers effective dynamic risk prediction when leveraging high-dimensional multi-source longitudinal data.

Classification-augmented survival estimation (CASE): A novel method for individualized long-term survival prediction with application to liver transplantation

Survival analysis is critical in many fields, particularly in healthcare where it can guide medical decisions. Conventional survival analysis methods like Kaplan-Meier and Cox proportional hazards models to generate survival curves indicating probability of survival v. time have limitations, especially for long-term prediction, due to assumptions that all instances follow a general population-level survival curve. Machine learning classification models, even those designed for survival predictions like random survival forest (RSF), also struggle to provide accurate long-term predictions due to class imbalance. We improve upon traditional survival machine learning approaches through a novel framework called classification-augmented survival estimation (CASE), which treats survival as a classification task that ultimately yields survival curves, beginning with dataset augmentation to improve class imbalance for use with any classification model. Unlike other approaches, CASE additionally provides an exact survival time prediction. We demonstrate CASE on a liver transplant case study to predict >20 years survival post-transplant, finding that CASE dataset augmentation improved AUCs from 0.69 to 0.88 and F1 scores from 0.32 to 0.73. Compared to Kaplan-Meier, Cox, and RSF survival models, the CASE framework demonstrated better performance across various existing survival metrics, as well as our novel metric, mean of individual areas under the survival curve (mAUSC). Further, we develop novel temporal feature importance methods to understand how different features may vary in survival importance over time, potentially providing actionable insights in real-world survival problems.

Machine Learning-Based Prediction of Death and Hospitalization in Patients With Implantable Cardioverter Defibrillators

BACKGROUND

Predicting the clinical trajectory of individual patients with implantable cardioverter-defibrillators (ICDs) is essential to inform clinical care. Machine learning approaches can potentially overcome the limitations of conventional statistical methods and provide more accurate, personalized risk estimates.

OBJECTIVES

The authors sought to develop and externally validate a novel machine learning algorithm for predicting all-cause mortality and/or heart failure (HF) hospitalization in ICD patients with and without cardiac resynchronization therapy (CRT) using variables that are readily available to treating clinicians. We also sought to identify key factors that separate patients along a continuum of risk.

METHODS

Random forest for survival, longitudinal, and multivariate (RF-SLAM) data analysis was applied to predict 3-month and 1-year risks for all-cause mortality and a composite outcome of death/HF hospitalization during the first 5 years of device implant. Models were trained using a nationwide cohort from the Veterans Health Administration. Three models were sequentially tested, and external validation was performed in a separate nonveteran clinical registry.

RESULTS

The training and validation cohorts included 12,043 patients (age 67.5 ± 9.4 years) and 1,394 patients (age 66.3 ± 11.9 years), respectively. Median follow-up was 3.3 years for the training cohort and 3.6 years for validation cohort. The most accurate models for both outcomes included baseline demographics entered at the time of ICD implant (age, sex, CRT therapy) and time-varying ICD data with area under the receiver-operating characteristic curve for predicting death at 3 months (0.91; 95% CI: 0.87-0.94) and 1 year (0.80; 95% CI: 0.78-0.82); death/HF hospitalization at 3 months (0.81; 95% CI: 0.79-0.83) and 1 year (0.71; 95% CI: 0.70-0.72). Models demonstrated high discrimination and good calibration in the validation cohort. Additionally, time-varying physiologic data from ICDs, especially daily physical activity, had substantial importance in predicting outcomes.

CONCLUSIONS

The RF-SLAM algorithm accurately predicted all-cause mortality and death/HF hospitalization at 3 months and 1 year during the first 5 years of device implant, demonstrating good internal and external validity. Prospective studies and randomized trials are needed to evaluate model performance in other populations and settings and to determine its impact on patient outcomes.

Application of machine learning in breast cancer survival prediction using a multimethod approach

Breast cancer is one of the most prevalent cancers with an increasing trend in both incidence and mortality rates in Iran. Survival analysis is a pivotal measure in setting appropriate care plans.  To the best of our knowledge, this study is pioneering in Iran, introducing a multi-method approach using a Deep Neural Network (DNN) and 11 conventional machine learning (ML) methods to predict the 5 year survival of women with breast cancer. Supplying data from two centers comprising a total of 2644 records and incorporating external validation further distinguishes the study. Thirty-four features were selected based on a literature review and common variables in both datasets. Feature selection was also performed using a p value criterion (< 0.05) and a survey involving oncologists. A total of 108 models were trained. According to external validation, the DNN model trained with the Shiraz dataset, considering all features, exhibited the highest accuracy (85.56%). While the DNN model showed superior accuracy in external validation, it did not consistently achieve the highest performance across all evaluation metrics. Notably, models trained with the Shiraz dataset outperformed those trained with the Tehran dataset, possibly due to the lower number of missing values in the Shiraz dataset.

A comparison of computational algorithms for the Bayesian analysis of clinical trials

BACKGROUND

Clinical trials are increasingly using Bayesian methods for their design and analysis. Inference in Bayesian trials typically uses simulation-based approaches such as Markov Chain Monte Carlo methods. Markov Chain Monte Carlo has high computational cost and can be complex to implement. The Integrated Nested Laplace Approximations algorithm provides approximate Bayesian inference without the need for computationally complex simulations, making it more efficient than Markov Chain Monte Carlo. The practical properties of Integrated Nested Laplace Approximations compared to Markov Chain Monte Carlo have not been considered for clinical trials. Using data from a published clinical trial, we aim to investigate whether Integrated Nested Laplace Approximations is a feasible and accurate alternative to Markov Chain Monte Carlo and provide practical guidance for trialists interested in Bayesian trial design.

METHODS

Data from an international Bayesian multi-platform adaptive trial that compared therapeutic-dose anticoagulation with heparin to usual care in non-critically ill patients hospitalized for COVID-19 were used to fit Bayesian hierarchical generalized mixed models. Integrated Nested Laplace Approximations was compared to two Markov Chain Monte Carlo algorithms, implemented in the software JAGS and stan, using packages available in the statistical software R. Seven outcomes were analysed: organ-support free days (an ordinal outcome), five binary outcomes related to survival and length of hospital stay, and a time-to-event outcome. The posterior distributions for the treatment and sex effects and the variances for the hierarchical effects of age, site and time period were obtained. We summarized these posteriors by calculating the mean, standard deviations and the 95% equitailed credible intervals and presenting the results graphically. The computation time for each algorithm was recorded.

RESULTS

The average overlap of the 95% credible interval for the treatment and sex effects estimated using Integrated Nested Laplace Approximations was 96% and 97.6% compared with stan, respectively. The graphical posterior densities for these effects overlapped for all three algorithms. The posterior mean for the variance of the hierarchical effects of age, site and time estimated using Integrated Nested Laplace Approximations are within the 95% credible interval estimated using Markov Chain Monte Carlo but the average overlap of the credible interval is lower, 77%, 85.6% and 91.3%, respectively, for Integrated Nested Laplace Approximations compared to stan. Integrated Nested Laplace Approximations and stan were easily implemented in clear, well-established packages in R, while JAGS required the direct specification of the model. Integrated Nested Laplace Approximations was between 85 and 269 times faster than stan and 26 and 1852 times faster than JAGS.

CONCLUSION

Integrated Nested Laplace Approximations could reduce the computational complexity of Bayesian analysis in clinical trials as it is easy to implement in R, substantially faster than Markov Chain Monte Carlo methods implemented in JAGS and stan, and provides near identical approximations to the posterior distributions for the treatment effect. Integrated Nested Laplace Approximations was less accurate when estimating the posterior distribution for the variance of hierarchical effects, particularly for the proportional odds model, and future work should determine if the Integrated Nested Laplace Approximations algorithm can be adjusted to improve this estimation.

Prediction models for identifying medication overuse or medication overuse headache in migraine patients: a systematic review

BACKGROUND

Migraine is a debilitating neurological disorder that presents significant management challenges, resulting in underdiagnosis and inappropriate treatments, leaving patients at risk of medication overuse (MO). MO contributes to disease progression and the development of medication overuse headache (MOH). Predicting which migraine patients are at risk of MO/MOH is crucial for effective management. Thus, this systematic review aims to review and critique available prediction models for MO/MOH in migraine patients.

METHODS

A systematic search was conducted using Embase, Scopus, Medline/PubMed, ACM Digital Library, and IEEE databases from inception to April 22, 2024. The risk of bias was assessed using the prediction model risk of bias assessment tool.

RESULTS

Out of 1,579 articles, six studies with nine models met the inclusion criteria. Three studies developed new prediction models, while the remaining validated existing scores. Most studies utilized cross-sectional and prospective data collection in specific headache settings and migraine types. The models included up to 53 predictors, with sample sizes from 17 to 1,419 participants. Traditional statistical models (logistic regression and least absolute shrinkage and selection operator regression) were used in two studies, while one utilized a machine learning (ML) technique (support vector machines). Receiver operating characteristic analysis was employed to validate existing scores. The area under the receiver operating characteristic (AUROC) for the ML model (0.83) outperformed the traditional statistical model (0.62) in internal validation. The AUROCs ranged from 0.84 to 0.85 for the validation of existing scores. Common predictors included age and gender; genetic data and questionnaire evaluations were also included. All studies demonstrated a high risk of bias in model construction and high concerns regarding applicability to participants.

CONCLUSION

This review identified promising results for MO/MOH prediction models in migraine patients, although the field remains limited. Future research should incorporate important risk factors, assess discrimination and calibration, and perform external validation. Further studies with robust designs, appropriate settings, high-quality and quantity data, and rigorous methodologies are necessary to advance this field.

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.

Enhancing diagnostic accuracy in symptom-based health checkers: a comprehensive machine learning approach with clinical vignettes and benchmarking

Introduction

The development of machine learning models for symptom-based health checkers is a rapidly evolving area with significant implications for healthcare. Accurate and efficient diagnostic tools can enhance patient outcomes and optimize healthcare resources. This study focuses on evaluating and optimizing machine learning models using a dataset of 10 diseases and 9,572 samples.

Methods

The dataset was divided into training and testing sets to facilitate model training and evaluation. The following models were selected and optimized: Decision Tree, Random Forest, Naive Bayes, Logistic Regression and K-Nearest Neighbors. Evaluation metrics included accuracy, F1 scores, and 10-fold cross-validation. ROC-AUC and precision-recall curves were also utilized to assess model performance, particularly in scenarios with imbalanced datasets. Clinical vignettes were employed to gauge the real-world applicability of the models.

Results

The performance of the models was evaluated using accuracy, F1 scores, and 10-fold cross-validation. The use of ROC-AUC curves revealed that model performance improved with increasing complexity. Precision-recall curves were particularly useful in evaluating model sensitivity in imbalanced dataset scenarios. Clinical vignettes demonstrated the robustness of the models in providing accurate diagnoses.

Discussion

The study underscores the importance of comprehensive model evaluation techniques. The use of clinical vignette testing and analysis of ROC-AUC and precision-recall curves are crucial in ensuring the reliability and sensitivity of symptom-based health checkers. These techniques provide a more nuanced understanding of model performance and highlight areas for further improvement.

Conclusion

This study highlights the significance of employing diverse evaluation metrics and methods to ensure the robustness and accuracy of machine learning models in symptom-based health checkers. The integration of clinical vignettes and the analysis of ROC-AUC and precision-recall curves are essential steps in developing reliable and sensitive diagnostic tools.

An inverse lomax-uniform poisson distribution and joint modeling of repeatedly measured and time-to-event data in the health sectors

Methodological developments in different sectors like health, biomedical and biological areas are the recent burning issue in the statistical literature. The approach of implementing declining hazard function which is obtained by compounding truncated Poisson distribution and a lifetime distribution is a special concern in a few studies. In this paper we proposed a newly introduced distribution called inverse Lomax-Uniform Poisson distribution mostly applied in the health, biomedical, biological, and related sectors. Some basic statistical properties of the distribution are discussed. The capability of the model is checked by comparing it with six potential models by using a practical real data set. Based on the comparison techniques, the newly proposed model out performs all its counterparts. The simulation study is also conducted. Furthermore, the joint modelling of repeatedly measured and time-to-vent processes is discussed in detail with the real data set in the health sector.

Development of an AI-Based Predictive Algorithm for Early Diagnosis of High-Risk Dementia Groups among the Elderly: Utilizing Health Lifelog Data

BACKGROUND/OBJECTIVES

This study aimed to develop a predictive algorithm for the early diagnosis of dementia in the high-risk group of older adults using artificial intelligence technologies. The objective is to create an accessible diagnostic method that does not rely on traditional medical equipment, thereby improving the early detection and management of dementia.

METHODS

Lifelog data from wearable devices targeting this high-risk group were collected from the AI Hub platform. Various indicators from these data were analyzed to develop a dementia diagnostic model. Machine learning techniques such as Logistic Regression, Random Forest, LightGBM, and Support Vector Machine were employed. Data augmentation techniques were applied to address data imbalance, thereby enhancing the model performance.

RESULTS

Data augmentation significantly improved the model's accuracy in classifying dementia cases. Specifically, in gait data, the SVM model performed with an accuracy of 0.879. In sleep data, a Logistic Regression was performed, yielding an accuracy of 0.818. This indicates that the lifelog data can effectively contribute to the early diagnosis of dementia, providing a practical solution that can be easily integrated into healthcare systems.

CONCLUSIONS

This study demonstrates that lifelog data, which are easily collected in daily life, can significantly enhance the accessibility and efficiency of dementia diagnosis, aiding in the effective use of medical resources and potentially delaying disease progression.

Interpretable artificial intelligence to optimise use of imatinib after resection in patients with localised gastrointestinal stromal tumours: an observational cohort study

BACKGROUND

Current guidelines recommend use of adjuvant imatinib therapy for many patients with gastrointestinal stromal tumours (GISTs); however, its optimal treatment duration is unknown and some patient groups do not benefit from the therapy. We aimed to apply state-of-the-art, interpretable artificial intelligence (ie, predictions or prescription logic that can be easily understood) methods on real-world data to establish which groups of patients with GISTs should receive adjuvant imatinib, its optimal treatment duration, and the benefits conferred by this therapy.

METHODS

In this observational cohort study, we considered for inclusion all patients who underwent resection of primary, non-metastatic GISTs at the Memorial Sloan Kettering Cancer Center (MSKCC; New York, NY, USA) between Oct 1, 1982, and Dec 31, 2017, and who were classified as intermediate or high risk according to the Armed Forces Institute of Pathology Miettinen criteria and had complete follow-up data with no missing entries. A counterfactual random forest model, which used predictors of recurrence (mitotic count, tumour size, and tumour site) and imatinib duration to infer the probability of recurrence at 7 years for a given patient under each duration of imatinib treatment, was trained in the MSKCC cohort. Optimal policy trees (OPTs), a state-of-the-art interpretable AI-based method, were used to read the counterfactual random forest model by training a decision tree with the counterfactual predictions. The OPT recommendations were externally validated in two cohorts of patients from Poland (the Polish Clinical GIST Registry), who underwent GIST resection between Dec 1, 1981, and Dec 31, 2011, and from Spain (the Spanish Group for Research in Sarcomas), who underwent resection between Oct 1, 1987, and Jan 30, 2011.

FINDINGS

Among 1007 patients who underwent GIST surgery in MSKCC, 117 were included in the internal cohort; for the external cohorts, the Polish cohort comprised 363 patients and the Spanish cohort comprised 239 patients. The OPT did not recommend imatinib for patients with GISTs of gastric origin measuring less than 15·9 cm with a mitotic count of less than 11·5 mitoses per 5 mm2 or for those with small GISTs (<5·4 cm) of any site with a count of less than 11·5 mitoses per 5 mm2. In this cohort, the OPT cutoffs had a sensitivity of 92·7% (95% CI 82·4-98·0) and a specificity of 33·9% (22·3-47·0). The application of these cutoffs in the two external cohorts would have spared 38 (29%) of 131 patients in the Spanish cohort and 44 (35%) of 126 patients in the Polish cohort from unnecessary treatment with imatinib. Meanwhile, the risk of undertreating patients in these cohorts was minimal (sensitivity 95·4% [95% CI 89·5-98·5] in the Spanish cohort and 92·4% [88·3-95·4] in the Polish cohort). The OPT tested 33 different durations of imatinib treatment (<5 years) and found that 5 years of treatment conferred the most benefit.

INTERPRETATION

If the identified patient subgroups were applied in clinical practice, as many as a third of the current cohort of candidates who do not benefit from adjuvant imatinib would be encouraged to not receive imatinib, subsequently avoiding unnecessary toxicity on patients and financial strain on health-care systems. Our finding that 5 years is the optimal duration of imatinib treatment could be the best source of evidence to inform clinical practice until 2028, when a randomised controlled trial with the same aims is expected to report its findings.

FUNDING

National Cancer Institute.

Random survival forest for predicting the combined effects of multiple physiological risk factors on all-cause mortality

Understanding the combined effects of risk factors on all-cause mortality is crucial for implementing effective risk stratification and designing targeted interventions, but such combined effects are understudied. We aim to use survival-tree based machine learning models as more flexible nonparametric techniques to examine the combined effects of multiple physiological risk factors on mortality. More specifically, we (1) study the combined effects between multiple physiological factors and all-cause mortality, (2) identify the five most influential factors and visualize their combined influence on all-cause mortality, and (3) compare the mortality cut-offs with the current clinical thresholds. Data from the 1999-2014 NHANES Survey were linked to National Death Index data with follow-up through 2015 for 17,790 adults. We observed that the five most influential factors affecting mortality are the tobacco smoking biomarker cotinine, glomerular filtration rate (GFR), plasma glucose, sex, and white blood cell count. Specifically, high mortality risk is associated with being male, active smoking, low GFR, elevated plasma glucose levels, and high white blood cell count. The identified mortality-based cutoffs for these factors are mostly consistent with relevant studies and current clinical thresholds. This approach enabled us to identify important cutoffs and provide enhanced risk prediction as an important basis to inform clinical practice and develop new strategies for precision medicine.

Dynamic Prediction of Post-Acute Care Needs for Hospitalized Medicine Patients

OBJECTIVES

Use patient demographic and clinical characteristics at admission and time-varying in-hospital measures of patient mobility to predict patient post-acute care (PAC) discharge.

DESIGN

Retrospective cohort analysis of electronic medical records.

SETTING AND PARTICIPANTS

Patients admitted to the two participating Hospitals from November 2016 through December 2019 with ≥72 hours in a general medicine service.

METHODS

Discharge location (PAC vs home) was the primary outcome, and 2 time-varying measures of patient mobility, Activity Measure for Post-Acute Care (AM-PAC) Mobility "6-clicks" and Johns Hopkins Highest Level of Mobility, were the primary predictors. Other predictors included demographic and clinical characteristics. For each day of hospitalization, we predicted discharge to PAC using the demographic and clinical characteristics and most recent mobility data within a random forest (RF) for survival, longitudinal, and multivariate (RF-SLAM) data. A regression tree for the daily predicted probabilities of discharge to PAC was constructed to represent a global summary of the RF.

RESULTS

There were 23,090 total patients and compared to PAC, those discharged home were younger (64 vs 71), had shorter length of stay (5 vs 8 days), higher AM-PAC at admission (43 vs 32), and average AM-PAC throughout hospitalization (45 vs 35). AM-PAC was the most important predictor, followed by age, and whether the patient lives alone. The area under the hospital day-specific receiver operating characteristic curve ranged from 0.76 to 0.79 during the first 5 days. The global summary tree explained 75% of the variation in predicted probabilities for PAC from the RF. Sensitivity (75%), specificity (70%), and accuracy (72%) were maximized at a PAC probability threshold of 40%.

CONCLUSIONS AND IMPLICATIONS

Daily assessment of patient mobility should be part of routine practice to help inform care planning by hospital teams. Our prediction model could be used as a valuable tool by multidisciplinary teams in the discharge planning process.

Explainable text-tabular models for predicting mortality risk in companion animals

As interest in using machine learning models to support clinical decision-making increases, explainability is an unequivocal priority for clinicians, researchers and regulators to comprehend and trust their results. With many clinical datasets containing a range of modalities, from the free-text of clinician notes to structured tabular data entries, there is a need for frameworks capable of providing comprehensive explanation values across diverse modalities. Here, we present a multimodal masking framework to extend the reach of SHapley Additive exPlanations (SHAP) to text and tabular datasets to identify risk factors for companion animal mortality in first-opinion veterinary electronic health records (EHRs) from across the United Kingdom. The framework is designed to treat each modality consistently, ensuring uniform and consistent treatment of features and thereby fostering predictability in unimodal and multimodal contexts. We present five multimodality approaches, with the best-performing method utilising PetBERT, a language model pre-trained on a veterinary dataset. Utilising our framework, we shed light for the first time on the reasons each model makes its decision and identify the inclination of PetBERT towards a more pronounced engagement with free-text narratives compared to BERT-base's predominant emphasis on tabular data. The investigation also explores the important features on a more granular level, identifying distinct words and phrases that substantially influenced an animal's life status prediction. PetBERT showcased a heightened ability to grasp phrases associated with veterinary clinical nomenclature, signalling the productivity of additional pre-training of language models.

Using Bayesian statistics in confirmatory clinical trials in the regulatory setting: a tutorial review

Bayesian statistics plays a pivotal role in advancing medical science by enabling healthcare companies, regulators, and stakeholders to assess the safety and efficacy of new treatments, interventions, and medical procedures. The Bayesian framework offers a unique advantage over the classical framework, especially when incorporating prior information into a new trial with quality external data, such as historical data or another source of co-data. In recent years, there has been a significant increase in regulatory submissions using Bayesian statistics due to its flexibility and ability to provide valuable insights for decision-making, addressing the modern complexity of clinical trials where frequentist trials are inadequate. For regulatory submissions, companies often need to consider the frequentist operating characteristics of the Bayesian analysis strategy, regardless of the design complexity. In particular, the focus is on the frequentist type I error rate and power for all realistic alternatives. This tutorial review aims to provide a comprehensive overview of the use of Bayesian statistics in sample size determination, control of type I error rate, multiplicity adjustments, external data borrowing, etc., in the regulatory environment of clinical trials. Fundamental concepts of Bayesian sample size determination and illustrative examples are provided to serve as a valuable resource for researchers, clinicians, and statisticians seeking to develop more complex and innovative designs.

New genetic insights into immunotherapy outcomes in gastric cancer via single-cell RNA sequencing and random forest model

OBJECTIVE

The high mortality rate of gastric cancer, traditionally managed through surgery, underscores the urgent need for advanced therapeutic strategies. Despite advancements in treatment modalities, outcomes remain suboptimal, necessitating the identification of novel biomarkers to predict sensitivity to immunotherapy. This study focuses on utilizing single-cell sequencing for gene identification and developing a random forest model to predict immunotherapy sensitivity in gastric cancer patients.

METHODS

Differentially expressed genes were identified using single-cell RNA sequencing (scRNA-seq) and gene set enrichment analysis (GESA). A random forest model was constructed based on these genes, and its effectiveness was validated through prognostic analysis. Further, analyses of immune cell infiltration, immune checkpoints, and the random forest model provided deeper insights.

RESULTS

High METTL1 expression was found to correlate with improved survival rates in gastric cancer patients (P = 0.042), and the random forest model, based on METTL1 and associated prognostic genes, achieved a significant predictive performance (AUC = 0.863). It showed associations with various immune cell types and negative correlations with CTLA4 and PDCD1 immune checkpoints. Experiments in vitro and in vivo demonstrated that METTL1 enhances gastric cancer cell activity by suppressing T cell proliferation and upregulating CTLA4 and PDCD1.

CONCLUSION

The random forest model, based on scRNA-seq, shows high predictive value for survival and immunotherapy sensitivity in gastric cancer patients. This study underscores the potential of METTL1 as a biomarker in enhancing the efficacy of gastric cancer immunotherapy.

Dynamic prediction of malignant ventricular arrhythmias using neural networks in patients with an implantable cardioverter-defibrillator

BACKGROUND

Risk stratification for ventricular arrhythmias currently relies on static measurements that fail to adequately capture dynamic interactions between arrhythmic substrate and triggers over time. We trained and internally validated a dynamic machine learning (ML) model and neural network that extracted features from longitudinally collected electrocardiograms (ECG), and used these to predict the risk of malignant ventricular arrhythmias.

METHODS

A multicentre study in patients implanted with an implantable cardioverter-defibrillator (ICD) between 2007 and 2021 in two academic hospitals was performed. Variational autoencoders (VAEs), which combine neural networks with variational inference principles, and can learn patterns and structure in data without explicit labelling, were trained to encode the mean ECG waveforms from the limb leads into 16 variables. Supervised dynamic ML models using these latent ECG representations and clinical baseline information were trained to predict malignant ventricular arrhythmias treated by the ICD. Model performance was evaluated on a hold-out set, using time-dependent receiver operating characteristic (ROC) and calibration curves.

FINDINGS

2942 patients (61.7 ± 13.9 years, 25.5% female) were included, with a total of 32,129 ECG recordings during a mean follow-up of 43.9 ± 35.9 months. The mean time-varying area under the ROC curve for the dynamic model was 0.738 ± 0.07, compared to 0.639 ± 0.03 for a static (i.e. baseline-only model). Feature analyses indicated dynamic changes in latent ECG representations, particularly those affecting the T-wave morphology, were of highest importance for model predictions.

INTERPRETATION

Dynamic ML models and neural networks effectively leverage routinely collected longitudinal ECG recordings for personalised and updated predictions of malignant ventricular arrhythmias, outperforming static models.

FUNDING

This publication is part of the project DEEP RISK ICD (with project number 452019308) of the research programme Rubicon which is (partly) financed by the Dutch Research Council (NWO). This research is partly funded by the Amsterdam Cardiovascular Sciences (personal grant F.V.Y.T).

Use of survival support vector machine combined with random survival forest to predict the survival of nasopharyngeal carcinoma patients

Background

The Cox regression model is not sufficiently accurate to predict the survival prognosis of nasopharyngeal carcinoma (NPC) patients. It is impossible to calculate and rank the importance of impact factors due to the low predictive accuracy of the Cox regression model. So, we developed a system. Using the SEER (The Surveillance, Epidemiology, and End Results) database data on NPC patients, we proposed the use of random survival forest (RSF) and survival-support vector machine (SVM) from the machine learning methods to develop a survival prediction system specifically for NPC patients. This approach aimed to make up for the insufficiency of the Cox regression model. We also used the Cox regression model to validate the development of the nomogram and compared it with machine learning methods.

Methods

A total of 1,683 NPC patients were extracted from the SEER database from January 2010 to December 2015. We used R language for modeling work, established the nomogram of survival prognosis of NPC patients by Cox regression model, ranked the correlation of influencing factors by RSF model VIMP (variable important) method, developed a survival prognosis system for NPC patients based on survival-SVM, and used C-index for model evaluation and performance comparison.

Results

Although the Cox regression models can be developed to predict the prognosis of NPC patients, their accuracy was lower than that of machine learning methods. When we substituted the data for the Cox model, the C-index for the training set was only 0.740, and the C-index for the test set was 0.721. In contrast, the C index of the survival-SVM model was 0.785. The C-index of the RSF model was 0.729. The importance ranking of each variable could be obtained according to the VIMP method.

Conclusions

The prediction results from the Cox model are not as good as those of the RSF method and survival-SVM based on the machine learning method. For the survival prognosis of NPC patients, the machine learning method can be considered for clinical application.

Futility monitoring for randomized clinical trials with non-proportional hazards: An optimal conditional power approach

BACKGROUND

Standard futility analyses designed for a proportional hazards setting may have serious drawbacks when non-proportional hazards are present. One important type of non-proportional hazards occurs when the treatment effect is delayed. That is, there is little or no early treatment effect but a substantial later effect.

METHODS

We define optimality criteria for futility analyses in this setting and propose simple search procedures for deriving such rules in practice.

RESULTS

We demonstrate the advantages of the optimal rules over commonly used rules in reducing the average number of events, the average sample size, or the average study duration under the null hypothesis with minimal power loss under the alternative hypothesis.

CONCLUSION

Optimal futility rules can be derived for a non-proportional hazards setting that control the loss of power under the alternative hypothesis while maximizing the gain in early stopping under the null hypothesis.

Application of machine learning in predicting survival outcomes involving real-world data: a scoping review

BACKGROUND

Despite the interest in machine learning (ML) algorithms for analyzing real-world data (RWD) in healthcare, the use of ML in predicting time-to-event data, a common scenario in clinical practice, is less explored. ML models are capable of algorithmically learning from large, complex datasets and can offer advantages in predicting time-to-event data. We reviewed the recent applications of ML for survival analysis using RWD in healthcare.

METHODS

PUBMED and EMBASE were searched from database inception through March 2023 to identify peer-reviewed English-language studies of ML models for predicting time-to-event outcomes using the RWD. Two reviewers extracted information on the data source, patient population, survival outcome, ML algorithms, and the Area Under the Curve (AUC).

RESULTS

Of 257 citations, 28 publications were included. Random survival forests (N = 16, 57%) and neural networks (N = 11, 39%) were the most popular ML algorithms. There was variability across AUC for these ML models (median 0.789, range 0.6-0.950). ML algorithms were predominately considered for predicting overall survival in oncology (N = 12, 43%). ML survival models were often used to predict disease prognosis or clinical events (N = 27, 96%) in the oncology, while less were used for treatment outcomes (N = 1, 4%).

CONCLUSIONS

The ML algorithms, random survival forests and neural networks, are mainly used for RWD to predict survival outcomes such as disease prognosis or clinical events in the oncology. This review shows that more opportunities remain to apply these ML algorithms to inform treatment decision-making in clinical practice. More methodological work is also needed to ensure the utility and applicability of ML models in survival outcomes.

Students learning performance prediction based on feature extraction algorithm and attention-based bidirectional gated recurrent unit network

With the development of information technology construction in schools, predicting student grades has become a hot area of application in current educational research. Using data mining to analyze the influencing factors of students' performance and predict their grades can help students identify their shortcomings, optimize teachers' teaching methods and enable parents to guide their children's progress. However, there are no models that can achieve satisfactory predictions for education-related public datasets, and most of these weakly correlated factors in the datasets can still adversely affect the predictive effect of the model. To solve this issue and provide effective policy recommendations for the modernization of education, this paper seeks to find the best grade prediction model based on data mining. Firstly, the study uses the Factor Analyze (FA) model to extract features from the original data and achieve dimension reduction. Then, the Bidirectional Gate Recurrent Unit (BiGRU) model and attention mechanism are utilized to predict grades. Lastly, Comparing the prediction results of ablation experiments and other single models, such as linear regression (LR), back propagation neural network (BP), random forest (RF), and Gate Recurrent Unit (GRU), the FA-BiGRU-attention model achieves the best prediction effect and performs equally well in different multi-step predictions. Previously, problems with students' grades were only detected when they had already appeared. However, the methods presented in this paper enable the prediction of students' learning in advance and the identification of factors affecting their grades. Therefore, this study has great potential to provide data support for the improvement of educational programs, transform the traditional education industry, and ensure the sustainable development of national talents.

Optimization of sports effect evaluation technology from random forest algorithm and elastic network algorithm

This study leverages advanced data mining and machine learning techniques to delve deeper into the impact of sports activities on physical health and provide a scientific foundation for informed sports selection and health promotion. Guided by the Elastic Net algorithm, a sports performance assessment model is meticulously constructed. In contrast to the conventional Least Absolute Shrinkage and Selection Operator (Lasso) algorithm, this model seeks to elucidate the factors influencing physical health indicators due to sports activities. Additionally, the incorporation of the Random Forest algorithm facilitates a comprehensive evaluation of sports performance across distinct dimensions: wrestling-type sports, soccer-type sports, skill-based sports, and school physical education. Employing the Top-K criterion for evaluation and juxtaposing it with the high-performance Support Vector Machine (SVM) algorithm, the accuracy is scrutinized under three distinct criteria: Top-3, Top-5, and Top-10. The pivotal innovation of this study resides in the amalgamation of the Elastic Net and Random Forest algorithms, permitting a holistic contemplation of the influencing factors of diverse sports activities on physical health indicators. Through this integrated methodology, the research achieves a more precise assessment of the effects of sports activities, unveiling a range of impacts various sports have on physical health. Consequently, a more refined assessment tool for sports performance detection and health development is established. Capitalizing on the Elastic Net algorithm, this research optimizes model construction during the pivotal feature selection phase, effectively capturing the crucial influencing factors associated with different sports activities. Concurrently, the integration of the Random Forest algorithm augments the predictive prowess of the model, enabling the sports performance assessment model to comprehensively unveil the extent of impact stemming from various sports activities. This study stands as a noteworthy contribution to the arena of sports performance assessment, offering substantial insights and advancements to both sports health and research methodologies.

An Artificial Intelligence Approach to Predicting Unplanned Intubation Following Anterior Cervical Discectomy and Fusion

STUDY DESIGN

Level III retrospective database study.

OBJECTIVES

The purpose of this study is to determine if machine learning algorithms are effective in predicting unplanned intubation following anterior cervical discectomy and fusion (ACDF).

METHODS

The National Surgical Quality Initiative Program (NSQIP) was queried to select patients who had undergone ACDF. Machine learning analysis was conducted in Python and multivariate regression analysis was conducted in R. C-Statistics area under the curve (AUC) and prediction accuracy were used to measure the classifier's effectiveness in distinguishing cases.

RESULTS

In total, 54 502 patients met the study criteria. Of these patients, .51% underwent an unplanned re-intubation. Machine learning algorithms accurately classified between 72%-100% of the test cases with AUC values of between .52-.77. Multivariable regression indicated that the number of levels fused, male sex, COPD, American Society of Anesthesiologists (ASA) > 2, increased operating time, Age > 65, pre-operative weight loss, dialysis, and disseminated cancer were associated with increased risk of unplanned intubation.

CONCLUSIONS

The models presented here achieved high accuracy in predicting risk factors for re-intubation following ACDF surgery. Machine learning analysis may be useful in identifying patients who are at a higher risk of unplanned post-operative re-intubation and their treatment plans can be modified to prophylactically prevent respiratory compromise and consequently unplanned re-intubation.

Derivation and validation of risk prediction for posttraumatic stress symptoms following trauma exposure

BACKGROUND

Posttraumatic stress symptoms (PTSS) are common following traumatic stress exposure (TSE). Identification of individuals with PTSS risk in the early aftermath of TSE is important to enable targeted administration of preventive interventions. In this study, we used baseline survey data from two prospective cohort studies to identify the most influential predictors of substantial PTSS.

METHODS

Self-identifying black and white American women and men (n = 1546) presenting to one of 16 emergency departments (EDs) within 24 h of motor vehicle collision (MVC) TSE were enrolled. Individuals with substantial PTSS (⩾33, Impact of Events Scale - Revised) 6 months after MVC were identified via follow-up questionnaire. Sociodemographic, pain, general health, event, and psychological/cognitive characteristics were collected in the ED and used in prediction modeling. Ensemble learning methods and Monte Carlo cross-validation were used for feature selection and to determine prediction accuracy. External validation was performed on a hold-out sample (30% of total sample).

RESULTS

Twenty-five percent (n = 394) of individuals reported PTSS 6 months following MVC. Regularized linear regression was the top performing learning method. The top 30 factors together showed good reliability in predicting PTSS in the external sample (Area under the curve = 0.79 ± 0.002). Top predictors included acute pain severity, recovery expectations, socioeconomic status, self-reported race, and psychological symptoms.

CONCLUSIONS

These analyses add to a growing literature indicating that influential predictors of PTSS can be identified and risk for future PTSS estimated from characteristics easily available/assessable at the time of ED presentation following TSE.

Artificial intelligence to investigate predictors and prognostic impact of time to surgery in colon cancer

BACKGROUND AND OBJECTIVES

The role of time to surgery (TTS) for long-term outcomes in colon cancer (CC) remains ill-defined. We sought to utilize artificial intelligence (AI) to characterize the drivers of TTS and its prognostic impact.

METHODS

The National Cancer Database was utilized to identify patients diagnosed with non-metastatic CC between 2004 and 2018. AI models were employed to rank the importance of several sociodemographic, facility, and tumor characteristics in determining TTS, and postoperative survival.

RESULTS

Among 518 983 patients, 137 902 (26.6%) received intraoperative diagnosis of CC (TTS = 0), while 381 081 (74.4%) underwent elective surgery (TTS > 0) with median TTS of 19.0 days (interquartile range [IQR]: 7.0-33.0). An AI model, identified tumor stage, receipt of adequate lymphadenectomy, histologic grade, lymphovascular invasion, and insurance status as the most important variables associated with TTS = 0. Conversely, the type and location of treating facility and receipt of adjuvant therapy were among the most important variables for TTS > 0. Notably, TTS was among the most important variables associated with survival, and TTS > 3 weeks was associated with an incremental increase in mortality risk.

CONCLUSIONS

The identification of factors associated with TTS can help stratify patients most likely to suffer poor outcomes due to prolonged TTS, as well as guide quality improvement initiatives related to timely surgical care.

Factors contributing to preventing operating room "never events": a machine learning analysis

BACKGROUND

A surgical "Never Event" is a preventable error occurring immediately before, during or immediately following surgery. Various factors contribute to the occurrence of major Never Events, but little is known about their quantified risk in relation to a surgery's characteristics. Our study uses machine learning to reveal and quantify risk factors with the goal of improving patient safety and quality of care.

METHODS

We used data from 9,234 observations on safety standards and 101 root-cause analyses from actual, major "Never Events" including wrong site surgery and retained foreign item, and three random forest supervised machine learning models to identify risk factors. Using a standard 10-cross validation technique, we evaluated the models' metrics, measuring their impact on the occurrence of the two types of Never Events through Gini impurity.

RESULTS

We identified 24 contributing factors in six surgical departments: two had an impact of > 900% in Urology, Orthopedics, and General Surgery; six had an impact of 0-900% in Gynecology, Urology, and Cardiology; and 17 had an impact of < 0%. Combining factors revealed 15-20 pairs with an increased probability in five departments: Gynecology, 875-1900%; Urology, 1900-2600%; Cardiology, 833-1500%; Orthopedics,1825-4225%; and General Surgery, 2720-13,600%. Five factors affected wrong site surgery's occurrence (-60.96 to 503.92%) and five affected retained foreign body (-74.65 to 151.43%): two nurses (66.26-87.92%), surgery length < 1 h (85.56-122.91%), and surgery length 1-2 h (-60.96 to 85.56%).

CONCLUSIONS

Using machine learning, we could quantify the risk factors' potential impact on wrong site surgeries and retained foreign items in relation to a surgery's characteristics, suggesting that safety standards should be adjusted to surgery's characteristics based on risk assessment in each operating room. .

TRIAL REGISTRATION NUMBER

MOH 032-2019.

Neural network based ensemble model to predict radiation induced lymphopenia after concurrent chemo-radiotherapy for non-small cell lung cancer from two institutions

The use of adjuvant Immune Checkpoint Inhibitors (ICI) after concurrent chemo-radiation therapy (CCRT) has become the standard of care for locally advanced non-small cell lung cancer (LA-NSCLC). However, prolonged radiotherapy regimens are known to cause radiation-induced lymphopenia (RIL), a long-neglected toxicity that has been shown to correlate with response to ICIs and survival of patients treated with adjuvant ICI after CCRT. In this study, we aim to develop a novel neural network (NN) approach that integrates patient characteristics, treatment related variables, and differential dose volume histograms (dDVH) of lung and heart to predict the incidence of RIL at the end of treatment. Multi-institutional data of 139 LA-NSCLC patients from two hospitals were collected for training and validation of our suggested model. Ensemble learning was combined with a bootstrap strategy to stabilize the model, which was evaluated internally using repeated cross validation. The performance of our proposed model was compared to conventional models using the same input features, such as Logistic Regression (LR) and Random Forests (RF), using the Area Under the Curve (AUC) of Receiver Operating Characteristics (ROC) curves. Our suggested model (AUC=0.77) outperformed the comparison models (AUC=0.72, 0.74) in terms of absolute performance, indicating that the convolutional structure of the network successfully abstracts additional information from the differential DVHs, which we studied using Gradient-weighted Class Activation Map. This study shows that clinical factors combined with dDVHs can be used to predict the risk of RIL for an individual patient and shows a path toward preventing lymphopenia using patient-specific modifications of the radiotherapy plan.

Segmental motor recovery after cervical spinal cord injury relates to density and integrity of corticospinal tract projections

Cervical spinal cord injury (SCI) causes extensive impairments for individuals which may include dextrous hand function. Although prior work has focused on the recovery at the person-level, the factors determining the recovery of individual muscles are poorly understood. Here, we investigate the muscle-specific recovery after cervical spinal cord injury in a retrospective analysis of 748 individuals from the European Multicenter Study about Spinal Cord Injury (NCT01571531). We show associations between corticospinal tract (CST) sparing and upper extremity recovery in SCI, which improves the prediction of hand muscle strength recovery. Our findings suggest that assessment strategies for muscle-specific motor recovery in acute spinal cord injury are improved by accounting for CST sparing, and complement person-level predictions.

On the use of historical estimates

The use of historical, i.e., already existing, estimates in current studies is common in a wide variety of application areas. Nevertheless, despite their routine use, the uncertainty associated with historical estimates is rarely properly accounted for in the analysis. In this communication, we review common practices and then provide a mathematical formulation and a principled frequentist methodology for addressing the problem of drawing inferences in the presence of historical estimates. Three distinct variants are investigated in detail; the corresponding limiting distributions are found and compared. The design of future studies, given historical data, is also explored and relations with a variety of other well-studied statistical problems discussed.

Forty years on: a new national study of hearing in England and implications for global hearing health policy

OBJECTIVE

We aimed to update the prevalence estimates of hearing loss in older adults in England using a nationally representative sample of adults aged 50 years old and older.

DESIGN

A comparative cross-sectional study design was implemented. Hearing loss was defined as ≥35 dB HL at 3.0 kHz, as measured via Hearcheck in the better-hearing ear.

STUDY SAMPLE

We compared the estimates based on the English census in 2015 to estimates from psychoacoustic hearing data available for 8,263 participants in the English Longitudinal Study of Ageing (ELSA) Wave 7 (2014-2015).

RESULTS

Marked regional variability in hearing loss prevalence was revealed among participants with similar age profiles. The regional differences in hearing outcomes reached up to 13.53% in those belonging to the 71-80 years old group; the prevalence of hearing loss was 49.22% in the North East of England (95%CI 48.0-50.4), versus 35.69% in the South East (95%CI 34.8-36.50).

CONCLUSION

A socio-spatial approach in planning sustainable models of hearing care based on the actual populations' needs and not on age demographics might offer a viable opportunity for healthier lives. Regular assessment of the extent and causality of the population's different audiological needs within the country is strongly supported.

Artificial intelligence and radiomics in evaluation of kidney lesions: a comprehensive literature review

Radiomics and artificial intelligence (AI) may increase the differentiation of benign from malignant kidney lesions, differentiation of angiomyolipoma (AML) from renal cell carcinoma (RCC), differentiation of oncocytoma from RCC, differentiation of different subtypes of RCC, to predict Fuhrman grade, to predict gene mutation through molecular biomarkers and to predict treatment response in metastatic RCC undergoing immunotherapy. Neural networks analyze imaging data. Statistical, geometrical, textural features derived are giving quantitative data of contour, internal heterogeneity and gray zone features of lesions. A comprehensive literature review was performed, until July 2022. Studies investigating the diagnostic value of radiomics in differentiation of renal lesions, grade prediction, gene alterations, molecular biomarkers and ongoing clinical trials have been analyzed. The application of AI and radiomics could lead to improved sensitivity, specificity, accuracy in detecting and differentiating between renal lesions. Standardization of scanner protocols will improve preoperative differentiation between benign, low-risk cancers and clinically significant renal cancers and holds the premises to enhance the diagnostic ability of imaging tools to characterize renal lesions.

The roles of individual differences in time perspective, promotion focus, and innovativeness: Testing technology acceptance model

The goal of this study is to examine the roles of Zimbardo's time perspective along with other individual differences such as promotion focus and innovativeness in perceived ease of use, perceived usefulness, and attitude toward SNSs (social networking sites) in the technology acceptance model (TAM). A total of 234 participants joined this online study in South Korea. As predicted, past positive time perspective (TP) positively affected promotion focus and innovativeness, whereas past negative TP negatively affected them. Present hedonic TP positively affected innovativeness, and present fatalistic TP negatively affected promotion focus each. Future TP also positively related to promotion focus and innovativeness. In addition, simple and serial mediation effects of perceived ease of use and perceived usefulness independently and sequentially mediated the impact of TP on attitude toward SNSs. By considering TP along with promotion focus and innovativeness in conjunction with beliefs in the TAM, this study identifies psychological underpinnings of how individual differences affect technology adoption attitude and behavior. Research implications and future research suggestions will be discussed in detail.

A review of risk concepts and models for predicting the risk of primary stroke

Predicting an individual's risk of primary stroke is an important tool that can help to lower the burden of stroke for both the individual and society. There are a number of risk models and risk scores in existence but no review or classification designed to help the reader better understand how models differ and the reasoning behind these differences. In this paper we review the existing literature on primary stroke risk prediction models. From our literature review we identify key similarities and differences in the existing models. We find that models can differ in a number of ways, including the event type, the type of analysis, the model type and the time horizon. Based on these similarities and differences we have created a set of questions and a system to help answer those questions that modelers and readers alike can use to help classify and better understand the existing models as well as help to make necessary decisions when creating a new model.

An efficient approach for optimizing the cost-effective individualized treatment rule using conditional random forest

Evidence from observational studies has become increasingly important for supporting healthcare policy making via cost-effectiveness analyses. Similar as in comparative effectiveness studies, health economic evaluations that consider subject-level heterogeneity produce individualized treatment rules that are often more cost-effective than one-size-fits-all treatment. Thus, it is of great interest to develop statistical tools for learning such a cost-effective individualized treatment rule under the causal inference framework that allows proper handling of potential confounding and can be applied to both trials and observational studies. In this paper, we use the concept of net-monetary-benefit to assess the trade-off between health benefits and related costs. We estimate cost-effective individualized treatment rule as a function of patients' characteristics that, when implemented, optimizes the allocation of limited healthcare resources by maximizing health gains while minimizing treatment-related costs. We employ the conditional random forest approach and identify the optimal cost-effective individualized treatment rule using net-monetary-benefit-based classification algorithms, where two partitioned estimators are proposed for the subject-specific weights to effectively incorporate information from censored individuals. We conduct simulation studies to evaluate the performance of our proposals. We apply our top-performing algorithm to the NIH-funded Systolic Blood Pressure Intervention Trial to illustrate the cost-effectiveness gains of assigning customized intensive blood pressure therapy.

Development and validation of a joint model for dynamic prediction of overall survival in nasopharyngeal carcinoma based on longitudinal post-treatment plasma cell-free Epstein-Barr virus DNA load

OBJECTIVES

To develop and validate a joint model for dynamic prediction of overall survival (OS) in nasopharyngeal carcinoma (NPC) based on longitudinal post-treatment plasma cell-free Epstein-Barr virus (cfEBV) DNA load.

PATIENTS AND METHODS

We analyzed 695 patients with non-metastatic NPC and detectable post-treatment cfEBV DNA load who did not receive adjuvant therapy. We fitted the trajectories of post-treatment cfEBV DNA load as a function of time into a linear mixed-effect model and fitted a Cox regression model with covariates including age, T and N stages, and lactate dehydrogenase level. Finally, we combined both via joint modeling to develop and validate our dynamic model.

RESULTS

A strong positive correlation was found between the individual longitudinal post-treatment cfEBV DNA load and the risk of death from any cause (P < 0.001). We developed a joint model capable of providing subject-specific dynamic prediction of conditional OS based on the evolution of the individual plasma cfEBV DNA load trajectory. The joint model showed reliable performance in both training and validation cohorts, with a large area under the curve (interquartile range [IQR]: training cohort, 0.775-0.850; validation cohort, 0.826-0.900) and low prediction errors (IQR: training cohort, 0.017-0.078; validation cohort, 0.034 -0.071). An increasing amount of data on cfEBV DNA load was associated with better model performance.

CONCLUSION

Our model provided reliable subject-specific dynamic prediction of conditional OS, which could help guide individualized post-treatment surveillance, risk stratification, and management of NPC in the future.

Advanced imaging for risk stratification for ventricular arrhythmias and sudden cardiac death

Sudden cardiac death (SCD) is a leading cause of mortality, comprising approximately half of all deaths from cardiovascular disease. In the US, the majority of SCD (85%) occurs in patients with ischemic cardiomyopathy (ICM) and a subset in patients with non-ischemic cardiomyopathy (NICM), who tend to be younger and whose risk of mortality is less clearly delineated than in ischemic cardiomyopathies. The conventional means of SCD risk stratification has been the determination of the ejection fraction (EF), typically via echocardiography, which is currently a means of determining candidacy for primary prevention in the form of implantable cardiac defibrillators (ICDs). Advanced cardiac imaging methods such as cardiac magnetic resonance imaging (CMR), single-photon emission computerized tomography (SPECT) and positron emission tomography (PET), and computed tomography (CT) have emerged as promising and non-invasive means of risk stratification for sudden death through their characterization of the underlying myocardial substrate that predisposes to SCD. Late gadolinium enhancement (LGE) on CMR detects myocardial scar, which can inform ICD decision-making. Overall scar burden, region-specific scar burden, and scar heterogeneity have all been studied in risk stratification. PET and SPECT are nuclear methods that determine myocardial viability and innervation, as well as inflammation. CT can be used for assessment of myocardial fat and its association with reentrant circuits. Emerging methodologies include the development of "virtual hearts" using complex electrophysiologic modeling derived from CMR to attempt to predict arrhythmic susceptibility. Recent developments have paired novel machine learning (ML) algorithms with established imaging techniques to improve predictive performance. The use of advanced imaging to augment risk stratification for sudden death is increasingly well-established and may soon have an expanded role in clinical decision-making. ML could help shift this paradigm further by advancing variable discovery and data analysis.

On model-based time trend adjustments in platform trials with non-concurrent controls

BACKGROUND

Platform trials can evaluate the efficacy of several experimental treatments compared to a control. The number of experimental treatments is not fixed, as arms may be added or removed as the trial progresses. Platform trials are more efficient than independent parallel group trials because of using shared control groups. However, for a treatment entering the trial at a later time point, the control group is divided into concurrent controls, consisting of patients randomised to control when that treatment arm is in the platform, and non-concurrent controls, patients randomised before. Using non-concurrent controls in addition to concurrent controls can improve the trial's efficiency by increasing power and reducing the required sample size, but can introduce bias due to time trends.

METHODS

We focus on a platform trial with two treatment arms and a common control arm. Assuming that the second treatment arm is added at a later time, we assess the robustness of recently proposed model-based approaches to adjust for time trends when utilizing non-concurrent controls. In particular, we consider approaches where time trends are modeled either as linear in time or as a step function, with steps at time points where treatments enter or leave the platform trial. For trials with continuous or binary outcomes, we investigate the type 1 error rate and power of testing the efficacy of the newly added arm, as well as the bias and root mean squared error of treatment effect estimates under a range of scenarios. In addition to scenarios where time trends are equal across arms, we investigate settings with different time trends or time trends that are not additive in the scale of the model.

RESULTS

A step function model, fitted on data from all treatment arms, gives increased power while controlling the type 1 error, as long as the time trends are equal for the different arms and additive on the model scale. This holds even if the shape of the time trend deviates from a step function when patients are allocated to arms by block randomisation. However, if time trends differ between arms or are not additive to treatment effects in the scale of the model, the type 1 error rate may be inflated.

CONCLUSIONS

The efficiency gained by using step function models to incorporate non-concurrent controls can outweigh potential risks of biases, especially in settings with small sample sizes. Such biases may arise if the model assumptions of equality and additivity of time trends are not satisfied. However, the specifics of the trial, scientific plausibility of different time trends, and robustness of results should be carefully considered.

Two-stage matching-adjusted indirect comparison

BACKGROUND

Anchored covariate-adjusted indirect comparisons inform reimbursement decisions where there are no head-to-head trials between the treatments of interest, there is a common comparator arm shared by the studies, and there are patient-level data limitations. Matching-adjusted indirect comparison (MAIC), based on propensity score weighting, is the most widely used covariate-adjusted indirect comparison method in health technology assessment. MAIC has poor precision and is inefficient when the effective sample size after weighting is small.

METHODS

A modular extension to MAIC, termed two-stage matching-adjusted indirect comparison (2SMAIC), is proposed. This uses two parametric models. One estimates the treatment assignment mechanism in the study with individual patient data (IPD), the other estimates the trial assignment mechanism. The first model produces inverse probability weights that are combined with the odds weights produced by the second model. The resulting weights seek to balance covariates between treatment arms and across studies. A simulation study provides proof-of-principle in an indirect comparison performed across two randomized trials. Nevertheless, 2SMAIC can be applied in situations where the IPD trial is observational, by including potential confounders in the treatment assignment model. The simulation study also explores the use of weight truncation in combination with MAIC for the first time.

RESULTS

Despite enforcing randomization and knowing the true treatment assignment mechanism in the IPD trial, 2SMAIC yields improved precision and efficiency with respect to MAIC in all scenarios, while maintaining similarly low levels of bias. The two-stage approach is effective when sample sizes in the IPD trial are low, as it controls for chance imbalances in prognostic baseline covariates between study arms. It is not as effective when overlap between the trials' target populations is poor and the extremity of the weights is high. In these scenarios, truncation leads to substantial precision and efficiency gains but induces considerable bias. The combination of a two-stage approach with truncation produces the highest precision and efficiency improvements.

CONCLUSIONS

Two-stage approaches to MAIC can increase precision and efficiency with respect to the standard approach by adjusting for empirical imbalances in prognostic covariates in the IPD trial. Further modules could be incorporated for additional variance reduction or to account for missingness and non-compliance in the IPD trial.

Ensemble methods for survival function estimation with time-varying covariates

Survival data with time-varying covariates are common in practice. If relevant, they can improve on the estimation of a survival function. However, the traditional survival forests-conditional inference forest, relative risk forest and random survival forest-have accommodated only time-invariant covariates. We generalize the conditional inference and relative risk forests to allow time-varying covariates. We also propose a general framework for estimation of a survival function in the presence of time-varying covariates. We compare their performance with that of the Cox model and transformation forest, adapted here to accommodate time-varying covariates, through a comprehensive simulation study in which the Kaplan-Meier estimate serves as a benchmark, and performance is compared using the integrated L2 difference between the true and estimated survival functions. In general, the performance of the two proposed forests substantially improves over the Kaplan-Meier estimate. Taking into account all other factors, under the proportional hazard setting, the best method is always one of the two proposed forests, while under the non-proportional hazard setting, it is the adapted transformation forest. K-fold cross-validation is used as an effective tool to choose between the methods in practice.

Frontal cortical activation during emotional and non-emotional verbal fluency tests

There has been growing recognition of the utility of combining the verbal fluency test and functional near-infrared spectroscopy (fNIRS) to assess brain functioning and to screen for psychiatric disorders. Recently, an emotional analogue of the semantic fluency test (SFT) has been developed that taps partly different processes from conventional verbal fluency tests. Nevertheless, neural processing during the emotional SFT remains elusive. Here, fNIRS was used to compare frontal cortical activation during emotional and non-emotional SFTs. The goal was to determine whether the emotional SFT activated overlapping yet distinct frontal cortical regions compared with the conventional, non-emotional SFT. Forty-three healthy young adults performed the emotional and non-emotional SFTs while hemodynamic changes in the bilateral frontopolar, dorsomedial, dorsolateral, ventrolateral, and posterolateral frontal cortices were measured by fNIRS. There were significant increases in oxyhemoglobin concentration and significant decreases in deoxyhemoglobin concentration (i.e., activation) in frontopolar, dorsolateral, and ventrolateral frontal regions during both the non-emotional and emotional SFTs. Also, complementary analyses conducted on changes in the two chromophores using classical and Bayesian hypothesis testing suggested that comparable frontal cortical regions were activated while performing the two tests. This similarity in activation occurred in a context where non-emotional and emotional SFT performances exhibited differential relationships with the overall level of negative mood symptoms. In conclusion, frontal cortical activation during the emotional SFT is similar to that during the conventional, non-emotional SFT. Given that there is evidence for discriminant validity for the emotional SFT, the neural mechanisms underlying the uniqueness of this test warrant further investigation.

Learning and Predicting from Dynamic Models for COVID-19 Patient Monitoring

COVID-19 has challenged health systems to learn how to learn. This paper describes the context, methods and challenges for learning to improve COVID-19 care at one academic health center. Challenges to learning include: (1) choosing a right clinical target; (2) designing methods for accurate predictions by borrowing strength from prior patients' experiences; (3) communicating the methodology to clinicians so they understand and trust it; (4) communicating the predictions to the patient at the moment of clinical decision; and (5) continuously evaluating and revising the methods so they adapt to changing patients and clinical demands. To illustrate these challenges, this paper contrasts two statistical modeling approaches - prospective longitudinal models in common use and retrospective analogues complementary in the COVID-19 context - for predicting future biomarker trajectories and major clinical events. The methods are applied to and validated on a cohort of 1,678 patients who were hospitalized with COVID-19 during the early months of the pandemic. We emphasize graphical tools to promote physician learning and inform clinical decision making.

The importance of chronic conditions for potentially avoidable hospitalizations among non-Hispanic Black and non-Hispanic White older adults in the US: a cross-sectional observational study

Background

Non-Hispanic (NH) Black older adults experience substantially higher rates of potentially avoidable hospitalization compared to NH White older adults. This study explores the top three chronic conditions preceding hospitalization and potentially avoidable hospitalization among NH White and NH Black Medicare beneficiaries in the United States.

Methods

Data on 4993 individuals (4,420 NH White and 573 NH Black individuals) aged ≥ 65 years from 2014 Medicare claims were linked with sociodemographic data from previous rounds of the Health and Retirement Study. Conditional inference random forests were used to rank the importance of chronic conditions in predicting hospitalization and potentially avoidable hospitalization separately for NH White and NH Black beneficiaries. Multivariable logistic regression with the top three chronic diseases for each outcome adjusted for sociodemographic characteristics were conducted to quantify the associations.

Results

In total, 22.1% of NH White and 24.9% of NH Black beneficiaries had at least one hospitalization during 2014. Among those with hospitalization, 21.3% of NH White and 29.6% of NH Black beneficiaries experienced at least one potentially avoidable hospitalization. For hospitalizations, chronic kidney disease, heart failure, and atrial fibrillation were the top three contributors among NH White beneficiaries and acute myocardial infarction, chronic obstructive pulmonary disease (COPD), and chronic kidney disease were the top three contributors among NH Black beneficiaries. These chronic conditions were associated with increased odds of hospitalization for both groups. For potentially avoidable hospitalizations, asthma, COPD, and heart failure were the top three contributors among NH White beneficiaries and fibromyalgia/chronic pain/fatigue, COPD, and asthma were the top three contributors among NH Black beneficiaries. COPD and heart failure were associated with increased odds of potentially avoidable hospitalization among NH White beneficiaries, whereas only COPD was associated with increased odds of potentially avoidable hospitalizations among NH Black beneficiaries.

Conclusion

Having at least one hospitalization and at least one potentially avoidable hospitalization was more prevalent among NH Black than NH White Medicare beneficiaries. This suggests greater opportunity for increasing prevention efforts among NH Black beneficiaries. The importance of COPD for potentially avoidable hospitalizations further highlights the need to focus on prevention of exacerbations for patients with COPD, possibly through greater access to primary care and continuity of care.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-022-07849-y.

Improved proprioception does not benefit visuomotor adaptation

Visuomotor adaptation arises when reaching in an altered visual environment, where one's seen hand position does not match their felt (i.e., proprioceptive) hand position in space. Here, we asked if proprioceptive training benefits visuomotor adaptation, and if these benefits arise due to implicit (unconscious) or explicit (conscious strategy) processes. Seventy-two participants were divided equally into 3 groups: proprioceptive training with feedback (PTWF), proprioceptive training no feedback (PTNF), and Control (CTRL). The PTWF and PTNF groups completed passive proprioceptive training, where a participant's hand was moved to an unknown reference location and they judged the felt position of their unseen hand relative to their body midline on every trial. The PTWF group received verbal feedback with respect to their response accuracy on the middle 60% of trials, whereas the PTNF did not receive any feedback during training. The CTRL group did not complete proprioceptive training and instead sat quietly during this time. Following proprioceptive training or time delay, all three groups reached when seeing a cursor that was rotated 30° clockwise relative to their hand motion. The experiment ended with participants completing a series of no-cursor reaches to assess implicit and explicit adaptation. Results indicated that the PTWF group improved the accuracy of their sense of felt hand position following proprioceptive training. However, this improved proprioceptive acuity (i.e., the accuracy of their sense of felt hand) did not benefit visuomotor adaptation, as all three groups showed similar visuomotor adaptation across rotated reach training trials. Visuomotor adaptation arose implicitly, with minimal explicit contribution for all three groups. Together, these results suggest that passive proprioceptive training does not benefit, nor hinder, the extent of implicit visuomotor adaptation established immediately following reach training with a 30° cursor rotation.

Impact of the bidirectional relationship between communication and cognitive efficacy on orthopedic patient adherence behavior

Background

There is growing interest in patient autonomy, and communication between physicians and patients has become the essential cornerstone for improving the quality of healthcare services. Previous research has concentrated on the direct effect of physician-patient communication on service outcomes. In the present study, we examined the influence among constructs in the service process and the impact on healthcare outcomes. The present study used behavioral theory to expand the process aspect of the Donabedian healthcare service quality structure-process-outcome model to examine the impact of cognitive changes and communication feedback on patients’ adherence behavior. In addition, the moderating effect of hospital facility levels is examined.

Methods

A conceptual model was developed and tested using a questionnaire administered to patients in eight hospitals. A total of 397 respondents returned usable surveys, with a response rate of 92.11%. Structural equation modeling was used to analyze the data in two steps that involved a measurement model and a structural model. The former was applied to estimate the Cronbach’s alphas, intercorrelations of factors, and descriptive statistics; the latter was used to test the hypothesized relationships of the constructs.

Results

The results identified three mediators of the healthcare process within the healthcare services framework: physician-patient communication, cognitive efficacy, and adherence behavior. Physician-patient communication influenced cognitive efficacy (β = 0.16, p < 0.001), and cognitive efficacy influenced physician-patient communication (β = 0.18, p < 0.001). The effect of this bidirectional relationship on adherence behavior was positive (β = 0.38, p < 0.001). The healthcare structure influenced healthcare outcomes via these three healthcare process constructs. The adherence behavior of patients who were treated in the medical center has greater influences by the structure and physician-patient communication than it was treated in the regional hospitals.

Conclusions

This study revealed a complex pattern in relationships among process constructs for healthcare services. The findings of this study acknowledge the important potential interrelationships among the healthcare service constructs to improve the quality of healthcare outcomes.

Trial registration

CRREC104107. Date: 22/01/2016. Prospectively Registered.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-022-07575-5.

Temporally informed random forests for suicide risk prediction

OBJECTIVE

Suicide is one of the leading causes of death worldwide, yet clinicians find it difficult to reliably identify individuals at high risk for suicide. Algorithmic approaches for suicide risk detection have been developed in recent years, mostly based on data from electronic health records (EHRs). Significant room for improvement remains in the way these models take advantage of temporal information to improve predictions.

MATERIALS AND METHODS

We propose a temporally enhanced variant of the random forest (RF) model-Omni-Temporal Balanced Random Forests (OT-BRFs)-that incorporates temporal information in every tree within the forest. We develop and validate this model using longitudinal EHRs and clinician notes from the Mass General Brigham Health System recorded between 1998 and 2018, and compare its performance to a baseline Naive Bayes Classifier and 2 standard versions of balanced RFs.

RESULTS

Temporal variables were found to be associated with suicide risk: Elevated suicide risk was observed in individuals with a higher total number of visits as well as those with a low rate of visits over time, while lower suicide risk was observed in individuals with a longer period of EHR coverage. RF models were more accurate than Naive Bayesian classifiers at predicting suicide risk in advance (area under the receiver operating curve = 0.824 vs. 0.754, respectively). The proposed OT-BRF model performed best among all RF approaches, yielding a sensitivity of 0.339 at 95% specificity, compared to 0.290 and 0.286 for the other 2 RF models. Temporal variables were assigned high importance by the models that incorporated them.

DISCUSSION

We demonstrate that temporal variables have an important role to play in suicide risk detection and that requiring their inclusion in all RF trees leads to increased predictive performance. Integrating temporal information into risk prediction models helps the models interpret patient data in temporal context, improving predictive performance.