Latent Profile Analysis of Canadian Military Veterans With Chronic Pain Identifies 5 Meaningful Classes Through Self-Report Measures

The purpose of this study was to identify meaningful response patterns in self-report survey data collected from Canadian military veterans with chronic pain and to create an algorithm intended to facilitate triage and prioritization of veterans to the most appropriate interventions. An online survey was presented to former members of the Canadian military who self-identified as having chronic pain. Variables collected were related to pain, physical and mental interference, prior traumatic experiences, and indicators from each of the 7 potential drivers of the pain experience. Maximum likelihood estimation-based latent profile analysis was used to identify clinically and statistically meaningful profiles using the 7-axis variables, and classification and regression tree (CRT) analysis was then conducted to identify the most parsimonious set of indicators that could be used to accurately classify respondents into the most relevant profile group. Data from N = 322 veterans were available for analysis. The results of maximum likelihood estimation-based latent profile analysis indicated a 5-profile structure was optimal for explaining the patterns of responses within the data. These were: Mood-Dominant (13%), Localized Physical (24%), Neurosensory-Dominant (33%), Central-Dominant with complex mood and neurosensory symptoms (16%), and Trauma- and mood-dominant (14%). From CRT analysis, an algorithm requiring only 3 self-report tools (central symptoms, mood screening, bodily coherence) achieved 83% classification accuracy across the 5 profiles. The new classification algorithm requiring 16 total items may be helpful for clinicians and veterans in pain to identify the most dominant drivers of their pain experience that may be useful for prioritizing intervention strategies, targets, and relevant health care disciplines. PERSPECTIVE: This article presents the results of latent profile (cluster) analysis of responses to standardized self-report questionnaires by Canadian military veterans with chronic pain. It identified 5 clusters that appear to represent different drivers of the pain experience. The results could be useful for triaging veterans to the most appropriate pain care providers.

Application of machine learning algorithms to predict dead on arrival of broiler chickens raised without antibiotic program

Understanding the factors of dead-on-arrival (DOA) incidents during pre-slaughter handling is crucial for informed decision-making, improving broiler welfare, and optimizing farm profitability. In this study, 3 different machine learning (ML) algorithms - least absolute shrinkage and selection operator (LASSO), classification tree (CT), and random forest (RF) - were used together with 4 sampling techniques to optimize imbalanced data. The dataset comes from 22,115 broiler truckloads from a large producer in Thailand (2021-2022) and includes 14 independent variables covering the rearing, catching, and transportation stages. The study focuses on DOA% in the range of 0.10 to 1.20%, with a threshold for high DOA% above 0.3%, and records DOA% per truckload during pre-slaughter ante-mortem inspection. With a high DOA rate of 25.2%, the imbalanced dataset prompts the implementation of 4 methods to tune the imbalance parameters: random over sampling (ROS), random under sampling (RUS), both sampling (BOTH), and synthetic sampling or random over sampling example (ROSE). The aim is to improve the performance of the prediction model in classifying and predicting high DOA%. The comparative analysis of the different error metrics shows that RF outperforms the other models in a balanced dataset. In particular, RUS shows a significant improvement in prediction performance across all models compared to the original unbalanced dataset. The identification of the 4 most important variables for predicting high DOA percentages - mortality and culling rate, rearing stocking density, season, and mean body weight - emphasizes their importance for broiler production. This study provides valuable insights into the prediction of DOA status using an ML approach and contributes to the development of more effective strategies to mitigate high DOA percentages in commercial broiler production.

Migration timing and tributary use of spawning flannelmouth sucker (Catostomus latipinnis)

Spawning phenology and associated migrations of fishes are often regulated by factors such as temperature and stream discharge, but flow regulation of mainstem rivers coupled with climate change might disrupt these cues and affect fitness. Flannelmouth sucker (Catostomus latipinnis) persisting in heavily modified river networks are known to spawn in tributaries that might provide better spawning habitat than neighboring mainstem rivers subject to habitat degradation (e.g., embedded sediments, altered thermal regimes, and disconnected floodplains). PIT tag data and radio telemetry were used to quantify the timing and duration of flannelmouth sucker tributary spawning migrations in relation to environmental cues in McElmo Creek, a tributary of the San Juan River in the American Southwest. We also tested the extent of the tributary migration and assessed mainstem movements prior to and after tributary migrations. Additionally, multiyear data sets of PIT detections from other tributaries in the Colorado River basin were used to quantify interannual and cross-site variation in the timing of flannelmouth sucker spawning migrations in relation to environmental cues. The arrival and residence times of fish spawning in McElmo Creek varied among years, with earlier migration and a 3-week increase in residence time in relatively wet years compared to drier years. Classification tree analysis suggested a combination of discharge- and temperature-determined arrival timing. Of fish PIT tagged in the fall, 56% tagged within 10 km of McElmo Creek spawned in the tributary the following spring, as did 60% of radio-tagged fish, with a decline in its use corresponding to increased distance of tagging location. A broader analysis of four tributaries in the Colorado River basin, including McElmo Creek, found photoperiod and temperature of tributary and mainstem rivers were the most important variables in determining migration timing, but tributary and mainstem discharge also aided in classification success. The largest tributary, the Little Colorado River, had more residential fish or fish that stayed for longer periods (median = 30 days), whereas McElmo Creek fish stayed an average of just 10 days in 2022. Our results generally suggest that higher discharge, across years or across sites, results in extended use of tributaries by flannelmouth suckers. Conservation actions that limit water extraction and maintain natural flow regimes in tributaries, while maintaining open connection with mainstem rivers, may benefit migratory species, including flannelmouth suckers.

Content of clinker and other materials in personal thoracic aerosol samples from cement plants estimated by scanning electron microscopy and energy-dispersive X-ray microanalysis

OBJECTIVES

To estimate the composition and exposure to clinker and other specific components in personal thoracic dust samples of cement production workers.

METHODS

A procedure for the classification of airborne particles in cement production plants was developed based on classification trees. For this purpose, the chemical compositions of 27,217 particles in 29 material samples (clinker, limestone, gypsum, clay, quartz, bauxite, iron source, coal fly ash, and coal) were determined automatically by scanning electron microscopy (SEM) and energy-dispersive X-ray microanalysis (EDX). The concentrations of the major elements in cement (calcium, aluminium, silicon, iron, and sulphur) were used for the classifications. The split criteria of the classification trees obtained in the material samples were used to classify 44,176 particles in 34 personal thoracic aerosol samples. The contents of clinker and other materials were estimated, and the clinker contents were analysed statistically for differences between job types and job tasks.

RESULTS

Between 64% and 88% of the particles from material samples were classified as actual materials. The material types with variable composition (clay, coal fly ash, and coal) were classified with the lowest consistency (64% to 67%), while materials with a more limited compositional variation (clinker, gypsum, and quartz) were classified more consistently (76% to 85%). The arithmetic mean (AM) of the clinker content in personal samples was 62.1%, the median was 55.3%, and 95% confidence interval (CI) was 42.6% to 68.1%. No significant differences were observed between job types. However, the clinker content in samples when workers handled materials with high clinker content was significantly higher than when materials with lower clinker content were handled, 85% versus 65% (P = 0.02). The limestone content was AM 14.8%, median 13.2% (95% CI 5.5 to 20.9), whereas the other materials were present with relative abundances of median ≤ 6.4%.

DISCUSSION

Automated particle analysis by SEM-EDX followed by classification tree analysis quantified clinker with fairly high consistency when evaluated together with raw materials that are expected to be airborne in cement production plants. The clinker proportions for job types were similar. Tasks a priori ranked by assumed clinker content were significantly different and according to expectations, which supports the validity of the chosen methodology.

CONCLUSIONS

The composition of personal samples of mineral aerosols in the cement production industry could be estimated by automated single particle analysis with SEM-EDX and classification by a classification tree procedure. Clinker was the major component in the thoracic aerosol that cement production workers were exposed to. Differences between job types were relatively small and not significant. The clinker content from tasks was in agreement with assumptions.

Classification tree analysis to evaluate the most useful magnetic resonance image type in the differentiation between early and progressed hepatocellular carcinoma

AIM

Using classification tree analysis, we evaluated the most useful magnetic resonance (MR) image type in the differentiation between early and progressed hepatocellular carcinoma (eHCC and pHCC).

METHODS

We included pathologically proven 214 HCCs (28 eHCCs and 186 pHCCs) in 144 patients. The signal intensity of HCCs was assessed on in-phase (T1in) and opposed-phase T1-weighted images (T1op), ultrafast T2-weighted images (ufT2WI), fat-saturated T2-weighted images (fsT2WI), diffusion-weighted images (DWI), contrast enhanced T1-weighted images in the arterial phase (AP), portal venous phase (PVP), and the hepatobiliary phase. Fat content and washout were also evaluated. Fisher's exact test was performed to evaluate usefulness for the differentiation. Then, we chose MR images using binary logistic regression analysis and performed classification and regression tree analysis with them. Diagnostic performances of the classification tree were evaluated using a stratified 10-fold cross-validation method.

RESULTS

T1in, ufT2WI, fsT2WI, DWI, AP, PVP, fat content, and washout were all useful for the differentiation (p < 0.05), and AP and T1in were finally chosen for creating classification trees (p < 0.05). AP appeared in the first node in the tree. The area under the curve, sensitivity and specificity for eHCC, and balanced accuracy of the classification tree were 0.83 (95% CI 0.74-0.91), 0.64 (18/28, 95% CI 0.46-0.82), 0.94 (174/186, 95% CI 0.90-0.97), and 0.79 (95% CI 0.70-0.87), respectively.

CONCLUSIONS

AP is the most useful MR image type and T1in the second in the differentiation between eHCC and pHCC.

Diagnosis of comorbid migraine without aura in patients with idiopathic/genetic epilepsy based on the gray zone approach to the International Classification of Headache Disorders 3 criteria

Background

Migraine without aura (MwoA) is a very frequent and remarkable comorbidity in patients with idiopathic/genetic epilepsy (I/GE). Frequently in clinical practice, diagnosis of MwoA may be challenging despite the guidance of current diagnostic criteria of the International Classification of Headache Disorders 3 (ICHD-3). In this study, we aimed to disclose the diagnostic gaps in the diagnosis of comorbid MwoA, using a zone concept, in patients with I/GEs with headaches who were diagnosed by an experienced headache expert.

Methods

In this multicenter study including 809 consecutive patients with a diagnosis of I/GE with or without headache, 163 patients who were diagnosed by an experienced headache expert as having a comorbid MwoA were reevaluated. Eligible patients were divided into three subgroups, namely, full diagnosis, zone I, and zone II according to their status of fulfilling the ICHD-3 criteria. A Classification and Regression Tree (CART) analysis was performed to bring out the meaningful predictors when evaluating patients with I/GEs for MwoA comorbidity, using the variables that were significant in the univariate analysis.

Results

Longer headache duration (<4 h) followed by throbbing pain, higher visual analog scale (VAS) scores, increase of pain by physical activity, nausea/vomiting, and photophobia and/or phonophobia are the main distinguishing clinical characteristics of comorbid MwoA in patients with I/GE, for being classified in the full diagnosis group. Despite being not a part of the main ICHD-3 criteria, the presence of associated symptoms mainly osmophobia and also vertigo/dizziness had the distinguishing capability of being classified into zone subgroups. The most common epilepsy syndromes fulfilling full diagnosis criteria (n = 62) in the CART analysis were 48.39% Juvenile myoclonic epilepsy followed by 25.81% epilepsy with generalized tonic-clonic seizures alone.

Conclusion

Longer headache duration, throbbing pain, increase of pain by physical activity, photophobia and/or phonophobia, presence of vertigo/dizziness, osmophobia, and higher VAS scores are the main supportive associated factors when applying the ICHD-3 criteria for the comorbid MwoA diagnosis in patients with I/GEs. Evaluating these characteristics could be helpful to close the diagnostic gaps in everyday clinical practice and fasten the diagnostic process of comorbid MwoA in patients with I/GEs.

Prevalence and Correlates of Insomnia Symptoms in Older Chinese Adults During the COVID-19 Outbreak: A Classification Tree Analysis

OBJECTIVE

To examine prevalence and correlates of insomnia symptoms in older Chinese adults (OCAs) during the COVID-19 outbreak.

BACKGROUND

During the COVID-19 pandemic, insomnia is a major health concern of elderly individuals, but its subtypes have not been investigated.

METHODS

Altogether, 590 OCAs (50+ years) were recruited via snowball sampling during the COVID-19 outbreak. Standardized self-report questions were used to assess the presence of difficulty initiating sleep (DIS), difficulty maintaining sleep (DMS), and early morning awakening (EMA). Classification tree analysis (CTA) was used to identify correlates of insomnia.

RESULTS

The one-month prevalence (95% confidence interval) of any subtype of insomnia symptoms was 23.4% (20.0-26.8%), with DIS, DMS, and EMA being 15.4% (12.5-18.3%), 17.1% (14.1-20.2%), and 11.2% (8.64-13.7%), respectively. Worry about being infected with COVID-19 emerged as the most salient correlate of insomnia (P < .001); compared to participants who were not worried about being infected, those who were worried and very worried were 3.2-fold (24.3% vs 7.5%) and 5.5-fold (24.3% vs 7.5%) more likely to have insomnia, respectively. Among participants in the "very worried" branch, those residing in Wuhan were 1.8-fold more likely to have insomnia than those residing in other places (50.0% vs 27.5%, P = .011). Among participants in the "worried" branch, unemployed persons were 2.0-fold more likely to have insomnia than employed persons (37.0% vs 18.1%, P < .001).

CONCLUSIONS

Insomnia symptoms were prevalent among OCAs during the COVID-19 outbreak. Selective intervention programs targeting elderly individuals who are worried about being infected, living in the epicenter of COVID-19, and unemployed might be effective.

Do we need a more flexbible use of Team Timeout calling? Evidence from the Handball Bundesliga

Calling a Team Timeout (TTO) is one of the coaches' most important tools. Given the key competitive advantage to determine your own timing, it is crucial to make a good decision when to use a TTO. Existing research shows that teams can benefit in general from TTOs and that they are called at the end of the game and when trailing (Gomes et al., 2014; Gutiérrez-Aguilar et al., 2016; Prieto et al., 2016). However, to generate relevant findings, situational variables must be included (Fernandez-Navarro et al., 2020; Gómez, Lago-Peñas et al., 2015). By integrating situational variables like scoring streak and player difference and higher-order interactions, this study aims to identify specific game situations where TTOs are most effective. Based on 850 games of the German Handball Bundesliga, game situations are identified by Classification Tree Analysis and efficacies are evaluated. Findings indicate a strong impact of timing. Frequently used TTOs, e.g., at the end of periods, are beneficial to the teams. However, strongest effect occurs for TTOs taken at the early stages of the game and with a positive run. Results indicate that TTO is a powerful tactical tool and an application at uncommon timings may even enhance the success rate.

Prediction of Recurrence by Machine Learning in Salivary Gland Cancer Patients After Adjuvant (Chemo)Radiotherapy

BACKGROUND/AIM

To investigate survival outcomes and recurrence patterns using machine learning in patients with salivary gland malignant tumor (SGMT) undergoing adjuvant chemoradiotherapy (CRT).

PATIENTS AND METHODS

Consecutive SGMT patients were identified, and a data set included nine predictor variables and a dependent variable [disease-free survival (DFS) event] was standardized. The open-source R software was used. Survival outcomes were estimated by the Kaplan-Meier method. The random forest approach was used to select the important explanatory variables. A classification tree that optimally partitioned SGMT patients with different DFS rates was built.

RESULTS

In total, 54 SGMT patients were included in the final analysis. Five-year DFS was 62.1%. The top two important variables identified were pathologic node (pN) and pathologic tumor (pT). Based on these explanatory variables, patients were partitioned in three groups, including pN0, pT1-2 pN+ and pT3-4 pN+ with 26%, 38% and 75% probability of recurrence, respectively. Accordingly, 5-year DFS rates were 73.7%, 57.1% and 34.3%, respectively.

CONCLUSION

The proposed decision tree algorithm is an appropriate tool to partition SGMT patients. It can guide decision-making and future research in the SGMT field.

A survival tree based on stabilized score tests for high-dimensional covariates

A survival tree can classify subjects into different survival prognostic groups. However, when data contains high-dimensional covariates, the two popular classification trees exhibit fatal drawbacks. The logrank tree is unstable and tends to have false nodes; the conditional inference tree is difficult to interpret the adjusted P-value for high-dimensional tests. Motivated by these problems, we propose a new survival tree based on the stabilized score tests. We propose a novel matrix-based algorithm in order to tests a number of nodes simultaneously via stabilized score tests. We propose a recursive partitioning algorithm to construct a survival tree and develop our original R package uni.survival.tree (https://cran.r-project.org/package=uni.survival.tree) for implementation. Simulations are performed to demonstrate the superiority of the proposed method over the existing methods. The lung cancer data analysis demonstrates the usefulness of the proposed method.

Lymph node metastases >5 and metastatic lymph node ratio >0.30 of differentiated thyroid cancer predict response to radioactive iodine

PURPOSE

The study was designed to elucidate the predictive value of the number of lymph node metastases (LNMs) and lymph node ratio (LNR) for response to therapy restratification system (RTRS).

METHODS

From December 2015 to December 2019, 1228 patients who accepted radioactive iodine (RAI) were collected in the study. After 6-8 months, response to RAI was evaluated as complete response (excellent response) and incomplete response (indeterminate, biochemical, and structural incomplete response). The study developed classification tree to determine the optimum LNMs and LNR that predicted response to RAI. Multivariate logistic regression analyses were further analyzed to find independent factors of response to RAI.

RESULT

The mean age of patients was 44 ± 12 and 71.09% (873/1228) were females. The best cutoff value of LNMs to affect RAI treatment response determined by classification tree was 5. Further in 388 patients with LNMs >5, the best cutoff value of LNR to affect RAI treatment response determined by classification tree was 0.30. With multivariate analysis, the study found that LNMs (>5), gender, lymph node dissection, and American Thyroid Association (ATA) risk classification were independent predictors of response to RAI for all 1228 patients; and LNR (>0.30), gender, and ATA risk classification for 388 patients with LNMs >5. The sensitivity analysis indicated that whether patients with LNM or not were included, the multivariate logistic regression model was kept stable. On subgroup analysis, no significant interactions were observed between the effect of LNMs/LNR and gender, N stage, ATA risk classification, lymph node dissection, or T stage.

CONCLUSIONS

With classification tree, the study found that LNMs and LNR could predict initial response to RAI, and their optimal cutoff values were 5 and 0.30, separately.

A classification tree to assist with routine scoring of the Clinical Frailty Scale

Background

the Clinical Frailty Scale (CFS) was originally developed to summarise a Comprehensive Geriatric Assessment and yield a care plan. Especially since COVID-19, the CFS is being used widely by health care professionals without training in frailty care as a resource allocation tool and for care rationing. CFS scoring by inexperienced raters might not always reflect expert judgement. For these raters, we developed a new classification tree to assist with routine CFS scoring. Here, we test that tree against clinical scoring.

Objective/Methods

we examined agreement between the CFS classification tree and CFS scoring by novice raters (clerks/residents), and the CFS classification tree and CFS scoring by experienced raters (geriatricians) in 115 older adults (mean age 78.0 ± 7.3; 47% females) from a single centre.

Results

the intraclass correlation coefficient (ICC) for the CFS classification tree was 0.833 (95% CI: 0.768–0.882) when compared with the geriatricians’ CFS scoring. In 93%, the classification tree rating was the same or differed by at most one level with the expert geriatrician ratings. The ICC was 0.805 (0.685–0.883) when CFS scores from the classification tree were compared with the clerk/resident scores; 88.5% of the ratings were the same or ±1 level.

Conclusions

a classification tree for scoring the CFS can help with reliable scoring by relatively inexperienced raters. Though an incomplete remedy, a classification tree is a useful support to decision-making and could be used to aid routine scoring of the CFS.

Inflammation-Related Factors Identified as Biomarkers of Dehydration and Subsequent Acute Kidney Injury in Agricultural Workers

Globally, there is increasing recognition that agricultural workers are at risk for chronic kidney disease of unknown etiology (CKDu). Recurrent heat exposure, physical exertion, dehydration, muscle damage, and inflammation are hypothesized to contribute to the development of CKDu, but the relative importance of these processes and the interactions among them remain unclear. Moreover, there is a need to identify biomarkers that could distinguish individuals who are at greatest risk for kidney damage to target preventative interventions for CKDu. In this study, we evaluated dehydration and markers of inflammation, muscle damage, and renal function in agricultural workers at a non-workday baseline assessment. Urine specific gravity and kidney function were measured before and after work shifts on three subsequent days, and heat index, core body temperature, and heart rate were monitored during the work shifts. A combination of direct comparisons and machine learning algorithms revealed that reduced levels of uromodulin and sodium in urine and increased levels of interleukin-6 and C-reactive protein in serum were indicative of dehydration at baseline, and that dehydration, high body mass index, reduced urine uromodulin, and increased serum interleukin-6, C-reactive protein, and lipopolysaccharide-binding protein at baseline were predictive of acute kidney injury on subsequent workdays. Our findings suggest a method for identifying agricultural workers at greatest risk for kidney injury and reveal potential mechanisms responsible for this process, including pathways overlapping in dehydration and kidney injury. These results will guide future studies confirming these mechanisms and introducing interventions to protect kidney health in this vulnerable population.

Informing patterns of health and social care utilisation in Irish older people according to the Clinical Frailty Scale

Background: There is increasing policy interest in the consideration of frailty measures (rather than chronological age alone) to inform more equitable allocation of health and social care resources. In this study the Clinical Frailty Scale (CFS) classification tree was applied to data from The Irish Longitudinal Study on Ageing (TILDA) and correlated with health and social care utilisation. CFS transitions over time were also explored. Methods: Applying the CFS classification tree algorithm, secondary analyses of TILDA data were performed to examine distributions of health and social care by CFS categories using descriptive statistics weighted to the population of Ireland aged ≥65 years at Wave 5 (n=3,441; mean age 74.5 (SD ±7.0) years, 54.7% female). CFS transitions over 8 years and (Waves 1-5) were investigated using multi-state Markov models and alluvial charts. Results: The prevalence of CFS categories at Wave 5 were: 6% 'very fit', 36% 'fit', 31% 'managing well', 16% 'vulnerable', 6% 'mildly frail', 4% 'moderately frail' and 1% 'severely frail'. No participants were 'very severely frail' or 'terminally ill'. Increasing CFS categories were associated with increasing hospital and community health services use and increasing hours of formal and informal social care provision. The transitions analyses suggested CFS transitions are dynamic, with 2-year probability of transitioning from 'fit' (CFS1-3) to 'vulnerable' (CFS4), and 'fit' to 'frail' (CFS5+) at 34% and 6%, respectively. 'Vulnerable' and 'frail' had a 22% and 17% probability of reversal to 'fit' and 'vulnerable', respectively. Conclusions: Our results suggest that the CFS classification tree stratified the TILDA population aged ≥65 years into subgroups with increasing health and social care needs. The CFS could be used to aid the allocation of health and social care resources in older people in Ireland. We recommend that CFS status in individuals is reviewed at least every 2 years.

Enhancing COVID-19 Vaccines Acceptance: Results from a Survey on Vaccine Hesitancy in Northern Italy

In March 2021, the coronavirus disease 2019 (COVID-19) pandemic still poses a threat to the global population, and is a public health challenge that needs to be overcome. Now more than ever, action is needed to tackle vaccine hesitancy, especially in light of the availability of effective and safe vaccines. A cross-sectional online survey was carried out on a representative random sample of 1011 citizens from the Emilia-Romagna region, in Italy, in January 2021. The questionnaire collected information on socio-demographics, comorbidities, past vaccination refusal, COVID-19-related experiences, risk perception of infection, and likelihood to accept COVID-19 vaccination. Multiple logistic regression analyses and classification tree analyses were performed to identify significant predictors of vaccine hesitancy and to distinguish groups with different levels of hesitancy. Overall, 31.1% of the sample reported hesitancy. Past vaccination refusal was the key discriminating variable followed by perceived risk of infection. Other significant predictors of hesitancy were: ages between 35 and 54 years, female gender, low educational level, low income, and absence of comorbidities. The most common concerns about the COVID-19 vaccine involved safety (54%) and efficacy (27%). Studying the main determinants of vaccine hesitancy can help with targeting vaccination strategies, in order to gain widespread acceptance-a key path to ensure a rapid way out of the current pandemic emergency.

Computational identification of chemical compounds with potential anti-Chagas activity using a classification tree

Chagas disease is endemic to 21 Latin American countries and is a great public health problem in that region. Current chemotherapy remains unsatisfactory; consequently the need to search for new drugs persists. Here we present a new approach to identify novel compounds with potential anti-chagasic action. A large dataset of 584 compounds, obtained from the Drugs for Neglected Diseases initiative, was selected to develop the computational model. Dragon software was used to calculate the molecular descriptors and WEKA software to obtain the classification tree. The best model shows accuracy greater than 93.4% for the training set; the tree was also validated using a 10-fold cross-validation procedure and through a test set, achieving accuracy values over 90.5% and 92.2%, correspondingly. The values of sensitivity and specificity were around 90% in all series; also the false alarm rate values were under 10.5% for all sets. In addition, a simulated ligand-based virtual screening for several compounds recently reported as promising anti-chagasic agents was carried out, yielding good agreement between predictions and experimental results. Finally, the present work constitutes an example of how this rational computer-based method can help reduce the cost and increase the rate in which novel compounds are developed against Chagas disease.

Obtaining optimal cutoff values for tree classifiers using multiple biomarkers

In biomedical practices, multiple biomarkers are often combined using a prespecified classification rule with tree structure for diagnostic decisions. The classification structure and cutoff point at each node of a tree are usually chosen on an ad hoc basis, depending on decision makers' experience. There is a lack of analytical approaches that lead to optimal prediction performance, and that guide the choice of optimal cutoff points in a pre-specified classification tree. In this paper, we propose to search for and estimate the optimal decision rule through an approach of rank correlation maximization. The proposed method is flexible, theoretically sound, and computationally feasible when many biomarkers are available for classification or prediction. Using the proposed approach, for a prespecified tree-structured classification rule, we can guide the choice of optimal cutoff points at tree nodes and estimate optimal prediction performance from multiple biomarkers combined.

Classification trees identify shared and distinct correlates of nonsuicidal self-injury and suicidal ideation across gender identities in emerging adults

College students have an elevated risk for self-injurious thoughts and behaviours (SITBs), and there are robust differences in prevalence rates for SITBs across gender identities. Although numerous constructs have been implicated as risk factors, researchers have not significantly improved at predicting SITBs, possibly owing to constraints of confirmatory analyses. Classification trees are exploratory, person-centred analyses that enable joint examination of numerous correlates and their interactions. Thus, classification trees may discern previously unstudied risk factors and identify distinct subpopulations with elevated risk for SITBs. We tested classification trees that evaluated 298 potential correlates of nonsuicidal self-injury and suicidal ideation across self-identified women and men. Data came from 5,131 college students who completed the National College Health Assessment, which assesses a wide range of health-related constructs. Models produced parsimonious decision trees that accounted for a substantial amount of outcome variability (38.3-51.5%). Psychopathology, poorer psychological well-being, and other SITBs emerged as important correlates for all participants. Trauma, disordered eating, and heavy alcohol use were salient among women, whereas alcohol use norms were important correlates among men. Importantly, models identified several constructs that may be amenable to intervention. Results support the use of exploratory analyses to explicate heterogeneity among individuals who engage in SITBs and suggest that gender identity is an important moderator for certain risk factors.

Using Machine Learning Methods to Develop a Short Tree-Based Adaptive Classification Test: Case Study With a High-Dimensional Item Pool and Imbalanced Data

This study explores advanced techniques in machine learning to develop a short tree-based adaptive classification test based on an existing lengthy instrument. A case study was carried out for an assessment of risk for juvenile delinquency. Two unique facts of this case are (a) the items in the original instrument measure a large number of distinctive constructs; (b) the target outcomes are of low prevalence, which renders imbalanced training data. Due to the high dimensionality of the items, traditional item response theory (IRT)-based adaptive testing approaches may not work well, whereas decision trees, which are developed in the machine learning discipline, present as a promising alternative solution for adaptive tests. A cross-validation study was carried out to compare eight tree-based adaptive test constructions with five benchmark methods using data from a sample of 3,975 subjects. The findings reveal that the best-performing tree-based adaptive tests yielded better classification accuracy than the benchmark method IRT scoring with optimal cutpoints, and yielded comparable or better classification accuracy than the best benchmark method, random forest with balanced sampling. The competitive classification accuracy of the tree-based adaptive tests also come with an over 30-fold reduction in the length of the instrument, only administering between 3 to 6 items to any individual. This study suggests that tree-based adaptive tests have an enormous potential when used to shorten instruments that measure a large variety of constructs.

Development of Classification Models for the Prediction of Alzheimer's Disease Utilizing Circulating Sex Hormone Ratios

BACKGROUND

While sex hormones are essential for normal cognitive health, those individuals with greater endocrine dyscrasia around menopause and with andropause are more likely to develop cognitive loss and Alzheimer's disease (AD).

OBJECTIVE

To assess whether circulating sex hormones may provide an etiologically significant, surrogate biomarker, for cognitive decline.

METHODS

Plasma (n = 152) and serum (n = 107) samples from age- and gender-matched AD and control subjects from the Wisconsin Alzheimer's Disease Research Center (ADRC) were analyzed for 11 steroids and follicle-stimulating hormone. Logistic regression (LR), correlation analyses, and recursive partitioning (RP) were used to examine the interactions of hormones and hormone ratios and their association with AD. Models generated were then tested on an additional 43 ADRC samples.

RESULTS

The wide variation and substantial overlap in the concentrations of all circulating sex steroids across control and AD groups precluded their use for predicting AD. Classification tree analyses (RP) revealed interactions among single hormones and hormone ratios that associated with AD status, the most predictive including only the hormone ratios identified by LR. The strongest associations were observed between cortisol, cortisone, and androstenedione with AD, with contributions from progesterone and 17β-estradiol. Utilizing this model, we correctly predicted 81% of AD test cases and 64% of control test cases.

CONCLUSION

We have developed a diagnostic model for AD, the Wisconsin Hormone Algorithm Test for Cognition (WHAT-Cog), that utilizes classification tree analyses of hormone ratios. Further refinement of this technology could provide a quick and cheap diagnostic method for screening those with AD.

Sustainable level of human performance with regard to actual availability in different professions

BACKGROUND

In a real working environment, workers' performance depends on the level of competence, psychological and health condition, motivation, and perceived stress. These are the attributes of actual availability. It is crucial to identify the most influential attributes to develop an adequate level of worker's performance.

OBJECTIVE

The purpose of this paper is to upgrade the Availability-Humanization-Model (AH-Model) with an implementation of the artificial intelligence classification tree to identify influencing factors of the well-being attributes on human performance, where the identified influencing factors are gripping points for maintaining sustainable performance in real-life conditions of different professions.

METHODS

Well-being attributes are collected with the Questionnaire Actual Availability (QAA) from AH-Model and then analysed by implementation of the decision trees classification algorithms. An embedded clustering analysis of QAA ensures an efficient feature construction and selection. It negates the need of applying tree pruning or any other noise reduction algorithms.

RESULTS

An implementation of the machine learning algorithms reflects the real conditions of working environments: (a) real performance of workers depends on the perception of well-being and availability and (b) the most influencing factors explicitly reflect the content of work in a specific domain (Fintech, health, forestry, traffic) with a high level of stress.

CONCLUSIONS

The presented approach offers a possibility to identify the most important well-being attributes to determine an adequate efficiency and to improve the performance level in the real working conditions.

Intent Prediction of Multi-axial Ankle Motion Using Limited EMG Signals

Background: In this study, different intent prediction strategies were explored with the objective of determining the best approach to predicting continuous multi-axial user motion based solely on surface EMG (electromyography) data. These strategies were explored as the first step to better facilitating control of a multi-axis transtibial powered prosthesis.

Methods: Based on data acquired from gait experiments, different data sets, prediction approaches and classification algorithms were explored. The effect of varying EMG electrode positioning was also tested. EMG data measured from three lower leg muscles was the sole data type used for making intent predictions. The motions to be predicted were along both the sagittal plane (foot dorsiflexion and plantarflexion) and the frontal plane (foot eversion and inversion).

Results: The deviation of EMG data from its optimal pattern led to a decrease in prediction accuracy of up to 23%. However, using features that were calculated based on a participant's specific walking pattern limited this loss of prediction accuracy as a result of EMG electrode placement. A decoupled data set, one wherein the terrain type was accounted for beforehand, yielded the highest intent prediction accuracy of 77.2%.

Conclusions: The results of this study highlighted the challenges faced when using very limited EMG data to predict multi-axial ankle motion. They also indicated that approaches that are more user-centric by design could led to more accurate motion predictions, possibly enabling more intuitive control.

Homicide Profiles Based on Crime Scene and Victim Characteristics

One of the current trends in homicide research includes developing works based on scientific study and empirical evidence, which offer conclusions that can be used in an operational manner during police investigations. The objective of this study was to identify homicide characteristics from behaviors carried out on the crime scene and victim characteristics associated with those of the perpetrators of these crimes in Spain. The sample consisted of 448 homicide cases from the database of the Homicide Revision Project led by the Office of Coordination and Studies of the Secretary of State and Security. After creating six classification tree models, it was found that the modus operandi of the aggressor and the victim characteristics may permit hypothesizing about the demographic characteristics of the perpetrator (gender, age, and country of origin), his/her criminal record, and the type of relationship with the victim. Furthermore, the importance of the study of victimology during a criminal investigation is highlighted, as it may indirectly offer information about the potential perpetrator. The findings of this study suggest that criminal profiling contributes notably to the decision-making process to establish more rigorous suspect prioritization, improve the management of human resources and materials, and increase the efficiency of criminal investigations.

Differentiation of avocados according to their botanical variety using liquid chromatographic fingerprinting and multivariate classification tree

BACKGROUND

The oil content, composition and marketing threshold value of an avocado depends on the cultivar hence, identifying the cultivar of the avocado fruit is desirable. However, analytical methods have not been reported with this aim.

RESULTS

A multivariate classification tree method was proposed to discriminate three commercial botanical varieties of avocado: Hass, Fuerte and Bacon, using high-performance liquid chromatography coupled to a charged aerosol detector (HPLC-CAD). Prior to the chromatographic analysis the avocados were lyophilized and then the oil fraction was extracted using a pressurized liquid extraction system. Normal and reverse phase liquid chromatography were applied in order to obtain the chromatographic fingerprint for each sample. Soft independent modelling of class analogies (SIMCA) and partial least-squares discriminant analysis (PLS-DA) were applied. Classification quality metrics were determined to evaluate the performance of the classification. Several strategies to develop the classification models were employed. Finally, the useful application of 'classification trees' methodology, which has been scarcely applied in the field of analytical food control, was evaluated to perform a multiclass classification.

CONCLUSION

Discrimination of the three botanical varieties was achieved. The best classification was obtained when the PLS-DA is applied on the normal-phase chromatographic fingerprints. Classification trees are showed to be useful tools that provide complementary information to single concatenated models showing different results from the same prediction sample set. © 2019 Society of Chemical Industry.

Profiling older community-dwellers with hypovitaminosis D: A classification tree analysis

Despite the high prevalence of hypovitaminosis D in older adults, universal vitamin D supplementation is not recommended due to potential risk of intoxication. Our aim here was to determine the clinical profiles of older community-dwellers with hypovitaminosis D. The perspective is to build novel strategies to screen for and supplement those with hypovitaminosis D. A classification tree (CHAID analysis) was performed on multiple datasets standardizedly collected from 1991 older French community-dwelling volunteers ≥ 65 years in 2009-2012. Hypovitaminosis D was defined as serum 25-hydroxyvitamin D ≤ 50 nmol/L. CHAID analysis retained 5 clinical profiles of older community-dwellers with different risks of hypovitaminosis D up to 87.3%, based on various combinations of the following characteristics: polymorbidity, obesity, sadness and gait disorders. For instance, the probability of hypovitaminosis D was 1.42-fold higher [95CI: 1.27-1.59] for those with polymorbidity and gait disorders compared to those with no polymorbidity, no obesity and no sadness. In conclusion, these easily-recordable measures may be used in clinical routine to identify older community-dwellers for whom vitamin D supplementation should be initiated.

Using Classification Trees to Examine Predictors of Marijuana Use Among a Wide Range of Variables

Background: Young adults have elevated risk for negative marijuana use-related outcomes, and there is heterogeneity among users. Identifying risk factors for marijuana user status will improve understanding of different populations of users, which may inform prediction of individuals most likely to experience negative outcomes. Objectives: To identify predictors of marijuana use initiation in young adults. We simultaneously examined a broad range of potential predictors and all their possible interactions, including constructs that have not been previously studied in substance use initiation research. Methods: Data were repeated cross-sectional survey responses from college students in Colorado (N = 4052, 77% White, 61% female, mean age = 22.77). Measures came from the National College Health Assessment, which assesses numerous health and behavioral constructs. We used recursive partitioning and random forest models to identify predictors of ever having used marijuana out of 206 variables. Results: Classification trees identified engagement in increased alcohol use and sexual behavior as salient correlates of marijuana use initiation. Parsimonious recursive partitioning trees explained a substantial amount of variability in marijuana user status (39% in the full model and 24% when alcohol variables were excluded). Random forest models predicted user status with 74.11% and 66.91% accuracy in the full model and when alcohol variables were excluded, respectively. Conclusions: Results support the use of exploratory analyses to explain heterogeneity among marijuana users and non-users. Since engagement in other health-risk behaviors were salient predictors of use initiation, prevention efforts to reduce harm from marijuana use may benefit from targeting risk factors for health-risk behaviors in general.

Red-Light-Running Crashes' Classification, Comparison, and Risk Analysis Based on General Estimates System (GES) Crash Database

Red-light running (RLR) has been identified as one of the prominent contributing factors involved in signalized intersection crashes. In order to reduce RLR crashes, primarily, a better understanding of RLR behavior and crashes is needed. In this study, three RLR crash types were extracted from the general estimates system (GES), including go-straight (GS) RLR vehicle colliding with go-straight non-RLR vehicle, go-straight RLR vehicle colliding with left-turn (LT) non-RLR vehicle, and left-turn RLR vehicle colliding with go-straight non-RLR vehicle. Then, crash features within each crash type scenario were compared, and risk analyses of GS RLR and LT RLR were also conducted. The results indicated that for the GS RLR driver, the speed limit displayed the highest effects on the percentages of GS RLR collision scenarios. For the LT RLR driver, the number of lanes displayed the highest effects on the percentages of LT RLR collision scenarios. Additionally, the drivers who were older than 50 years, distracted, and had a limited view were more likely to be involved in LT RLR accidents. Furthermore, the speeding drivers were more likely to be involved in GS RLR accidents. These findings could give a comprehensive understanding of RLR crash features and propensities for each RLR crash type.

Computational development of a molecular-based approach to improve risk stratification of endometrial cancer patients

Histological classification and staging are the gold standard for the prognosis of endometrial cancer (EC). However, in morphologically intermediate and doubtful cases this approach results largely insufficient, defining the need for better classification criteria.

In this work we developed an algorithm that based on EC genetic alterations and in combination with the current histological classification, improves EC patients prognostic stratification, in particular in doubtful cases. A panel of 26 cancer related genes was analyzed in 89 EC patients and somatic functional mutations were investigated in association with different histology and outcome.

An unsupervised hierarchical clustering analysis revealed that two groups of patients with different tumor grade and different prognosis can be distinguished by mutational profile. In particular, the mutational status of APC, CTNNB1, PIK3CA, PTEN, SMAD4 and TP53 resulted to be principal drivers of prognostic clustering. Consistently, a decisional tree generated by a data mining approach summarizes the consequential molecular criteria for patients prognostic stratification.

The model proposed by this work provides the clinician with a tool able to support the prognosis of EC patients and consequently drives the choice of the most appropriated therapeutic strategy and follow up.

Predicting Long-Term Global Outcome after Traumatic Brain Injury: Development of a Practical Prognostic Tool Using the Traumatic Brain Injury Model Systems National Database

For patients surviving serious traumatic brain injury (TBI), families and other stakeholders often desire information on long-term functional prognosis, but accurate and easy-to-use clinical tools are lacking. We aimed to build utilitarian decision trees from commonly collected clinical variables to predict Glasgow Outcome Scale (GOS) functional levels at 1, 2, and 5 years after moderate-to-severe closed TBI. Flexible classification tree statistical modeling was used on prospectively collected data from the TBI-Model Systems (TBIMS) inception cohort study. Enrollments occurred at 17 designated, or previously designated, TBIMS inpatient rehabilitation facilities. Analysis included all participants with nonpenetrating TBI injured between January 1997 and January 2017. Sample sizes were 10,125 (year-1), 8,821 (year-2), and 6,165 (year-5) after cross-sectional exclusions (death, vegetative state, insufficient post-injury time, and unavailable outcome). In our final models, post-traumatic amnesia (PTA) duration consistently dominated branching hierarchy and was the lone injury characteristic significantly contributing to GOS predictability. Lower-order variables that added predictability were age, pre-morbid education, productivity, and occupational category. Generally, patient outcomes improved with shorter PTA, younger age, greater pre-morbid productivity, and higher pre-morbid vocational or educational achievement. Across all prognostic groups, the best and worst good recovery rates were 65.7% and 10.9%, respectively, and the best and worst severe disability rates were 3.9% and 64.1%. Predictability in test data sets ranged from C-statistic of 0.691 (year-1; confidence interval [CI], 0.675, 0.711) to 0.731 (year-2; CI, 0.724, 0.738). In conclusion, we developed a clinically useful tool to provide prognostic information on long-term functional outcomes for adult survivors of moderate and severe closed TBI. Predictive accuracy for GOS level was demonstrated in an independent test sample. Length of PTA, a clinical marker of injury severity, was by far the most critical outcome determinant.

Subgroup identification in clinical trials: an overview of available methods and their implementations with R

Randomized controlled trials (RCTs) usually enroll heterogeneous study population, and thus it is interesting to identify subgroups of patients for whom the treatment may be beneficial or harmful. A variety of methods have been developed to do such kind of post hoc analyses. Conventional generalized linear model is able to include prognostic variables as a main effect and predictive variables in an interaction with treatment variable. A statistically significant and large interaction effect usually indicates potential subgroups that may have different responses to the treatment. However, the conventional regression method requires to specify the interaction term, which requires knowledge of predictive variables or becomes infeasible when there is a large number of feature variables. The Least Absolute Shrinkage and Selection Operator (LASSO) method does variable selection by shrinking less clear effects (including interaction effects) to zero and in this way selects only certain variables and interactions for the model. There are many tree-based methods for subgroup identification. For example, model-based recursive partitioning incorporates parametric models such as generalized linear models into trees. The model incorporated is usually a simple model with only the treatment as covariate. Predictive and prognostic variables are found and incorporated automatically via the tree. The present article gives an overview of these methods and explains how to perform them using the free software environment for statistical computing R (version 3.3.2). A simulated dataset is employed for illustrating the performance of these methods.

Team Performance Indicators Explain Outcome during Women's Basketball Matches at the Olympic Games

The Olympic Games is the pinnacle international sporting competition with team sport coaches interested in key performance indicators to assist the development of match strategies for success. This study examined the relationship between team performance indicators and match outcome during the women’s basketball tournament at the Olympic Games. Team performance indicators were collated from all women’s basketball matches during the 2004–2016 Olympic Games (n = 156) and analyzed via linear (binary logistic regression) and non-linear (conditional interference (CI) classification tree) statistical techniques. The most parsimonious linear model retained “defensive rebounds”, “field-goal percentage”, “offensive rebounds”, “fouls”, “steals”, and “turnovers” with a classification accuracy of 85.6%. The CI classification tree retained four performance indicators with a classification accuracy of 86.2%. The combination of “field-goal percentage”, “defensive rebounds”, “steals”, and “turnovers” provided the greatest probability of winning (91.1%), while a combination of “field-goal percentage”, “steals”, and “turnovers” provided the greatest probability of losing (96.7%). Shooting proficiency and defensive actions were identified as key team performance indicators for Olympic female basketball success. The development of key defensive strategies and/or the selection of athletes highly proficient in defensive actions may strengthen Olympic match success. Incorporation of non-linear analyses may provide teams with superior/practical approaches for elite sporting success.

Explaining Match Outcome During The Men's Basketball Tournament at The Olympic Games

In preparation for the Olympics, there is a limited opportunity for coaches and athletes to interact regularly with team performance indicators providing important guidance to coaches for enhanced match success at the elite level. This study examined the relationship between match outcome and team performance indicators during men's basketball tournaments at the Olympic Games. Twelve team performance indicators were collated from all men's teams and matches during the basketball tournament of the 2004-2016 Olympic Games (n = 156). Linear and non-linear analyses examined the relationship between match outcome and team performance indicator characteristics; namely, binary logistic regression and a conditional interference (CI) classification tree. The most parsimonious logistic regression model retained 'assists', 'defensive rebounds', 'field-goal percentage', 'fouls', 'fouls against', 'steals' and 'turnovers' (delta AIC <0.01; Akaike weight = 0.28) with a classification accuracy of 85.5%. Conversely, four performance indicators were retained with the CI classification tree with an average classification accuracy of 81.4%. However, it was the combination of 'field-goal percentage' and 'defensive rebounds' that provided the greatest probability of winning (93.2%). Match outcome during the men's basketball tournaments at the Olympic Games was identified by a unique combination of performance indicators. Despite the average model accuracy being marginally higher for the logistic regression analysis, the CI classification tree offered a greater practical utility for coaches through its resolution of non-linear phenomena to guide team success.

Weak Relationships between Stint Duration, Physical and Skilled Match Performance in Australian Football

Australian Rules football comprises physical and skilled performance for more than 90 min of play. The cognitive and physiological fatigue experienced by participants during a match may reduce performance. Consequently, the length of time an athlete is on the field before being interchanged (known as a stint), is a key tactic which could maximize the skill and physical output of the Australian Rules athlete. This study developed two methods to quantify the relationship between athlete time on field, skilled and physical output. Professional male athletes (n = 39) from a single elite Australian Rules football club participated, with physical output quantified via player tracking systems across 22 competitive matches. Skilled output was calculated as the sum of involvements performed by each athlete, collected from a commercial statistics company. A random intercept and slope model was built to identify how a team and individuals respond to physical outputs and stint lengths. Stint duration (mins), high intensity running (speeds >14.4 km · hr⁻¹) per minute, meterage per minute and very high intensity running (speeds >25 km·hr⁻¹) per minute had some relationship with skilled involvements. However, none of these relationships were strong, and the direction of influence for each player was varied. Three conditional inference trees were computed to identify the extent to which combinations of physical parameters altered the anticipated skilled output of players. Meterage per minute, player, round number and duration were all related to player involvement. All methods had an average error of 10 to 11 involvements, per player per match. Therefore, other factors aside from physical parameters extracted from wearable technologies may be needed to explain skilled output within Australian Rules football matches.

Predicting Quality of Life Changes in Hemodialysis Patients Using Machine Learning: Generation of an Early Warning System

Objective

To predict changes in the quality of life scores of hemodialysis patients for the coming month and the development of an early warning system using machine learning

Methods

It was a prospective cohort study (one-month duration) at the dialysis center of a tertiary care hospital in Pakistan. The study started on 1^st October 2016. About 78 patients have been enrolled till now. Bachelor of Medicine and Bachelor of Surgery (MBBS) qualified doctors administered a proforma with demographics and the validated Urdu version of World Health Organization Quality Of Life-BREF (WHOQOL-BREF). It was to be repeated after one month to the same patient by the same investigator. Simple statistics were computed using SPSS version 24 (IBM Corp., Armonk, NY) while machine learning was performed using R (version 3.0) and Orange (version 3.1).

Results

Using machine learning algorithms, two models (classification tree and Naïve Bayes) were generated to predict an increase or decrease of 5% in a patient’s WHOQOL-BREF score over one month. The classification tree was selected as the most accurate model with an area under curve (AUC) of 83.3% (accuracy: 81.9%) for the prediction of 5% increase in QOL and an AUC of 76.2% (accuracy: 81.8%) for the prediction of 5% decrease in QOL over the coming month. The factors associated with an increase of QOL by 5% or more over the next month included younger age (<19 years) and higher iron sucrose doses (>278mg/month). Drops in psychological, physical, and social domain scores lead to a decrease of 5% or more in QOL scores over the following month.

Conclusion

An early warning system, dialysis data interpretation for algorithmic-prediction on quality of life (DIAL) was built for the early detection of deteriorating QOL scores in the hemodialysis population using machine learning algorithms. The model pointed out that working on psychological and environmental domains, in particular, may prevent the drop in QOL scores from occurring. DIAL, if implemented on a larger scale, is expected to help patients in terms of ensuring a better QOL and in reducing the financial burden in the long term.

Weight Status and Dental Problems in Early Childhood: Classification Tree Analysis of a National Cohort

A poor quality diet may be a common risk factor for both obesity and dental problems such as caries. The aim of this paper is to use classification tree analysis (CTA) to identify predictors of dental problems in a nationally representative cohort of Irish pre-school children. CTA was used to classify variables and describe interactions between multiple variables including socio-demographics, dietary intake, health-related behaviour, body mass index (BMI) and a dental problem. Data were derived from the second (2010/2011) wave of the 'Growing Up in Ireland' study (GUI) infant cohort at 3 years, n = 9793. The prevalence of dental problems was 5.0% (n = 493). The CTA model showed a sensitivity of 67% and specificity of 58.5% and overall correctly classified 59% of children. Ethnicity was the most significant predictor of dental problems followed by longstanding illness or disability, mother's BMI and household income. The highest prevalence of dental problems was among children who were obese or underweight with a longstanding illness and an overweight mother. Frequency of intake of some foods showed interactions with the target variable. Results from this research highlight the interconnectedness of weight status, dental problems and general health and reinforce the importance of adopting a common risk factor approach when dealing with prevention of these diseases.

Neurocognitive dysfunction in addiction: Testing hypotheses of diffuse versus selective phenotypic dysfunction with a classification-based approach

Neurocognitive dysfunctions are frequently identified in the addictions. These dysfunctions may indicate either diffuse dysfunction or may represent separate facets that have differential importance to the addiction phenotype. In a sample (n = 260) of alcohol and/or stimulant users and controls we measured responses across 7 diverse neurocognitive measures. These measures were Continuous Performance, Delay Discounting, Iowa Gambling, Stroop, Tower, Wisconsin Card Sorting, and Letter Number Sequencing. Comparisons were then made between the drug-dependent groups and controls using analysis of variance and also using a machine learning approach to classify participants based on task performance as substance-dependent or controls in 1 tree and as alcohol and/or stimulant users or controls in a second tree. The analysis of variance showed significant differences between groups on the Delay Discounting (p < .001), Iowa Gambling (p < .001), Letter Number Sequencing (p < .001), and Wisconsin Card Sorting (p < .05) tasks. The first classification tree correctly classified between substance-dependent or controls for 88.3% of participants and classified between alcohol and/or stimulant users or controls for 63.9% of participants. Delay discounting was the first split in both trees and in the substance-dependent and control tree. The analysis of variance results largely replicate previous findings. The machine learning classification tree analysis provides evidence to support the hypothesis that different measures of neurocognitive dysfunction represent different processes. Among them, delay discounting was the most robust in categorizing drug dependence. (PsycINFO Database Record

A Geographic Information-Assisted Temporal Mixture Analysis for Addressing the Issue of Endmember Class and Endmember Spectra Variability

Spectral mixture analysis (SMA) is a common approach for parameterizing biophysical fractions of urban environment and widely applied in many fields. For successful SMA, the selection of endmember class and corresponding spectra has been assumed as the most important step. Thanks to the spatial heterogeneity of natural and urban landscapes, the variability of endmember class and corresponding spectra has been widely considered as the profound error source in SMA. To address the challenging problems, we proposed a geographic information-assisted temporal mixture analysis (GATMA). Specifically, a logistic regression analysis was applied to analyze the relationship between land use/land covers and surrounding socio-economic factors, and a classification tree method was used to identify the present status of endmember classes throughout the whole study area. Furthermore, an ordinary kriging analysis was employed to generate a spatially varying endmember spectra at all pixels in the remote sensing image. As a consequence, a fully constrained temporal mixture analysis was conducted for examining the fractional land use land covers. Results show that the proposed GATMA achieved a promising accuracy with an RMSE of 6.81%, SE of 1.29% and MAE of 2.6%. In addition, comparative analysis result illustrates that a significant accuracy improvement has been found in the whole study area and both developed and less developed areas, and this demonstrates that the variability of endmember class and endmember spectra is essential for unmixing analysis.

Sensitivity and specificity of multichannel surface electromyography in diagnosing fecal incontinence

BACKGROUND

Assessment of the neurocontrol of the external anal sphincter has long been restricted to investigating patients by invasive tools. Less invasive techniques have been regarded less suitable for diagnosis.

OBJECTIVE

The aim was to develop a surface electromyography-based algorithm to facilitate fecal incontinence diagnosis, and to assess its sensitivity and specificity.

DESIGN

Data analysis from a single center prospective study.

PATIENTS

All patients from colorectal surgery office were considered. They underwent a structured interview, a general physical and proctologic examination. Patients with diagnosed fecal incontinence (Fecal Incontinence Severity Index >10) were included into the study group. The control group consisted of healthy volunteers that scored 5 or less and had negative history and physical exam. Both groups underwent the same tests (rectoscopy, anorectal manometry, transanal ultrasonography, multichannel surface electromyography and assessment of anal reflexes).

METHODS

EMG results were analyzed to find parameters that would facilitate fecal incontinence diagnosis.

OUTCOME MEASURES

Sensitivity and specificity of surface electromyography, to diagnose fecal incontinence, were assessed.

RESULTS

A total of 49 patients were included in the study group (mean age ± SD 58.9 ± 13.8). The control group (n = 49) gender matched the study group (mean age ± SD 45.4 ± 15.1). The constructed classification tree, based on surface electromyography results, correctly classified 97% of cases. Thee sensitivity and specificity of this classification tree, to diagnose FI, was 96% and 98% respectively.

LIMITATIONS

The age of women in the control group differs significantly from mean age of other groups.

CONCLUSIONS

Surface electromyography is an good tool to facilitate diagnosing of fecal incontinence.

A right whale pootree: classification trees of faecal hormones identify reproductive states in North Atlantic right whales (Eubalaena glacialis)

Immunoassay of hormone metabolites extracted from faecal samples of free-ranging large whales can provide biologically relevant information on reproductive state and stress responses. North Atlantic right whales (Eubalaena glacialis Müller 1776) are an ideal model for testing the conservation value of faecal metabolites. Almost all North Atlantic right whales are individually identified, most of the population is sighted each year, and systematic survey effort extends back to 1986. North Atlantic right whales number <500 individuals and are subject to anthropogenic mortality, morbidity and other stressors, and scientific data to inform conservation planning are recognized as important. Here, we describe the use of classification trees as an alternative method of analysing multiple-hormone data sets, building on univariate models that have previously been used to describe hormone profiles of individual North Atlantic right whales of known reproductive state. Our tree correctly classified the age class, sex and reproductive state of 83% of 112 faecal samples from known individual whales. Pregnant females, lactating females and both mature and immature males were classified reliably using our model. Non-reproductive [i.e. 'resting' (not pregnant and not lactating) and immature] females proved the most unreliable to distinguish. There were three individual males that, given their age, would traditionally be considered immature but that our tree classed as mature males, possibly calling for a re-evaluation of their reproductive status. Our analysis reiterates the importance of considering the reproductive state of whales when assessing the relationship between cortisol concentrations and stress. Overall, these results confirm findings from previous univariate statistical analyses, but with a more robust multivariate approach that may prove useful for the multiple-analyte data sets that are increasingly used by conservation physiologists.

A novel statistical method for behaviour sequence analysis and its application to birdsong

Complex vocal signals, such as birdsong, contain acoustic elements that differ in both order and duration. These elements may convey socially relevant meaning, both independently and through their interactions, yet statistical methods that combine order and duration data to extract meaning have not, to our knowledge, been fully developed. Here we design novel semi-Markov methods, Bayesian estimation and classification trees to extract order and duration information from behavioural sequences and apply these methods to songs produced by male European starlings, Sturnus vulgaris, in two social contexts in which the function of song differs: a spring (breeding) and autumn (nonbreeding) context. Additionally, previous data indicate that damage to the medial preoptic nucleus (POM), a brain area known to regulate male sexually motivated behaviour, affects structural aspects of starling song such that males in a sexually relevant context (i.e. spring) sing shorter songs than appropriate for this context. We further test the utility of our statistical approach by comparing attributes of song structure in POM-lesioned males to song produced by control spring and autumn males. Spring and autumn songs were statistically separable based on the duration and order of phrase types. Males produced more structurally complex aspects of song in spring than in autumn. Spring song was also longer and more stereotyped than autumn song, both attributes used by females to select mates. Songs produced by POM-lesioned males in some cases fell between measures of spring and autumn songs but differed most from songs produced by autumn males. Overall, these statistical methods can effectively extract biologically meaningful information contained in many behavioural sequences given sufficient sample sizes and replication numbers.

A novel statistical method for behaviour sequence analysis and its application to birdsong

Complex vocal signals, such as birdsong, contain acoustic elements that differ in both order and duration. These elements may convey socially relevant meaning, both independently and through their interactions, yet statistical methods that combine order and duration data to extract meaning have not, to our knowledge, been fully developed. Here we design novel semi-Markov methods, Bayesian estimation and classification trees to extract order and duration information from behavioural sequences and apply these methods to songs produced by male European starlings, Sturnus vulgaris, in two social contexts in which the function of song differs: a spring (breeding) and autumn (nonbreeding) context. Additionally, previous data indicate that damage to the medial preoptic nucleus (POM), a brain area known to regulate male sexually motivated behaviour, affects structural aspects of starling song such that males in a sexually relevant context (i.e. spring) sing shorter songs than appropriate for this context. We further test the utility of our statistical approach by comparing attributes of song structure in POM-lesioned males to song produced by control spring and autumn males. Spring and autumn songs were statistically separable based on the duration and order of phrase types. Males produced more structurally complex aspects of song in spring than in autumn. Spring song was also longer and more stereotyped than autumn song, both attributes used by females to select mates. Songs produced by POM-lesioned males in some cases fell between measures of spring and autumn songs but differed most from songs produced by autumn males. Overall, these statistical methods can effectively extract biologically meaningful information contained in many behavioural sequences given sufficient sample sizes and replication numbers.

Can echocardiography and ECG discriminate hereditary transthyretin V30M amyloidosis from hypertrophic cardiomyopathy?

OBJECTIVE

Hereditary transthyretin (ATTR) amyloidosis with increased left ventricular wall thickness could easily be misdiagnosed by echocardiography as hypertrophic cardiomyopathy (HCM). Our aim was to create a diagnostic tool based on echocardiography and ECG that could optimise identification of ATTR amyloidosis.

METHODS

Data were analysed from 33 patients with biopsy proven ATTR amyloidosis and 30 patients with diagnosed HCM. Conventional features from ECG were acquired as well as two dimensional and Doppler echocardiography, speckle tracking derived strain and tissue characterisation analysis. Classification trees were used to select the most important variables for differentiation between ATTR amyloidosis and HCM.

RESULTS

The best classification was obtained using both ECG and echocardiographic features, where a QRS voltage >30 mm was diagnostic for HCM, whereas in patients with QRS voltage <30 mm, an interventricular septal/posterior wall thickness ratio (IVSt/PWt) >1.6 was consistent with HCM and a ratio <1.6 supported the diagnosis of ATTR amyloidosis. This classification presented both high sensitivity (0.939) and specificity (0.833).

CONCLUSION

Our study proposes an easily interpretable classification method for the differentiation between HCM and increased left ventricular myocardial thickness due to ATTR amyloidosis. Our combined echocardiographic and ECG model could increase the ability to identify ATTR cardiac amyloidosis in clinical practice.

Individualized Prediction of Heat Stress in Firefighters: A Data-Driven Approach Using Classification and Regression Trees

The purpose of this study was to explore data-driven models, based on decision trees, to develop practical and easy to use predictive models for early identification of firefighters who are likely to cross the threshold of hyperthermia during live-fire training. Predictive models were created for three consecutive live-fire training scenarios. The final predicted outcome was a categorical variable: will a firefighter cross the upper threshold of hyperthermia - Yes/No. Two tiers of models were built, one with and one without taking into account the outcome (whether a firefighter crossed hyperthermia or not) from the previous training scenario. First tier of models included age, baseline heart rate and core body temperature, body mass index, and duration of training scenario as predictors. The second tier of models included the outcome of the previous scenario in the prediction space, in addition to all the predictors from the first tier of models. Classification and regression trees were used independently for prediction. The response variable for the regression tree was the quantitative variable: core body temperature at the end of each scenario. The predicted quantitative variable from regression trees was compared to the upper threshold of hyperthermia (38°C) to predict whether a firefighter would enter hyperthermia. The performance of classification and regression tree models was satisfactory for the second (success rate = 79%) and third (success rate = 89%) training scenarios but not for the first (success rate = 43%). Data-driven models based on decision trees can be a useful tool for predicting physiological response without modeling the underlying physiological systems. Early prediction of heat stress coupled with proactive interventions, such as pre-cooling, can help reduce heat stress in firefighters.

Tumor progression driven by pathways activating matrix metalloproteinases and their inhibitors

BACKGROUND

Laryngeal squamous cell carcinoma (LSCC) is still a problem worldwide. In some publications interactions between the expression of matrix metalloproteinases (MMPs), particularly MMP-2 and MMP-9, and their tissue inhibitors (TIMPs) implicated during cancer progression were suggested.

METHODS

The immunohistochemical staining using primary antibody against MMP-2, MMP-9, TIMP-1, TIMP-2, and TIMP-3 were performed. The research group consists of primary N(0) LSCC (20 cases), primary N(+) LSCC (17 cases), and 18 cases of normal mucosa.

RESULTS

Studied MMPs and TIMPs were localized in tumor cells and tumor stroma compartment. MMP-2 expression was higher in stroma compared to tumor cells. MMP-9, TIMP-1, TIMP-2, and TIMP-3 expression was higher in tumor cells than in tumor stroma (P < 0.05). In tumor stroma MMP-2, MMP-9, TIMP-1, and TIMP-3 expression, in LSCC N(0) vs. LSCC N(+) was significantly higher (P < 0.05). The ratios between MMP-2 and TIMP-3 expression were statistically significant (N(0) vs. N(+); P = 0.012). The analyses using classification trees predicted the probability of metastases according to TIMP-3/MMP-14/MMP-2 and MMP-9/TIMP-1 expression levels.

CONCLUSIONS

The presence of MMP-2, MMP-9, TIMP-1, TIMP-2, TIMP-3 expression in tumor cells and in tumor stroma, and additionally different expression according to lymph node involvement suggested of their impact during cancer progression. The significant correlation between TIMP-3 expression and the presence of lymph node metastases and MMP-2 expression might suggest the importance of TIMP-3 as a prognostic factor during tumor progression. The evaluation of molecular markers which participate in MMP-2 activation pathway have a major impact during metastasis.

TARV: tree-based analysis of rare variants identifying risk modifying variants in CTNNA2 and CNTNAP2 for alcohol addiction

Since the development of next generation sequencing (NGS) technology, researchers have been extending their efforts on genome-wide association studies (GWAS) from common variants to rare variants to find the missing inheritance. Although various statistical methods have been proposed to analyze rare variants data, they generally face difficulties for complex disease models involving multiple genes. In this paper, we propose a tree-based analysis of rare variants (TARV) that adopts a nonparametric disease model and is capable of exploring gene-gene interactions. We found that TARV outperforms the sequence kernel association test (SKAT) in most of our simulation scenarios, and by notable margins in some cases. By applying TARV to the study of addiction: genetics and environment (SAGE) data, we successfully detected gene CTNNA2 and its 43 specific variants that increase the risk of alcoholism in women, with an odds ratio (OR) of 1.94. This gene has not been detected in the SAGE data. Post hoc literature search also supports the role of CTNNA2 as a likely risk gene for alcohol addiction. In addition, we also detected a plausible protective gene CNTNAP2, whose 97 rare variants can reduce the risk of alcoholism in women, with an OR of 0.55. These findings suggest that TARV can be effective in dissecting genetic variants for complex diseases using rare variants data.

Towards enhancing the performance of multi-parameter patient monitors

Multi-parameter patient monitors (MPMs) have become increasingly important in providing quality healthcare to patients. It is well known in the medical community that there exists an intrinsic relationship between different vital parameters in a healthy person, these include heart rate, blood pressure, respiration rate and oxygen saturation. For example, an increase in blood pressure would lead to a decrease in the heart rate, and vice versa. Although it is likely to improve the performance of MPM systems, this fact is not explored in engineering research. In this work, experiments show that deriving additional features to capture the intrinsic relationship between the vital parameters, the alarm accuracy (sensitivity), no-alarm accuracy (specificity) and the overall performance of MPMs can be improved. The geometric mean of the product of all the vital parameters taken in pairs of two was used to capture the intrinsic relationship between the different parameters. An improvement of 10.55% for sensitivity, 0.32% for specificity and an overall performance improvement of 1.03% was obtained, compared to the baseline system using classification and regression tree with the four vital parameters.

Clinical factors associated with statins prescription in acute ischemic stroke patients: findings from the Lombardia Stroke Registry

BACKGROUND

Statins, due to their well-established pleiotropic effects, have noteworthy benefits in stroke prevention. Despite this, a significant proportion of high-risk patients still do not receive the recommended therapeutic regimens, and many others discontinue treatment after being started on them. The causes of non-adherence to current guidelines are multifactorial, and depend on both physicians and patients. The aim of this study is to identify the factors influencing statin prescription at Stroke Unit (SU) discharge.

METHODS

This study included 12,750 patients enrolled on the web-based Lombardia Stroke Registry (LRS) from July 2009 to April 2012 and discharged alive, with a diagnosis of ischemic stroke or transient ischemic attack (TIA) and without contra-indication to statin therapy. By logistic regression analysis and classification trees, we evaluated the impact of demographic data, risk factors, tPA treatment, in-hospital procedures and complications on statin prescription rate at discharge.

RESULTS

We observed a slight increase in statins prescription during the study period (from 39.1 to 43.9%). Lower age, lower stroke severity and prestroke disability, the presence of atherothrombotic/lacunar risk factors, a diagnosis of non-cardioembolic stroke, tPA treatment, the absence of in-hospital complications, with the sole exception of hypertensive fits and hyperglycemia, were the patient-related predictors of adherence to guidelines by physicians. Overall, dyslipidemia appears as the leading factor, while TOAST classification does not reach statistical significance.

CONCLUSIONS

In our region, Lombardia, adherence to guidelines in statin prescription at Stroke Unit discharge is very different from international goals. The presence of dyslipidemia remains the main factor influencing statin prescription, while the presence of well-defined atherosclerotic etiopathogenesis of stroke does not enhance statin prescription. Some uncertainties about the risk/benefit of statin therapy in stroke etiology subtypes (cardioembolism, other or undetermined causes) may partially justify the underuse of statin in ischemic stroke. The differences that exist between current international guidelines may prevent a more widespread use of statin and should be clarified in a consensus.

Classification tree analysis of the factors influencing injury-related disability caused by the Wenchuan earthquake

OBJECTIVE

To identify the factors that influenced the risk of injury-related disability caused by the Wenchuan earthquake.

METHODS

A chi-squared automatic interaction detection (CHAID) classification tree analysis was used to retrospectively analyse clinical data from patients who underwent surgical treatment for earthquake-related injuries in the first 5 days after the earthquake. The CHAID classification tree explored the relationships between the development of disability and potential influencing factors including sex, age, time interval between injury and treatment, wound type, preoperative and postoperative haemoglobin levels, and operation time.

RESULTS

A total of 334 patients underwent surgery; of these, 113 (33.8%) were discharged with varying degrees of permanent disability. The CHAID classification tree showed that children (≤ 17 years old), a long time interval between injury and treatment, an open wound and a low preoperative haemoglobin level were significant risk factors for disability.

CONCLUSION

The results of this study can help to stratify patients according to their medical needs and to help allocate the available resources efficiently to ensure the best outcomes for injured patients during future earthquakes.

The process and utility of classification and regression tree methodology in nursing research

Aim

This paper presents a discussion of classification and regression tree analysis and its utility in nursing research.

Background

Classification and regression tree analysis is an exploratory research method used to illustrate associations between variables not suited to traditional regression analysis. Complex interactions are demonstrated between covariates and variables of interest in inverted tree diagrams.

Design

Discussion paper.

Data sources

English language literature was sourced from eBooks, Medline Complete and CINAHL Plus databases, Google and Google Scholar, hard copy research texts and retrieved reference lists for terms including classification and regression tree* and derivatives and recursive partitioning from 1984–2013.

Discussion

Classification and regression tree analysis is an important method used to identify previously unknown patterns amongst data. Whilst there are several reasons to embrace this method as a means of exploratory quantitative research, issues regarding quality of data as well as the usefulness and validity of the findings should be considered.

Implications for Nursing Research

Classification and regression tree analysis is a valuable tool to guide nurses to reduce gaps in the application of evidence to practice. With the ever-expanding availability of data, it is important that nurses understand the utility and limitations of the research method.

Conclusion

Classification and regression tree analysis is an easily interpreted method for modelling interactions between health-related variables that would otherwise remain obscured. Knowledge is presented graphically, providing insightful understanding of complex and hierarchical relationships in an accessible and useful way to nursing and other health professions.

Stroke damage detection using classification trees on electrical bioimpedance cerebral spectroscopy measurements

After cancer and cardio-vascular disease, stroke is the third greatest cause of death worldwide. Given the limitations of the current imaging technologies used for stroke diagnosis, the need for portable non-invasive and less expensive diagnostic tools is crucial. Previous studies have suggested that electrical bioimpedance (EBI) measurements from the head might contain useful clinical information related to changes produced in the cerebral tissue after the onset of stroke. In this study, we recorded 720 EBI Spectroscopy (EBIS) measurements from two different head regions of 18 hemispheres of nine subjects. Three of these subjects had suffered a unilateral haemorrhagic stroke. A number of features based on structural and intrinsic frequency-dependent properties of the cerebral tissue were extracted. These features were then fed into a classification tree. The results show that a full classification of damaged and undamaged cerebral tissue was achieved after three hierarchical classification steps. Lastly, the performance of the classification tree was assessed using Leave-One-Out Cross Validation (LOO-CV). Despite the fact that the results of this study are limited to a small database, and the observations obtained must be verified further with a larger cohort of patients, these findings confirm that EBI measurements contain useful information for   assessing on the health of brain tissue after stroke and supports the hypothesis that classification features based on Cole parameters, spectral information and the geometry of EBIS measurements are useful to differentiate between healthy and stroke damaged brain tissue.