Assessing effect heterogeneity of a randomized treatment using conditional inference trees

Treatment effect heterogeneity occurs when individual characteristics influence the effect of a treatment. We propose a novel approach that combines prognostic score matching and conditional inference trees to characterize effect heterogeneity of a randomized binary treatment. One key feature that distinguishes our method from alternative approaches is that it controls the Type I error rate, that is, the probability of identifying effect heterogeneity if none exists and retains the underlying subgroups. This feature makes our technique particularly appealing in the context of clinical trials, where there may be significant costs associated with erroneously declaring that effects differ across population subgroups. Treatment effect heterogeneity trees are able to identify heterogeneous subgroups, characterize the relevant subgroups and estimate the associated treatment effects. We demonstrate the efficacy of the proposed method using a comprehensive simulation study and illustrate our method using a nutrition trial dataset to evaluate effect heterogeneity within a patient population.

Trends and factors related to adolescent pregnancies: an incidence trend and conditional inference trees analysis of northern Nicaragua demographic surveillance data

Background We aimed to identify the 2001–2013 incidence trend, and characteristics associated with adolescent pregnancies reported by 20–24-year-old women. Methods A retrospective analysis of the Cuatro Santos Northern Nicaragua Health and Demographic Surveillance 2004–2014 data on women aged 15–19 and 20–24. To calculate adolescent birth and pregnancy rates, we used the first live birth at ages 10–14 and 15–19 years reported by women aged 15–19 and 20–24 years, respectively, along with estimates of annual incidence rates reported by women aged 20–24 years. We conducted conditional inference tree analyses using 52 variables to identify characteristics associated with adolescent pregnancies. Results The number of first live births reported by women aged 20–24 years was 361 during the study period. Adolescent pregnancies and live births decreased from 2004 to 2009 and thereafter increased up to 2014. The adolescent pregnancy incidence (persons-years) trend dropped from 2001 (75.1 per 1000) to 2007 (27.2 per 1000), followed by a steep upward trend from 2007 to 2008 (19.1 per 1000) that increased in 2013 (26.5 per 1000). Associated factors with adolescent pregnancy were living in low-education households, where most adults in the household were working, and high proportion of adolescent pregnancies in the local community. Wealth was not linked to teenage pregnancies. Conclusions Interventions to prevent adolescent pregnancy are imperative and must bear into account the context that influences the culture of early motherhood and lead to socioeconomic and health gains in resource-poor settings.

Compound Endoscopic Morphological Features for Identifying Non-Pedunculated Lesions ≥20 mm with Intramucosal Neoplasia

Background: The major limitation of piecemeal endoscopic mucosal resection (EMR) is the inaccurate histological assessment of the resected specimen, especially in cases of submucosal invasion. Objective: To classify non-pedunculated lesions ≥20 mm based on endoscopic morphological features, in order to identify those that present intramucosal neoplasia (includes low-grade neoplasia and high-grade neoplasia) and are suitable for piecemeal EMR. Design: A post-hoc analysis from an observational prospective multicentre study conducted by 58 endoscopists at 17 academic and community hospitals was performed. Unbiased conditional inference trees (CTREE) were fitted to analyse the association between intramucosal neoplasia and the lesions’ endoscopic characteristics. Result: 542 lesions from 517 patients were included in the analysis. Intramucosal neoplasia was present in 484 of 542 (89.3%) lesions. A conditional inference tree including all lesions’ characteristics assessed with white light imaging and narrow-band imaging (NBI) found that ulceration, pseudodepressed type and sessile morphology changed the accuracy for predicting intramucosal neoplasia. In ulcerated lesions, the probability of intramucosal neoplasia was 25% (95%CI: 8.3–52.6%; p < 0.001). In non-ulcerated lesions, its probability in lateral spreading lesions (LST) non-granular (NG) pseudodepressed-type lesions rose to 64.0% (95%CI: 42.6–81.3%; p < 0.001). Sessile morphology also raised the probability of intramucosal neoplasia to 86.3% (95%CI: 80.2–90.7%; p < 0.001). In the remaining 319 (58.9%) non-ulcerated lesions that were of the LST-granular (G) homogeneous type, LST-G nodular-mixed type, and LST-NG flat elevated morphology, the probability of intramucosal neoplasia was 96.2% (95%CI: 93.5–97.8%; p < 0.001). Conclusion: Non-ulcerated LST-G type and LST-NG flat elevated lesions are the most common non-pedunculated lesions ≥20 mm and are associated with a high probability of intramucosal neoplasia. This means that they are good candidates for piecemeal EMR. In the remaining lesions, further diagnostic techniques like magnification or diagnostic +/− therapeutic endoscopic submucosal dissection should be considered.

The semantics of conversion nouns and -ing nominalizations: A quantitative and theoretical perspective

This paper addresses a fundamental problem of derivational morphology: which meanings are possible for the words of a given morphological category, which forms can be chosen to express a given meaning, and what is the role of the base in these mappings of form and meaning? In a broad empirical study we examine the extent to which two types of nominalizations in English – conversion nouns and -ing nominalizations – can express either eventive or referential readings, can be quantified as either count or mass, and can be based on verbs of particular aspectual classes (state, activity, accomplishment, achievement, semelfactive). Past literature (for example, Grimshaw 1990 Brinton 1995, 1998 Borer 2013) has suggested an association between conversion nominalization, count quantification, and referential reading on the one hand, and between -ing nominalization, mass quantification and eventive reading on the other. Using a subset of the data reported in Andreou & Lieber (2020), we give statistical evidence that the relationship between morphological form, type of quantification, and aspectual class of base verb is neither categorical, as the literature suggests, nor completely free, but rather is probabilistic. We provide both a univariate analysis and a multivariate analysis (using conditional inference trees) that show that the relationship among the variables of morphological form, eventivity, quantification and aspectual class of base is complex. Tendencies sometimes go in the direction suggested by past literature (e.g. -ing forms tend to be eventive), but sometimes contradict past predictions (conversion also tends to be eventive). We also document that an important role is played by the specific verb underlying the nominalization rather than the aspectual class of verb. Finally, we consider what the pattern of polysemy that we uncover suggests with respect to theoretical modeling, looking at syntactic models (Distributed Morphology), lexical semantic models (the Lexical Semantic Framework), Analogical Models, and Distributional Semantics.

What Drives Task Performance in Fluency Tasks in People With HIV?

Introduction: Fluency tasks require language (i.e., semantics, phonological output lexicon, and phonological assembly) and executive functions (i.e., inhibition; mental set shifting; updating, and monitoring). Little is known about whether people with human immunodeficiency virus (HIV) are more impaired on a specific type of fluency task and what aspects of language and executive functions drive such performance.Aims: To understand (1) whether people with HIV are more impaired in animal, letter, or unconstrained fluency relative to a normative sample; (2) whether there exist differences between tasks relative to the total number of words; and (3) which aspects of executive function and language are involved in their performance.Methods: Data from animal, letter, and unconstrained fluency of 50 Spanish-speaking people with HIV were analyzed. The number of switches and mean cluster size for each task and 10 word properties (e.g., frequency, age of acquisition, length in graphemes) for each of the correct words were measured. A chi-square test was used to address Aim 1, linear mixed effects models for Aim 2, and random forests and conditional inference trees for Aim 3. The results were cross-validated with a normative sample.Results: People with HIV were not more impaired in animal, letter, or unconstrained fluency relative to a normative sample. People with HIV produced fewer words in letter fluency compared to animal and unconstrained fluency. In addition, they produced fewer words in animal fluency compared to unconstrained fluency. Number of switches emerged as the most important variable to predict the total number of correct words when considering the three tasks together and for each task separately. Word frequency was relevant to predict animal fluency, age of acquisition to predict letter fluency, and cluster size to predict unconstrained fluency. These results were cross-validated with the exception cluster size.Conclusion: People with HIV rely on language (phonological output lexicon, not necessarily semantics) and executive functioning (updating and monitoring) to produce words in fluency tasks. These results concur with the current literature. Future work may correlate fluency scores with other tests measuring language and executive functions or study other types of fluency tasks (e.g., action, cities, supermarket, and professions).

Right ventricular reverse remodeling assessed by echocardiography as a new goal-oriented treatment strategy in PAH patients

Background While in heart failure with reduced ejection fraction, left ventricular reverse remodeling assessed by transthoracic echocardiography (TTE) is associated with better prognosis, right ventricular reverse remodeling (RVRR) was less investigated in pulmonary arterial hypertension (PAH) Purpose We aimed to investigate whether RVVR assessed by echocardiography could help to stratify PAH patient's prognosis. Methods Between 2002 and 2019, all consecutive PAH patients were included, treated and followed in a single PAH center in accordance with the current ESC/ERS guidelines. In addition to regular risk stratification parameters, we measured several echocardiographic RV systolic function and size parameters, including tricuspid annular plane systolic excursion (TAPSE, mm) or RV-end diastolic area (cm2) from apical-4 chamber view both at baseline, 1-year of follow-up as well as their change. Primary composite outcome was three-year transplant-free survival and death from all cause from the 1-year evaluation. Conditional inference trees were used to determine which TTE parameters and cutoffs values were associated with primary outcome from hierarchy of multiple covariates in multivariable Cox regression analysis. Kaplan-Meier curves were then drawn and compared with log-rank test. Results 126 incident PAH patients were included (63% female, mean age 59±18 yo), mainly due to connectivite-tissue disease and idiopathic PAH (26% and 22% respectively). At baseline, mean pulmonary arterial pressure was 42 (33, 52) mmHg. At 1-y follow-up under pulmonary vasodilation therapy, NYHA (p&lt;0.01), NTproBNP (p&lt;0.01), mean pulmonary arterial pressure (p&lt;0.01) and cardiac index (p&lt;0.01) were significantly improved compared to baseline. Conditional inference trees showed that 1-year TAPSE gain &gt;1 mm and 1-year RV end-diastolic area decreased &gt;2 cm2 were associated with 3-year transplant-free survival in multivariable Cox regression analysis (HR=0.23, 95% CI [0.08–0.61] p=0,0035, HR=0.34, 95% CI [0.12–0.94], p=0.038). Simple score from 0 (absence of RVRR), 1 (partial RVRR) and 2 (complete RVRR), describing the number of TTE parameters reach at 1-year was then investigated. Patients with complete RVRR depicted better transplant-free survival than partial or absence of RVRR, log-rank p&lt;0.001 (figure). Conclusion Complete reverse remodeling from right ventricular size and function could represent a new goal-oriented treatment strategy in PAH patients. FUNDunding Acknowledgement Type of funding sources: None. RVRR survival curves

‘An Eye for an Aye’: linguistic and political backlash and conformity in eighteenth-century Scots

This study examines the effect of political change on the use of written Scots during the eighteenth century. In particular, it compares a cross-section of texts from literate Scottish society, with works from certain politically-active authors, who identified strongly as pro- or anti-Union following the creation of the British state in 1707. The proportion of Scots to English lexemes in their writing is explored using conditional inference trees and random forests, in a small, purpose-built corpus. Use of Scots is shown to differ between the two groups, with specific extralinguistic factors encouraging or suppressing the presence of written Scots. Frequency of Scots is also found to be influenced by the political ideology of the politicised authors. These results are linked to the Scottish political scene during the eighteenth century, as well as general processes of change over time.

Variation in negation in Seto

Seto is an exceptional language in the Uralic family due to its systematic use of postverbal negation, although preverbal and double negation marking are also used. Postverbal negation is still the most frequent and unmarked pattern occurring in about 74% of negative clauses in Seto. This paper analyzes variation between pre- and postverbal negation in East Seto (spoken in present-day Russia), based on data gathered during fieldwork trips in 2010–2013. By applying quantitative methods that are used in variationist studies (regression modelling, conditional inference trees, and random forests), we determine the variables affecting the choice between pre- and postverbal negation. Marked preverbal negation occurs more likely with first and third person, cognition verbs, and present tense, all of which are often used in fixed expressions like I don’t know. We also found a strong structural persistence effect in the data and remarkable differences between individual speakers.

Conditional inference trees in the assessment of tree mortality rates in the transitional mixed forests of Atlantic Canada

Long-term predictions of forest dynamics, including forecasts of tree growth and mortality, are central to sustainable forest-management planning. Although often difficult to evaluate, tree mortality rates under different abiotic and biotic conditions are vital in defining the long-term dynamics of forest ecosystems. In this study, we have modeled tree mortality rates using conditional inference trees (CTREE) and multi-year permanent sample plot data sourced from an inventory with coverage of New Brunswick (NB), Canada. The final CTREE mortality model was based on four tree- and three stand-level terms together with two climatic terms. The correlation coefficient (R2) between observed and predicted mortality rates was 0.67. High cumulative annual growing degree-days (GDD) was found to lead to increased mortality in 18 tree species, including Betula papyrifera, Picea mariana, Acer saccharum, and Larix laricina. In another ten species, including Abies balsamea, Tsuga canadensis, Fraxinus americana, and Fagus grandifolia, mortality rates tended to be higher in areas with high incident solar radiation. High amounts of precipitation in NB’s humid maritime climate were also found to contribute to heightened tree mortality. The relationship between high GDD, solar radiation, and high mortality rates was particularly strong when precipitation was also low. This would suggest that although excessive soil water can contribute to heightened tree mortality by reducing the supply of air to the roots, occasional drought in NB can also contribute to increased mortality events. These results would have significant implications when considered alongside regional climate projections which generally entail both components of warming and increased precipitation.

The canine counts! Significance of a craniodental measure to describe sexual dimorphism in canids: Golden jackals (Canis aureus) and African wolves (Canis lupaster)

Sexual dimorphism is a widespread phenomenon among mammals, including carnivorans. While sexual dimorphism in golden jackals (Canis aureus) has been analysed in the past, in the related and apparently convergent canid, the African wolf (Canis lupaster), it is poorly studied and showed to be relatively small. Previously, sexual size dimorphism (SSD) research in these species was mostly based on skull and body measurements. In our study, we also included dental measurements, namely the diameter of the canine. We used 11 measured sections of 104 adult specimens, comprising 61 golden jackal and 43 African wolf skulls. Data analyses were carried out through logistic regression and conditional inference trees (CIT). To compare the results of SSD to other species, sexual dimorphism indices (SDI) were calculated. Golden jackals and African wolves show significant sexual size dimorphism, both in cranial and dental size. The logistic regression revealed that the mesiodistal diameter of the upper canine is most effective in discerning the sexes. The difference in the calculated SDI of the canine diameter between the sexes amounted to 8.71 in golden jackals and 14.11 in African wolves, respectively—with regional diversity. Thus, the canine diameter is an important measure to investigate SSD as well as an easy tool to apply in the field.