Search

Traditional methods for evaluating a potential treatment have focused on the average treatment effect. However, there exist situations where individuals can experience significantly heterogeneous responses to a treatment. In these situations, one needs to account for the differences among individuals when estimating the treatment effect. Li et al. (2022) proposed a method based on random forest of interaction trees (RFIT) for a binary or categorical treatment variable, while incorporating the propensity score in the construction of random forest. Motivated by the need to evaluate the effect of tutoring sessions at a Math and Stat Learning Center (MSLC), we extend their approach to an ordinal treatment variable. Our approach improves upon RFIT for multiple treatments by incorporating the ordered structure of the treatment variable into the tree growing process. To illustrate the effectiveness of our proposed method, we conduct simulation studies where the results show that our proposed method has a lower mean squared error and higher optimal treatment classification, and is able to identify the most important variables that impact the treatment effect. We then apply the proposed method to estimate how the number of visits to the MSLC impacts an individual student’s probability of passing an introductory statistics course. Our results show that every student is recommended to go to the MSLC at least once and some can drastically improve their chance of passing the course by going the optimal number of times suggested by our analysis.

We expand methods for estimating an optimal treatment regime (OTR) from the personalized medicine literature to educational data mining applications. As part of this development, we detail and modify the current state-of-the-art, assess the efficacy of the approaches for student success studies, and provide practitioners the machinery to apply the methods in their specific problems. Our particular interest is to estimate an optimal treatment regime for students enrolled in an introductory statistics course at San Diego State University (SDSU). The available treatments are combinations of three programs SDSU implemented to foster student success in this large enrollment, bottleneck STEM course. We leverage tree-based reinforcement learning approaches based on either an inverse probability-weighted purity measure or an augmented probability-weighted purity measure. The thereby deduced OTR promises to significantly increase the average grade in the introductory course and also reveals the need for program recommendations to students as only very few, on their own, selected their optimal treatment.

Despite the importance of monsoon rainfall to over half of the world's population, many climate models of the current generation struggle to capture some of the major features of the various monsoon systems. Studies of the development of errors in several tropical regions have shown that they start to develop very quickly, within the first few days of a model simulation, and can then persist to climate timescales. Understanding the sources of such errors requires the combination of various modelling techniques and sensitivity experiments of varying complexity. Here, we demonstrate how such analysis can shed light on the way in which monsoon errors develop, their local and remote drivers and feedbacks. We make use of the seamless modelling approach adopted by the Met Office, whereby different applications of the Met Office Unified Model (MetUM) use essentially the same model configuration (dynamical core and physical parameterisations) across a range of spatial and temporal scales. Using the Asian summer monsoon (ASM) as an example, we show that error patterns in circulation and rainfall over the ASM region in the MetUM are similar between multidecadal climate simulations and seasonal hindcasts initialised in spring. Analysis of the development of these errors on both short-range and seasonal timescales following model initialisation suggests that both the Maritime Continent and the oceans around the Philippines play a role in the development of East Asian summer monsoon errors, with the Indian summer monsoon region providing an additional contribution, while the errors over the Indian summer monsoon region itself appear to arise locally. Regional modelling with various lateral boundary locations helps to separate local and remote contributions to the errors, while regional relaxation experiments shed light on the influence of errors developing within particular areas on the region as a whole.

Background and Objectives Selecting applications for college admission is critical for university operation and development. This paper leverages machine learning techniques to support enrollment management teams through data-informed decision-making in this otherwise laborious admissions processing. Research Design and Measures Two aspects of university admissions are considered. An ensemble learning approach, through the SuperLearner algorithm, is used to predict student show (yield) rate. The goal is to improve prediction accuracy to minimize over- or under-enrollment. A combinatorial optimization framework is proposed to weigh academic performance and experiential factors for ranking and selecting students for admission. This framework uses simulated annealing, and an efficacy study is presented to evaluate performance. Results The proposed framework is illustrated for selecting an incoming class by optimizing predicted graduation rate and by developing an eligibility index. Each example presents a selection process under potential academic performance and experiential factor targets a university may place on an admitted class. R code is provided for higher education researchers and practitioners to apply the proposed methods in their own settings.

Serotonin plays an important role in the development of brainstem circuits that control breathing. Here, we test the hypothesis that developmental nicotine exposure (DNE) alters the breathing-related motor response to serotonin (5HT). Pregnant rats were exposed to nicotine or saline, and brainstem-spinal cord preparations from 1- to 5-day-old pups were studied in a split-bath configuration, allowing drugs to be applied selectively to the medulla or spinal cord. The activity of the fourth cervical ventral nerve roots (C4VR), which contain axons of phrenic motoneurons, was recorded. We applied 5HT alone or together with antagonists of 5HT1A, 5HT2A, or 5HT7 receptor subtypes. In control preparations, 5HT applied to the medulla consistently reduced C4VR frequency and this reduction could not be blocked by any of the three antagonists. In DNE preparations, medullary 5HT caused a large and sustained frequency increase (10 min), followed by a sustained decrease. Notably, the transient increase in frequency could be blocked by the independent addition of any of the antagonists. Experiments with subtype-specific agonists suggest that the 5HT7 subtype may contribute to the increased frequency response in the DNE preparations. Changes in C4VR burst amplitude in response to brainstem 5HT were uninfluenced by DNE. Addition of 5HT to the caudal chamber modestly increased phasic and greatly increased tonic C4VR activity, but there were no effects of DNE. The data show that DNE alters serotonergic signaling within brainstem circuits that control respiratory frequency but does not functionally alter serotonin signaling in the phrenic motoneuron pool.

In this paper, we propose a new random forest method based on completely randomized splitting rules with an acceptance–rejection criterion for quality control. We show how the proposed acceptance–rejection (AR) algorithm can outperform the standard random forest algorithm (RF) and some of its variants including extremely randomized (ER) trees and smooth sigmoid surrogate (SSS) trees. Twenty datasets were analyzed to compare prediction performance and a simulated dataset was used to assess variable selection bias. In terms of prediction accuracy for classification problems, the proposed AR algorithm performed the best, with ER being the second best. For regression problems, RF and SSS performed the best, followed by AR, and then ER at the last. However, each algorithm was most accurate for at least one study. We investigate scenarios where the AR algorithm can yield better predictive performance. In terms of variable importance, both RF and SSS demonstrated selection bias in favor of variables with many possible splits, while both ER and AR largely removed this bias.

Background: When performing a medial patellofemoral ligament (MPFL) reconstruction, surgeons may place the MPFL graft under higher than anatomic tension to minimize the chance of recurrent instability. Purpose: To investigate whether a lateral retinacular release (LRR) significantly decreases patellofemoral contact pressures after an overtensioned (OT) MPFL reconstruction. Study Design: Controlled laboratory study. Methods: Mean and peak pressure across the patellofemoral joint at 30°, 45°, and 60° of flexion was assessed in 14 cadaveric knee specimens with intact MPFL, transected MPFL, reconstructed MPFL with graft OT, and OT MPFL with LRR. The Wilcoxon signed rank test was used to determine differences across states, with W and C values calculated when possible. Results: Mean pressure decreased significantly after MPFL transection compared with intact at 30° (456.9 ± 116.8 vs 410.9 ± 109.4 N, P = .006, W < 7) and 45° (404.9 ± 91.7 vs 369.4 ± 85.3 N, P = .005, W < 5) and increased significantly from intact to OT graft at 30° (456.9 ± 116.8 vs 563.0 ± 11.2 N, P = .003, W < 7), 45° (404.9 ± 91.7 vs 481.4 ± 14.8 N, P = .005, W < 5), and 60° (272.9 ± 139.0 vs 367.0 ± 53.7 N, P = .007, W < 3). Peak pressure increased significantly between intact and OT graft at 30° (1364.0 ± 478.2 vs 2094.4 ± 619.8 N, P = .002, W < 9), 45° (1224.7 ± 491.5 vs 1676.7 ± 779.1 N, P = .005, W < 5), and 60° (1117.7 ± 566.8 vs 1604.2 ± 772.9 N, W < 3). In knees with significantly increased mean pressure after overtensioning, mean pressure increased by 23.3% (11/14 knees) at 30°, 18.3% (10/14 knees) at 45°, and 35.0% (10/14 knees) at 60°. Peak pressure increased significantly by 35.3% (30°), 25.2% (45°), and 29.3% (60°). A significant decrease in mean pressure, toward but not to baseline, was observed between the OT and LRR states at 30° (563.0 ± 11.2 vs 501.5 ± 9.3 N, W < 7) and 60° (367.0 ± 53.7 vs 302.0 ± 13.8 N, W < 5) and a decrease in peak pressure at 30° (2094.4 ± 619.8 vs 1886.5 ± 655.3 N; W < 9). Conclusion: LRR led to a statistically significant decrease in pressure across the patellofemoral joint in knees that demonstrated increased contact pressures after an OT MPFL graft. Clinical Relevance: LRR after an MPFL reconstruction in which the MPFL graft has been OT may help reduce patellofemoral contact pressures at the time of surgery.

We tested the hypothesis that nicotine exposure in utero and after birth [developmental nicotine exposure (DNE)] disrupts development of glycinergic synaptic transmission to hypoglossal motoneurons (XIIMNs). Glycinergic spontaneous and miniature inhibitory postsynaptic currents (sIPSC/mIPSC) were recorded from XIIMNs in brain stem slices from 1- to 5-day-old rat pups of either sex, under baseline conditions and following stimulation of nicotinic acetylcholine (ACh) receptors with nicotine (i.e., an acute nicotine challenge). Under baseline conditions, there were no significant effects of DNE on the amplitude or frequency of either sIPSCs or mIPSCs. In addition, DNE did not alter the magnitude of the whole cell current evoked by bath application of glycine, consistent with an absence of change in postsynaptic glycine-mediated conductance. An acute nicotine challenge (bath application of 0.5 μM nicotine) increased sIPSC frequency in the DNE cells, but not control cells. In contrast, nicotine challenge did not change mIPSC frequency in either control or DNE cells. In addition, there were no significant changes in the amplitude of either sIPSCs or mIPSCs in response to nicotine challenge. The increased frequency of sIPSCs in response to an acute nicotine challenge in DNE cells reflects an enhancement of action potential-mediated input from glycinergic interneurons to hypoglossal motoneurons. This could lead to more intense inhibition of hypoglossal motoneurons in response to exogenous nicotine or endogenous ACh. The former would occur with smoking or e-cigarette use while the latter occurs with changes in sleep state and with hypercapnia.NEW & NOTEWORTHY Here we show that perinatal nicotine exposure does not impact baseline glycinergic neurotransmission to hypoglossal motoneurons but enhances glycinergic inputs to hypoglossal motoneurons in response to activation of nicotinic acetylcholine (ACh) receptors with acute nicotine. Given that ACh is the endogenous ligand for nicotinic ACh receptors, the latter reveals a potential mechanism whereby perinatal nicotine exposure alters motor function under conditions where ACh release increases, such as the transition from non-rapid-eye movement to rapid-eye movement sleep, and during hypercapnia.

Assessing heterogeneous treatment effects has become a growing interest in advancing precision medicine. Individualized treatment effects (ITE) play a critical role in such an endeavor. Concerning experimental data collected from randomized trials, we put forward a method, termed random forests of interaction trees (RFIT), for estimating ITE on the basis of interaction trees (Su et al., 2009). To this end, we first propose a smooth sigmoid surrogate (SSS) method, as an alternative to greedy search, to speed up tree construction. RFIT outperforms the traditional `separate regression' approach in estimating ITE. Furthermore, standard errors for the estimated ITE via RFIT can be obtained with the infinitesimal jackknife method. We assess and illustrate the use of RFIT via both simulation and the analysis of data from an acupuncture headache trial., 32 pages, 6 figures

Statistical predictive models play an important role in learning analytics. In this work, we seek to harness the power of predictive modeling methodology for the development of an analytics framewo...

KEY POINTS Critical homeostatic behaviours such as suckling, swallowing and breathing depend on the precise control of tongue muscle activity. Perinatal nicotine exposure has multiple effects on baseline inhibitory GABAergic neurotransmission to hypoglossal motoneurons (XIIMNs), consistent with homeostatic compensations directed at maintaining normal motoneuron output. Developmental nicotine exposure (DNE) alters how GABAergic neurotransmission is modulated by acute activation of nicotinic acetylcholine receptors, which may provide insight into mechanisms by which nicotine exposure alters motor function under conditions that result in increased release of GABA, such as hypoxia, or endogenous acetylcholine, as occurs in the transition from NREM to REM sleep, or in response to exogenous nicotine. ABSTRACT Nicotinic acetylcholine receptor (nAChR) signalling regulates neuronal differentiation and synaptogenesis. Here we test the hypothesis that developmental nicotine exposure (DNE) disrupts the development of GABAergic synaptic transmission to hypoglossal motoneurons (XIIMNs). GABAergic spontaneous and miniature inhibitory postsynaptic currents (sIPSCs/mIPSCs) were recorded from XIIMNs in brainstem slices from control and DNE rat pups of either sex, 1-5 days old, at baseline and following acute stimulation of nAChRs with nicotine. At baseline, sIPSCs were less frequent and smaller in DNE cells (consistent with decreased action potential-mediated GABA release), and mIPSCs were more frequent (consistent with increased vesicular GABA release from presynaptic terminals). Acute nicotine challenge increased sIPSC frequency in both groups, though the increase was greater in DNE cells. Acute nicotine challenge did not change the frequency of mIPSCs in either group, though mIPSC amplitude increased significantly in DNE cells, but not control cells. Stimulation of postsynaptic GABAA receptors with muscimol caused a significantly greater chloride current in DNE cells than in control cells. The increased quantal release of GABA, coupled with the rise in the strength of postsynaptic inhibition may be homeostatic adjustments to the decreased action-potential-mediated input from GABAergic interneurons. However, this will exaggerate synaptic inhibition under conditions where the release of GABA (e.g. hypoxia) or ACh (sleep-wake transitions) is increased. These findings reveal a mechanism that may explain why DNE is associated with deficits in the ability to respond appropriately to chemosensory stimuli or to changes in neuromodulation secondary to changes in central nervous system state.

Estimating the efficacy of different instructional modalities, techniques, and interventions is challenging because teaching style covaries with instructor, and the typical student only takes a course once. We introduce the individualized treatment effect (ITE) from analyses of personalized medicine as a means to quantify individual student performance under different instructional modalities or intervention strategies, despite the fact that each student may experience only one “treatment.” The ITE is presented within an ensemble machine learning approach to evaluate student performance, identify factors indicative of student success, and estimate persistence. A key element is the use of a priori student information from institutional records. The methods are motivated and illustrated by a comparison of online and standard face-to-face offerings of an upper division applied statistics course that is a curriculum bottleneck at San Diego State University. The ITE allows us to characterize students that benefit from either the online or the traditional offerings. We find that students in the online class performed at least as well as the traditional lecture class on a common final exam. We discuss the general implications of this analytics framework for assessing pedagogical innovations and intervention strategies, identifying and characterizing at-risk students, and optimizing the individualized student learning environment.

Monsoon low pressure systems (LPS) are synoptic-scale systems forming over the Indian monsoon trough region, contributing substantially to seasonal mean summer monsoon rainfall there. Many current global climate models (GCMs), including the Met Office Unified Model (MetUM), show deficient rainfall in this region, much of which has previously been attributed to remote systematic biases such as excessive equatorial Indian Ocean (EIO) convection, while also substantially under-representing LPS and associated rainfall as they travel westwards across India. Here the sources and sensitivities of LPS to local, remote and short-timescale forcing are examined, in order to understand the poor representation in GCMs. An LPS tracking method is presented using TRACK feature tracking software for comparison between re-analysis data-sets, MetUM GCM and regional climate model (RCM) simulations. RCM simulations, at similar horizontal resolution to the GCM and forced with re-analysis data at the lateral boundaries, are carried out with different domains to examine the effects of remote biases. The results suggest that remote biases contribute significantly to the poor simulation of LPS in the GCM. As these remote systematic biases are common amongst many current GCMs, it is likely that GCMs are intrinsically capable of representing LPS, even at relatively low resolution. The main problem areas are time-mean excessive EIO convection and poor representation of precursor disturbances transmitted from the Western Pacific. The important contribution of the latter is established using RCM simulations forced by climatological 6-hourly lateral boundary conditions, which also highlight the role of LPS in moving rainfall from steep orography towards Central India.

Supplemental Instruction (SI) is a voluntary, non-remedial, peer-facilitated, course-specific intervention that has been widely demonstrated to increase student success, yet concerns persist regarding the biasing effects of disproportionate participation by already higher-performing students. With a focus on maintaining access for all students, a large, public university in the Western United States used student demographic, performance, and SI participation data to evaluate the intervention’s efficacy while reducing selection bias. This analysis was conducted in the first year of SI implementation within a traditionally high-challenge introductory psychology course. Findings indicate a statistically significant relationship between student participation in SI and increased odds of successful course completion. Furthermore, the application of Coarsened Exact Matching reduced concerns that increased course performance was attributed to an over-representation of higher performing students who elected to attend SI Sessions.

Student success efficacy studies are aimed at assessing instructional practices and learning environments by evaluating the success of and characterizing student subgroups that may benefit from such modalities. We propose an ensemble learning approach to perform these analytics tasks with specific focus on estimating individualized treatment effects (ITE). ITE are a measure from the personalized medicine literature that can, for each student, quantify the impact of the intervention strategy on student performance, even though the given student either did or did not experience this intervention (i.e., is either in the treatment group or in the control group). We illustrate our learning analytics methods in the study of a supplemental instruction component for a large enrollment introductory statistics course recognized as a curriculum bottleneck at San Diego State University. As part of this application, we show how the ensemble estimate of the ITE may be used to assess the pedagogical reform (supplemental instruction), advise students into supplemental instruction at the beginning of the course, and quantify the impact of the supplemental instruction component on at-risk subgroups.