A note on the interpretation of tree-based regression models.

Tree-based models are a popular tool for predicting a response given a set of explanatory variables when the regression function is characterized by a certain degree of complexity. Sometimes, they are also used to identify important variables and for variable selection. We show that if the generating model contains chains of direct and indirect effects, then the typical variable importance measures suggest selecting as important mainly the background variables, which have a strong indirect effect, disregarding the variables that directly influence the response. This is attributable mainly to the variable choice in the first steps of the algorithm selecting the splitting variable and to the greedy nature of such search. This pitfall could be relevant when using tree-based algorithms for understanding the underlying generating process, for population segmentation and for causal inference.

[Analysis and future scenarios of the SARS-CoV-2 epidemic in Tuscany Region (Central Italy)]. / Analisi e previsioni dell'epidemia da SARS-CoV-2 in Toscana.

OBJECTIVES: about two months after the end of the lockdown imposed for the containment of the SARS-CoV-2 epidemic, the contagion dynamics in the Tuscany Region (Central Italy) have been assessed from the beginning of the emergency to the end of June through a compartmental model, and future medium-long term projections have been produced. DATA AND METHODS: this study used a SIRD model in which the infection reproduction number R0 varied over time, according to a piecewise constant function. The fatality parameter and the time from contagion to infection resolution (death or recovery) were fixed to ensure parameter identifiability, and the model was calibrated on the Covid-19 deaths notified from March 9th to June 30th 2020. The uncertainty around the estimates was quantified through parametric bootstrap. Finally, the resulting model was used to produce medium-long term projections of the epidemic dynamics. RESULTS: the date of the first infection in Tuscany was estimated as February 21st 2020. The value of R0(t) ranged from 7.78 (95%CI 7.55-7.89), at the beginning of the outbreak, to a value very close to 0 between April 27th and May 17th. Finally, R0(t) rose, reaching an average of 0.66 (0.32, 0.88) between May 18th and June 30th. At the epidemic peak, estimated at the beginning of April, the notified infected people circulating in the region were just 22% of those predicted by the model. According to the estimated SIRD, under the hypothetical scenario that R0(t) slightly exceeds 1 from the beginning of October 2020, a new wave of contagion could arise by next spring. CONCLUSIONS: the estimated trend of R0(t) is suggestive of a strong effect of the lockdown in Tuscany and of a mild increase of the contagion potentially attributable to the easing of the containment measures. Medium-long term projections unequivocally indicate that the danger of a new epidemic wave has not been averted.

The Less You Are, the More You Are Helped: Effect of Kinesio Tape on Temporal Coordination.

We investigated whether a relationship exists between the inherent timing skill of subject and the magnitude of effect of KT on timing performance by expanding the examination of the dataset described in our previous work. Healthy subjects, tested with and without KT (NKT), performed sets of repetitive wrist's flexion-extensions (IWFEs) with the dominant limb (DL) and the non-dominant limb (NDL) in a synchronization-continuation task at two inter-onset intervals (IOIs): 550-ms and 800-ms. Standard deviation (SD) of IWFEs was used to measure the unevenness of performance. Different patterns of response to KT were observed based on the participant's inherent precision. In the NDL the effect of KT was found significantly higher in the subgroups of individuals having SDNKT of IWFEs>38 ms (p=0.0024) in the 800-ms IOI or SDNKT of IWFEs>19 ms (p=0.0004) in the 550-ms IOI. In the DL the effect of KT was not influenced by the inherent timing skill. We propose KT to be tested for restoring motor control on subjects experiencing sensorimotor disorders associated with intense repetitive training. Also, our findings suggest that care should be taken when using groups of healthy subjects to test the effect of KT.

When Non-Dominant Is Better than Dominant: Kinesiotape Modulates Asymmetries in Timed Performance during a Synchronization-Continuation Task.

There is a growing consensus regarding the specialization of the non-dominant limb (NDL)/hemisphere system to employ proprioceptive feedback when executing motor actions. In a wide variety of rhythmic tasks the dominant limb (DL) has advantages in speed and timing consistency over the NDL. Recently, we demonstrated that the application of Kinesio® Tex (KT) tape, an elastic therapeutic device used for treating athletic injuries, improves significantly the timing consistency of isochronous wrist's flexion-extensions (IWFEs) of the DL. We argued that the augmented precision of IWFEs is determined by a more efficient motor control during movements due to the extra-proprioceptive effect provided by KT. In this study, we tested the effect of KT on timing precision of IWFEs performed with the DL and the NDL, and we evaluated the efficacy of KT to counteract possible timing precision difference between limbs. Young healthy subjects performed with and without KT (NKT) a synchronization-continuation task in which they first entrained IWFEs to paced auditory stimuli (synchronization phase), and subsequently continued to produce motor responses with the same temporal interval in the absence of the auditory stimulus (continuation phase). Two inter-onset intervals (IOIs) of 550-ms and 800-ms, one within and the other beyond the boundaries of the spontaneous motor tempo, were tested. Kinematics was recorded and temporal parameters were extracted and analyzed. Our results show that limb advantages in performing proficiently rhythmic movements are not side-locked but depend also on speed of movement. The application of KT significantly reduces the timing variability of IWFEs performed at 550-ms IOI. KT not only cancels the disadvantages of the NDL but also makes it even more precise than the DL without KT. The superior sensitivity of the NDL to use the extra-sensory information provided by KT is attributed to a greater competence of the NDL/hemisphere system to rely on sensory input. The findings in this study add a new piece of information to the context of motor timing literature. The performance asymmetries here demonstrated as preferred temporal environments could reflect limb differences in the choice of sensorimotor control strategies for the production of human movement.

Fertility intentions and outcomes: Implementing the Theory of Planned Behavior with graphical models.

This paper studies fertility intentions and their outcomes, analyzing the complete path leading to fertility behavior according to the social psychological model of Theory Planned Behavior (TPB). We move beyond existing research using graphical models to have a precise understanding, and a formal description, of the developmental fertility decision-making process. Our findings yield new results for the Italian case which are empirically robust and theoretically coherent, adding important insights to the effectiveness of the TPB for fertility research. In line with TPB, all intentions' primary antecedents are found to be determinants of the level of fertility intentions, but do not affect fertility outcomes, being pre-filtered by fertility intentions. Nevertheless, in contrast with TPB, background factors are not fully mediated by intentions' primary antecedents, influencing directly fertility intentions and even fertility behaviors.

A little elastic for a better performance: kinesiotaping of the motor effector modulates neural mechanisms for rhythmic movements.

A rhythmic motor performance is brought about by an integration of timing information with movements. Investigations on the millisecond time scale distinguish two forms of time control, event-based timing and emergent timing. While event-based timing asserts the existence of a central internal timekeeper for the control of repetitive movements, the emergent timing perspective claims that timing emerges from dynamic control of nontemporal movements parameters. We have recently demonstrated that the precision of an isochronous performance, defined as performance of repeated movements having a uniform duration, was insensible to auditory stimuli of various characteristics (Bravi et al., 2014). Such finding has led us to investigate whether the application of an elastic therapeutic tape (Kinesio® Tex taping; KTT) used for treating athletic injuries and a variety of physical disorders, is able to reduce the timing variability of repetitive rhythmic movement. Young healthy subjects, tested with and without KTT, have participated in sessions in which sets of repeated isochronous wrist's flexion-extensions (IWFEs) were performed under various auditory conditions and during their recall. Kinematics was recorded and temporal parameters were extracted and analyzed. Our results show that the application of KTT decreases the variability of rhythmic movements by a 2-fold effect: on the one hand KTT provides extra proprioceptive information activating cutaneous mechanoreceptors, on the other KTT biases toward the emergent timing thus modulating the processes for rhythmic movements. Therefore, KTT appears able to render movements less audio dependent by relieving, at least partially, the central structures from time control and making available more resources for an augmented performance.