A methodology to identify critical road sections by means of cyclist's fatigue

Critical road sections (CRS) are part of links of the road network, which have an obvious influence on urban transportation systems. Identifying CRS would contribute to improving efficient traffic management. However, existing studies pay less attention to the influence of traffic temporal dynamism on CRS and the spatial disparity of CRS in different temporal scenarios. We propose a method to identify CRS in urban road network. The new method takes sparse tensor decomposition and reconstruction for the imputation of driving speed that is calculated from trajectory data. Then, empirical mode decomposition is applied to calculate the weighted periodicity for each time series of driving speed. Finally, CRS are determined according to local spatial autocorrelation of the weighted periodicity. Taking the urban area of Xi’an City, China, as a case study, the result show that the new method could effectively achieve the imputation of speed information (R2 >0.67). The weighted periodicity could characterize the temporal dynamism of driving speed with considering the aliasing effect of traffic modes. The CRS reflect the multi-center characteristics of urban transportation systems, and show obvious spatial disparity in holiday and workday. The CRS identified by the proposed method could be applied to improving urban traffic management and maintaining efficient urban transportation.

Recently, short-term traffic prediction under conditions with corrupted or missing data has become a popular topic. Since a road section has predictive power regarding the adjacent roads at a specific location, this paper proposes a novel hybrid convolutional long short-term memory neural network model based on critical road sections (CRS-ConvLSTM NN) to predict the traffic evolution of global networks. The critical road sections that have the most powerful impact on the subnetwork are identified by a spatiotemporal correlation algorithm. Subsequently, the traffic speed of the critical road sections is used as the input to the ConvLSTM to predict the future traffic states of the entire network. The experimental results from a Beijing traffic network indicate that the CRS-ConvLSTM outperforms prevailing deep learning (DL) approaches for cases that consider critical road sections and the results validate the capability and generalizability of the model when predicting with different numbers of critical road sections.

Road intersections ranking for road safety improvement: Comparative analysis of multi-criteria decision making methods

Background : Professional women’s road cycling is growing in popularity and increasing in terms of numbers of participants and competitions, and in 2016 the Women’s World Tour was established. Female World Tour cyclists can cover between 13 000 and 18 000 km in training and competition each year, including up to 65 competition days ( Sanders et al., 2019 ). Interest and participation in cycling in Brazil is growing and the national cycling calendar contains 10 elite female events consisting of 20 days of competition, in addition to numerous regional and state competitions on a more regular basis. Nonetheless, little is known about professional female cycling in Brazil. We aimed to determine the training and competition demands of a professional Brazilian female cycling team throughout a competitive season.  Methods: Five female Brazilian cyclists (age 26 ± 4 y; body weight 53.6 ± 4.2 kg; height 1.64 ± 0.05 m) from the same professional cycling team, including the former Brazilian national time-trial champion and two current members of the Brazilian national road-race team, were monitored throughout their competitive season. A standardised exercise test was performed at three moments throughout the year (March, July and December 2018). The exercise protocol comprised an incremental cycling test until volitional exhaustion ( De Pauw et al., 2013 ) to determine maximal oxygen uptake (VO 2max ) using a breath-by-breath system (Quark, Cosmed, Italy), maximal power output (W max ) and maximal heart rate (HR). Training and competition data were acquired using everyone’s preferred GPS system (Garmin Connect, Strava, Training Peaks) and power data from two athletes was acquired (Garmin Vector, Garmin, USA).  Results: The athletes spent a total of 193 ± 56 days on the bike (range: 104 – 234 days), completing 164 ± 45 days training (range: 89 – 206 days) and 30 ± 16 days competing (range: 15 – 55). The total distance covered over the year was 11124 ± 2895 km, ranging between 7382 and 14698 km (Figure 1). Distance covered during training was higher compared to competition (9037 ± 2027 vs. 2111 ± 1253 km) for all athletes, as was time spent training (334 ± 72 vs. 61 ± 33 h; Figure 1).   Figure 1 . Distance covered by each athlete during training and competition throughout the season.  Mean power output during training was 113 ± 28 W for Athlete 1 and 128 ± 16 W for Athlete 4; peak power output during training was 469 ± 192 W for Athlete 1 and 472 ± 172 W for Athlete 4. Mean power output during competition was 174 ± 30 W for Athlete 1 and 167 ± 12 W for Athlete 4; peak power output during training was 803 ± 168 W for Athlete 1 and 795 ± 91 for Athlete 4.  Laboratorial exercise performance data showed that most athletes maintained their exercise capacity from the start of the season until midway, although the end of season showed a marked decline in these variables for all athletes who returned for testing (Table 1).  Table 1 . Exercise performance data from the incremental cycling test to exhaustion: Maximal oxygen uptake (VO 2max ), maximal power output (W max ) and maximal heart rate (Max HR).    Start Mid End       Athlete 1 VO 2max ml·kg -1 ·min -1 62.3 61.3 52.5  W max W 303 284 260  Max HR beats·min -1 194 193 191             Athlete 2 VO 2max ml·kg -1 ·min -1 53.7 - 44.4  W max W 214 - 184  Max HR beats·min -1 187 - 184       Athlete 3 VO 2max ml·kg -1 ·min -1 51.2 50.3 -  W max W 237 230 -  Max HR beats·min -1 187 183 -       Athlete 4 VO 2max ml·kg -1 ·min -1 54.3 56.0 50.9  W max W 252 255 240  Max HR beats·min -1 186 183 183       Athlete 5 VO 2max ml·kg -1 ·min -1 55.9 56.2 -  W max W 238 227 -  Max HR beats·min -1 187 176 -         Discussion: Our data show that several of these professional Brazilian cyclists covered similar distances to those covered by World Tour level cyclists, one athlete covering up to almost 15 000 km. Over 150 days were spent training while between 15 and 55 days were spent competing in a total of 58 different races over the year across 5 countries (Brazil, Belgium, Chile, Uruguay and Italy). Four of the 5 athletes travelled to Belgium for 10 weeks to compete in international events between April and June and 19 of the 56 races were disputed during this 10-week period. Athletes 1 and 2 travelled to Chile for the Pan American Track Cycling Championships in May. Athlete 2, who did not travel to Belgium, fractured her wrist prior to mid-season testing and did not take part in these evaluations. Two of the athletes were called up to represent Brazil at the 2018 UCI Road World Championships in the Women Elite Individual Time Trial (Athlete 1) and the Elite Road Race (Athlete 4). Unfortunately, Athlete 1 sustained a crash in early August, ruling her out of the World Championships and she also sustained a fall shortly following her return to training in September, fracturing her collarbone requiring surgery. Athlete number 3 became pregnant towards the end of the season and did not complete any end of season analyses. This highlight the complex nature of professional female cycling which involves substantial time training, competing and travelling, as well as significant risks of injury.  All athletes had high VO 2max values at the start of the season, classifying them as trained, well-trained or professional according to the criteria of Decroix et al. ( 2016 ). Mid-season testing revealed that the cyclists continued to maintain this high cardiorespiratory capacity, with everyone maintain values within 3%. However, end of season laboratory test performance was worse for all returning athletes, with reductions of between 6 and 17%. This was also reflected in the W max attained during the same incremental test. The explanation for the considerable decline in laboratorial test performance is unclear although we could speculate that it may be due to a plethora of factors including accumulated fatigue over the season and injuries. They were not associated with clinical parameters (data not shown). Monitoring fatigue and its foremost causes over a cycling season would allow greater understanding of the demands placed upon this athlete population and may also provide insight into optimising preparation (i.e. training, nutrition, schedule) for peak competitive performance.  Conclusion: These professional Brazilian female cyclists had training and competition schedules similar to female World Tour cyclists, competing in numerous national and international competitions. A reduced exercise capacity, as measured by laboratorial tests, at the end of the season is perhaps indicative of a gruelling year-long schedule although further research is warranted to assess the various demands on professional female cyclists throughout the season. Â

Background: The national cycling calendar in Brazil is extensive with numerous regional and state competitions on a weekly basis. Furthermore, several teams now compete in national, continental and international races throughout the calendar year. In light of Brazil’s increasing interest and progression in the world of cycling, we aimed to determine the characteristics of well-trained Brazilian state cyclists also competing at national and international level.  Methods: Thirty-eight well-trained Brazilian cyclists (six professionals and thirty-two amateurs) competing in regional (N = 38), national (N = 14) and international (N = 7) competition attended one laboratory session for the determination of the following characteristics: age, height, body mass, maximal oxygen uptake (VO2max), anaerobic threshold (ANT) and maximal power output (Wmax). Furthermore, individuals were also required to complete a questionnaire relating to their cycling frequency and history (ie hours cycled per week; distance cycled per week; years of experience; level of competition). Individual VO2max, ANT and Wmax was determined during a cycling capacity test to exhaustion on a cycle ergometer (Lode Excalibur, Germany). The test consisted of four submaximal 4-min stages starting at 75 W, increasing by 50 W each stage until 225 W. Thereafter, workload was increased by 30 W every minute until exhaustion. Ventilatory and gas exchange measurements were recorded using a portable breath-by-breath system (K4 b2, Cosmed, Italy); the highest value averaged over a 30-s period during the test was defined as VO2max. The last completed stage and the fraction of time spent in the final non-completed stage multiplied by 30 W was defined as an individual’s Wmax. Individual ANT was determined using the method described by Okano et al. (2006).  Discussion: Well-trained Brazilian male cyclists showed a wide range of values for physical and physiological characteristics. Interestingly, VO2max was well below those reported by several studies employing well-trained cyclists (>60 ml·kg·min-1; Jeukendrup et al., 2008; Bellinger and Minahan, 2014 & 2015), while the upper-most value shown in our cyclists is equivalent to the lower-end value in professional road cyclists (Mujika and Padilla, 2001). In addition, Wmax and ANT was also below those shown in reportedly well-trained (Chung et al., 2014; Jeukendrup et al., 2008) and professional (Mujika and Padilla, 2001) cyclists. These results are somewhat surprising considering the trained nature of our participants. All athletes included in this study were of a competitive level, several having secured podium finishes in national and international competition over the previous year, while weekly training loads were similar to, or higher than, those previously reported with well-trained cyclists (Chung et al., 2014; Bellinger et al., 2015). Although several individuals were only competing at state level, there were no differences in physiological measures between them and those competing at national level or higher (data not shown). A significant contributing factor may have been the age of the current individuals, since the average age of a professional road team is 26 years (Mujika and Padilla, 2001). Despite this, the highest individual VO2max in our cohort was achieved by an individual 41 years of age. Further investigation into the physiological characteristics of trained Brazilian cyclists is warranted.  Conclusion: These data suggest that well-trained Brazilian cyclists may be at a lower standard than their international counterparts. Nonetheless, future research should determine the physiological characteristics of Brazilian cyclists using a larger professional cohort. Â

Background: The demands of road cycling comprise rapid and frequent variations in power output. Recently, our lab developed a 60 min variable power cycling performance test (VCT) and established its reliability for measuring power output in trained male cyclists (Sharma et al, 2014: Reliability and validity of a new variable power performance test in road cyclists, 6th Exercise and Sport Science Australia Conference, Adelaide, Australia). The VCT was modelled on data from elite men’s road racing (Ebert et al., 2006: International Journal of Sports Physiology and Performance, 1, p. 324-335) and included substantial variation in power output. Purpose: The purpose of this investigation was to determine the differences in power output between national and club level cyclists during a new test of variable cycling performance (VCT). Methods: Nine national (N) level (mean ± s; age: 24 ± 5.2 years; mass: 71 ± 8.5 kg; peak power output: 5.3 ± 0.4 W/kg) and fourteen club (C) level cyclists (mean ± s; age: 33 ± 6.7 years; mass: 79 ± 5.5 kg; peak power output 4.6 ± 0.4 W/kg) each completed an incremental exercise test (for determination of peak power output) and a VCT (plus familiarisation). The VCT consisted of 10 x 6 min “laps”, with each lap consisting of self-paced periods where the participant were instructed to cycle at their perceived “recovery”, “hard” or “sprint” pace (figure 1). Power output was monitored continuously during the VCT. Results: Mean power output (W/kg) for the VCT was significantly greater in N than C (mean ± s; 3.53 ± 0.24 vs. 3.18 ± 0.17; p = 0.001). Power output at each lap (figure 2) was also significantly higher in SN (p < 0.01). Relative mean power (% PPO) for sprint efforts was greater in N (mean ± s; 185 ± 29% vs. 171 ± 29%; mean difference; 14.3%, 95% CI [12.9, 15.7]) and also for “hard” efforts (mean ± s; 143 ± 13% vs. 132 ± 19%; mean difference 10.8%, 95% CI [9.7, 11.9]). Discussion: National level cyclists were able to sustain a greater overall power output throughout the trial, as well as during sprints and hard efforts. Interestingly, the national cyclists adopted a parabolic pacing strategy, compared to the largely even pacing of the club cyclists (figure 2). Whilst both groups were able to increase power output in the latter laps, the national cyclists did so by a greater amount (figure 2).  Furthermore, during the sprint and “hard” effort sections of the VCT, the national cyclists were consistently higher in power output, which may suggest an improved lactate tolerance (Coleman, 2013. In Hopker and Jobson (Eds.), Performance cycling: the science of success, 13-32. London: Bloomsbury). Thus, muscle acid/base properties associated with fatigue resistance may be highly relevant to performance during variable power cycling, including muscle buffering capacity (Chicharro et al., 2000: British Journal of Sports Medicine, 34, p. 450-455) and clearance of lactate and other metabolites (Coleman, 2013). Conclusion: These findings suggest the VCT is a valid test of stochastic performance in trained cyclists.

Introduction Cycling performance depends on several physiological and biomechanical parameters (Faria et al. 2005). The influence of biomechanical factors such as pedalling technique is still an issue of debate (Leirdal and Ettema 2011). Several studies demonstrated that pedal force effectiveness (FE, %, ratio of the force perpendicular to the crank and the total force applied to the pedal) has also been used as a gold standard measure of pedalling technique in cycling. However, FE depends on several constraints such as power output (PO, W), pedalling cadence, body position, fatigue, and cycling experience (Bini et al. 2013). Studies showed that large increase in PO (i.e. from 60% to 98% of the maximal aerobic power [MAP]) led to higher FE. Additionally, a recent study showed that professional cyclists have better pedalling technique than elite or club cyclists (Garcia-Lopez et al. 2016). The purpose of this study was to assess the influence of PO on pedalling technique in cyclists of different competitive levels using FE as a performance clue.  Methods 37 male road cyclists of different competitive levels (elite [19] and professional [18]) performed all testing sessions on a Bikefitting ergometer (Shimano, Pedal Analyzer, Dynamics Lab, Sittard, Netherland) that has been used to assess FE in seated position. Firstly, the personal bike position of each cyclist was reported on the ergometer. Then, the cyclists were required to perform exercises at four level of PO (55, 70, 85 and 100% of MAP). In order to keep an individual combination between muscular force and pedalling cadence, the cyclists were asked to keep their preferred pedalling cadence during each level of PO. The main parameter measured was FE whereas the balance between propulsive and resistive forces (%) was also measured as a secondary parameter. A two-way ANOVA was used to analyse the influence of both the PO and the competitive level on FE.  Results The figure 1 shows an increase in FE with PO (+26.4% from 55 to 100% of MAP, p < 0.001). Even if the competitive level did not influence FE, the increase of FE according to PO was higher (+36.1%) in professional cyclists than elite cyclists (+19.1%). Additionally, the coefficient of variation (CV, %) decreased with PO (CV55% = 16.2 %, CV70% = 15.0 %, CV85% =  12.8 % and CV100% = 11.0%). For all PO and competitive levels, FE was correlated with both PO (r = 0.47, p < 0.001) and the balance between propulsive and resistive forces (r = 0.82, p  < 0.001). Finally, the resistive forces were significantly (p < 0.001) decreased from 15.8 to 5% between 55 and 100% of MAP during the upstroke.  Conclusions The main findings of this study show that FE was influenced by the level of PO and was independent of the competitive level. Yet, professional cyclists increased faster FE with the level of PO suggesting that their pedalling technique was further improved in high levels of PO as compared with elite cyclists. Previous studies showed that an increase in PO induced an increase in FE (Zameziati et al. 2006). In this study, for a given PO, the increase in FE was due to a lower resistive force. Concerning the effect of competitive level, previous studies have shown no difference in FE between cyclists of different competitive levels (Sanderson et al. 2000) whereas another study (Garcia-Lopez et al. 2016) demonstrated that professional cyclists had better pedalling technique than elite cyclists. Finally, it appears that with the increase of PO, the professional cyclists improve their pedalling efficiency both by increasing FE and decreasing the resistive force during the upstroke of the pedalling cycle.

Aim: Extensive research has been directed towards the identification of the threshold that demarcates fatiguing from non-fatiguing exercise during an incremental workout on a cycle ergometer. As the exercise intensity increases, fatigue compromises not only the cardiovascular and respiratory systems, but also the neuromuscular system. The objective of this work was to calculate the neuromuscular fatigue threshold from surface electromyographic amplitude (EMG-FT) and determine possible associations between this threshold and the metabolic (onset of blood lactate accumulation, OBLA) and ventilatory (ventilatory threshold, VT, and respiratory compensation point, RCP) thresholds.  Methods: Sixteen cyclists performed incremental cycle ergometer rides to exhaustion with bipolar surface sEMG signals recorded from the vastus lateralis. The participants were road cyclists engaged in regular training and amateur road races. On average, cyclists trained at least four times a week covering a weekly distance ranging between 400 and 600 km, plus competition or Sunday training. Cyclists had a national competitive experience of 4.3 (1.7) years and had performed an average of 20,000 km riding. Before the incremental exercise test started, cyclists performed 5 min of unloaded cycling. The test was initiated at a workload of 125 W and the load was increased by 25 W every 1 min until the subjects could no longer continue to exercise (Fig. 1a). At this time, the power achieved was referred to as the maximal power. To calculate the EMG-FT, the deVries’ model was employed using 1-min exercise periods. Blood samples were taken every two minutes during the test. OBLA was calculated as the power output corresponding to a blood lactate concentration of 4.0 mmol/L. During the incremental exercise, breath by breath analysis was performed using a turbine flow-meter connected to a face mask and data of oxygen uptake and carbon dioxide production and ventilatory parameters. One-way repeated-measures ANOVA was used to determine whether there were significant differences in average power output among the EMG-FT, OBLA, VT, and RCP. Pearson correlation coefficients were calculated to determine the relationships among EMG-FT, OBLA, VT, and RCP.  Results: The maximal power tolerated was 403.8 ± 36.5 W. The power output at EMG-FT (372 ± 25 W) was comparable (P>0.05) to that of RCP (381 ± 36 W), but significantly higher (P 0.05), but it was positively correlated with those of VT (0.78, P<0.001) and RCP (0.85, P<0.001).  Conclusions: Our results do not support the existence of a relationship between EMG-FT and lactate concentration. A significant correlation was found between the sEMG-based threshold and gas exchange thresholds, although this correlation may not be causative. Second, application of deVries’ model using of 1-min exercise periods could result in overestimation of the fatigue threshold since such short exercise periods could be insufficient to detect EMG signs of neuromuscular fatigue

Background: Critical power (CP) represents the highest metabolic rate at which oxygen uptake and blood lactate stabilizes during exercise and is strongly associated with endurance exercise performance. Determination of CP typically requires the completion of two to five constant power tests to exhaustion (TTE). An alternative test involves measurement of mean power during the final 30 s of a single 3-minute all-out cycling test, and this test agrees well with CP (Vanhatalo et al.,2007: Medicine and Science in Sports and Exercise, 39, 548-555). However, the 3-minute all-out test requires a preliminary Ramp test to determine the ergometer settings for the subsequent CP assessment. Purpose: To quantify the reliability of a single contiguous Ramp and 3-minute All-Out test (RAO), and to appraise validity relative to CP determined from serial TTE tests. Methods: All participants provided written informed consent prior to participation, tests were separated by 2 – 7 d, and studies by 6 – 12 months. Reliability study: Four male triathletes, six male and one female road cyclist ( 2peak 4.32 ± 0.70 L•min-1) completed two RAO (RAO 1 & 2) cycle ergometer tests (Lode Excalibur Sport, Groningen, The Netherlands). The RAO was a contiguous 5-min warm up at 50 W, a ramp-incremental (30 W•min-1) to exhaustion, and a 3-min all-out exercise test. Exhaustion was defined as a cadence ≤ 50 rpm. The ergometer resistance during the AO phase was defined prior to exercise using: (body mass (kg) x 3.5) / 852. Cadence and duration was withheld during the AO phase.  Validity study: Five triathletes and two road cyclists ( 2peak 3.96 ± 0.53 L•min-1) completed the RAO before and after a series of at least four constant power TTE designed to elicit exhaustion between ~ 3-15 min. For TTE tests the asymptote (CP) and curvature constant (W´) of the power-duration relationship were quantified using 2-parameter nonlinear regression models. For RAO tests the estimated CP (ECP) was calculated as the mean power output over the final 30 s of the RAO, and estimated W´ (EW´) was calculated as the power-time integral above ECP during the RAO. Statistics: Data are reported as mean, standard deviation. Data was appraised using a repeated measures general linear model for quantification of bias and typical error (TE). The relationship between protocols was quantified using coefficient of variation (CV). Results: Reliability study: ECP in RAO1 (289, 38 W) and RAO2 (292, 41 W) were not different (P = 0.24, TE 7 W, mean bias of 4 ± 19 W). ECP between repeats were very strongly related (CV 2%).  EW´ in RAO1 (12.6, 3.4 kJ) and RAO2 (12.0, 4.2 kJ) were not different (P = 0.49, TE 2.0 kJ, mean bias of -0.6 ± 5.5 kJ), and were related (CV 13%). Validity study: Durations of TTE ranged from 170 (34) to 861 (202) s. A 5 W (1.8%) standard error was associated with the estimate of CP using the 2-parameter model. ECP (280, 36 W) was not different to CP (279, 34 W; P = 0.95, TE 10 W). Comparisons revealed a mean bias of 1 ± 27 W, (CV 3%).  EW´ (17.1, 3.4 kJ) was not different to W´ (15.8, 2.8 kJ; P = 0.461, TE 1.8 kJ, mean bias of -1.35 ± 8.9 kJ), EW´ and W´ were related (CV 14%). Discussion: The RAO test parameters were sufficiently reliable for use in future studies. Similar to previous studies investigating alternative methods for power-duration curve estimation, the reproducibility of ECP was small (CV 2%), but EW´ was wider (13%). The mean difference between parameter estimates of CP and W´ using the traditional TTE test and single RAO test was negligible. The limits of agreement between the CP estimate derived from a series of maximal constant power TTE tests and the single test RAO protocol were moderately wide (27 W) but similar to previous reports of AO-exercise strategies for CP estimation. Conclusion: The CP and W´ estimates from a single test were similar to traditional testing methods that require multiple maximal exercise tests. Therefore, the RAO test may simplify CP estimation in competitive cyclists.

Cyclists are vulnerable road users and the most severe injury outcomes for on-road cyclists are from collisions involving a motor vehicle. Research undertaken in this thesis aimed to identify contributing factors in unsafe cyclist-driver events to inform efforts to reduce the incidence of cyclist-driver crashes and cyclist injury severity outcomes. The research was conducted in three stages, primarily in Melbourne, Victoria, Australia and is presented as a thesis by publication. The Safe System Framework was used as the theoretical model for the research and the research stages included i) an observational study using a covertly positioned video camera at signalised intersections across metropolitan Melbourne; ii) a naturalistic cycling study using a compact video camera attached to commuter cyclists’ helmets which recorded their trips to and from work; and, iii) a national online survey of drivers and cyclists of their cycling-related behaviours, knowledge and attitudes. The role of driver behaviour in cyclist-driver crashes and near-crash events was identified and was the most significant finding of this doctoral research. In-depth analysis of near-collision events revealed that drivers’ behaviour immediately prior to an event contributed to the majority of unsafe interactions between cyclists and drivers. The most frequent driver behaviour associated with near-collision events was turning left across a cyclists’ path. Three important components of this behaviour were: indicating (signalling) before turning, driver head checks before turning left and clearance distance when overtaking cyclists. These three behavioural components were investigated further, with a particular focus on the influence of cycling-related knowledge and attitudes. Findings supported the concept of safety in numbers which proposes a positive association between cycling participation and cyclist safety. A significant finding was that drivers who were also cyclists (driver-cyclists) were more likely than drivers who were not cyclists to report safe driving behaviours. Driver-cyclists also reported more positive attitudes towards cyclists and good knowledge of road rules for cycling facilities. Cyclist behaviour had also been identified as a potential crash risk factor, particularly red light running behaviour. Encouragingly, however, only a small proportion of observed cyclist infringed and predictive factors included direction of travel (left turn) and gender (male). The presence of other road users (cross traffic and in the same direction) had a deterrent effect. Last, the presence of cycling facilities was associated with cyclist-driver interactions. Cyclist and driver behaviour at two cycling facilities at intersections (bike boxes and continuous bike lane) was measured. Despite the high level of knowledge of bike boxes, many drivers were non-compliant at this type of facility. In contrast, both cyclists and drivers were more likely to be compliant at the facility that provided a continuous parallel bike lane compared with the bike box facility. Findings of this research provide new insights into the influence of behavioural factors and presence of cycling facilities on cyclist safety. Greater cyclist-related driver education and training are essential to improve cyclist safety. It is anticipated that the findings from this research will inform programs and initiatives that will improve the safety of on-road cyclists.

Identifying critical road sections that require prompt attention is essential for road agencies to prioritize monitoring, maintenance, and rehabilitation efforts and improve overall road conditions and safety. This study suggests a decision matrix with a hierarchical structure that factors in the pavement deterioration rate, infrastructure safety, and crash history to identify these sections. A Markov mixed hazard model was used to assess each section’s deterioration rate. The safety of the road sections was rated with the International Road Assessment Program star rating protocol considering all road users. Early detection of sections with fast deterioration and poor safety conditions allows for preventive measures to be taken and to reduce further deterioration and traffic crashes. Additionally, including crash history data in the decision matrix helps to understand the possible causes of a crash and is useful in developing safety policies. The proposed method is demonstrated using data from 4725 road sections, each 100 m, in Addis Ababa, Ethiopia. The case study results show that the proposed decision matrix can effectively identify critical road sections which need close attention and immediate action. As a result, the proposed method can assist road agencies in prioritizing inspections, maintenance, and rehabilitation decisions and effectively allocate budgets and resources.

Introduction: Bioelectrical impedance vector analysis (BIVA) and the phase-angle (PA), derived from bioelectrical impedance raw values (i.e., resistance, R, and reactance, Xc) allow qualitative and descriptive assessments of body composition and hydration status, independent of prediction equations, body weight and body composition models [3,4,5]. Bioelectrical impedance reference values for the healthy normal population, soccer players [6] and several clinical settings are well established [1,2], but are lacking for male cyclists. Therefore, the aim of the present study was to obtain reference bioelectrical impedance data characterizing professional, elite, junior, and amateur male road cyclists.  Methods: The study included 102 male professional riders (age: 25.6±4.7 yr, height: 178.3±5.9 cm, weight: 68.5±5.9 kg), 225 male amateur riders (age: 39±2 yr, height: 176.1±6.4 cm, weight: 71.1±9.5 kg), 46 male junior cyclists (age: 16.9±1.2 yr, height 176.60±5.94 cm, weight 65.62±7.27 kg) and 69 male elite road cyclists (age: 21±2.9 yr, height: 178.2±5.7 cm, weight: 69.03±7.6 kg). The controls were represented by normal values of healthy male Italian population (age: 48±17 yr, height: 170.4±8 cm, weight: 72.6±11.5 kg) [2]. Professional cyclists were classified into 3 groups according to their team role (sprinter, n=16; climber, n=35; all-rounder, n=51), whereas the juniors, amateurs and elite cyclists were all considered all-rounder. Whole-body impedance measurements (BIA- 101 Anniversary AKERN/RJL-Systems) were performed during the peak performance period of the cyclists. All measurements were obtained in compliance with the manufacturer’s guidelines and analyzed according to the BIVA method [1]. BIVA, compared to the estimated fat and fat-free masses with conventional BIA, does not use regression prediction equations, and was shown to adequately display differences in hydration and soft tissue mass in healthy people and athletes [7]. Furthermore, BIVA allows establishing population- specific norms (50, 75 and 95% tolerance ellipses).  Results: Compared to the amateur cyclists and the normal population, the group vector and the tolerance ellipses of the professional cyclists was displaced to the upper left (p<0.001) as well as the comparison of the amateurs and juniors (p<0.001), amateurs and elite cyclists (p<0.001) (fig. 1). Also, all categories showed a shift to the upper left in comparison of controls. Comparisons of professional cyclists to amateurs and elite cyclists to amateurs showed a higher phase-angle (7.1°±0.1 vs 6.5°±0.8 (p<0.001) and 7.0°±0.7 vs 6.5°±0 (p<0.001), respectively). Significant differences in vector position were found between sprinters and climbers (p<0.05) and between allrounders and climbers (p<0.02) (fig. 2).  Conclusion: The main finding of the present study is that road cyclists in general and specialists have specific bioimpedance values compared to non-cyclists. Muscle mass and function, as indicated by the left shifted vector and the phase-angle, increased with increasing performance level. The specific tolerance ellipses of the professionals might be used for classifying individual vectors of professional cyclists and to define target regions for lower level cyclists.

Introduction: It is recognized that vitamin D has multiple effects in health (Autier, Boniol et al. 2014) and in athletic performance (Ogan and Pritchett 2013). It has been shown that it may have a role in the immune function, protein synthesis, inflammatory responses, regulation of cell growth and skeletal muscle physiology. However, vitamin D deficiency is common either in general population (Holick and Chen 2008) as in athletes (Farrokhyar, Tabasinejad et al. 2014). The endogenous synthesis via ultraviolet-B radiation through exposure to sunlight is the major source of vitamin D in humans. The 25(OH) Vitamin D is used as the clinical measure of vitamin D status (Willis, Peterson et al. 2008). However optimal concentration thresholds are not yet well determined. Purpose to assess the vitamin D status of elite Portuguese cycling athletes. Methods: This study was conducted with a group of 65 elite Portuguese cyclists of different ages and modalities (road cycling, mountain bike - cross-country and mountain bike - downhill), selected from the Portuguese national team. All athletes were resident in mainland Portugal at latitudes 37-42°N. Over five months, from April to August 2014, peripheral blood samples were collected to assess serum levels of 25 (OH) vitamin D in the fasted state. Levels were classified in: deficit ( 32 ng/ml). Data were collected and analyzed in SPSS ® version 20. Results: The sample had a mean age of 21 ± 5,1 years. Most athletes were male (73,8%) and road cycling practitioners (56,9%). We found 4 athletes with 25 (OH) vitamin D deficits, 3 of them perform mountain bike - cross- country and 1 mountain bike – downhill (table 1). 25 (OH) vitamin D mean value was 29,2ng/ml ± 7,1. As shown in Fig.1 only 36,9% of the cyclists had a sufficient value of 25 (OH) vitamin D. The athletes blood samples were preferentially collected in April (33,8%) and June (24,6%). August was the month with fewer athletes evaluated (6,2%). We found no difference between the mean value of 25 (OH) vitamin D between genders [t(63)=0,853; p =,398]. There was a weak positive correlation between age and 25 (OH) vitamin D levels, however with no statistical significance (r=0,107; p =0,395). We found a statistical significant positive correlation between collection month and 25 (OH) Vitamin D levels (r=0,426; p <0,001). Discussion: The cyclists who participated in this study trained outdoor, however the results show that the vitamin D status is far from being the most suitable. We have to realize that, despite the sunny days in winter months, the cyclists wear more clothes to protect every part of the body from cold during the outdoor training session. So, the ultraviolet-B radiation will not be so effective during this period of time. As already mentioned (Cannell, Hollis et al. 2009), the values of vitamin D rise gradually by March till September, then the values starts to decrease again. If we consider the cut-off used in a recent meta-analysis (Farrokhyar, Tabasinejad et al. 2014) of vitamin D inadequacy in athletes we could say that 63,1% of our cyclists had a value of vitamin D inadequate, that is, almost two in each three cyclists. Considering other studies made with cyclists, our results are very similar. A French study (Guillaume, Chappard et al. 2012) with twenty nine professional road cyclists and a mean of 26,5 years, from a team who participated in the Tour de France, showed a mean value 29,8ng/ml ± 11,0 of 25(OH) Vitamin D . The blood samples were collected in different periods of the season. Similar results were obtained (Lombardi, Corsetti et al. 2014) from nine professional road cyclist of an Italian team during the Giro d’Italia 2011 in three periods of time of this competition between May 08 and May 29. The mean results were between 28,3ng/ml ± 5,1 and 33,7ng/ml ± 5,0 More recently, the results from a sample of twenty Spanish cyclists (ValtueA±a, Dominguez et al. 2014) indicate a mean value of 25(OH) Vitamin D much lower. The mean value is 20,9ng/ml ± 7,3. However, the date of analysis is not mentioned. In our sample we had three different types of cycling modalities. Unfortunately, it was not possible to compare the mean value of 25(OH) Vitamin D between the different modalities due to the small sample (mountain bike - downhill) and to the different collection date (road cycling mostly in June/July and mountain bike only in April and May). Conclusions: Physicians and sports nutritionists working with cyclists should consider to analyze serum 25 (OH) Vitamin D concentrations in winter and spring months. Supplementation with vitamin D is recommended in cyclists with low value of 25(OH) Vitamin D.

Short-term traffic prediction under corrupted or missing data for large-scale transportation networks has become an important and challenging topic in recent decades. Since the critical roads have predictive power on their adjacent roads, this paper proposes a novel hybrid short-term traffic state prediction method based on critical road selection optimization. First, the utility function of the quality of service (QoS) for the critical roads in a large-scale road network is proposed based on the coverage and the data score. Then, the critical road selection optimization model in the transportation networks is presented by selecting an appropriate set of critical roads with the maximum proportion of the total calculation resources to maximize the utility value of the QoS. Also, an innovative critical road selection method is introduced, which is considering the topological structure and the mobility of the urban road network. Subsequently, the traffic speed of the critical roads is regarded as the input of the convolutional long short-term memory neural network to predict the future traffic states of the entire network. Experiment results on the Beijing traffic network indicate that the proposed method outperforms prevailing DL approaches in the case of considering critical road sections.

Introduction Within the scientific literature there is little evidence available to provide practitioners with information on strength and power profiles of cyclists, resulting in a limited understanding of neuromuscular factors related to cycling performance. Information on the legs’ elastic energy utilisation, force-velocity and length-tension curves can inform training programs and aid in talent identification. Other sports where such information is more widely available have already successfully implemented this within preparation programmes (e.g. McBride et al., 1999). Methods A total of 44 cyclists were recruited for this project, of which 15 classified in a Novice category by having no racing experience at all (age 35.5 ± 11.4 yrs; height 177.4 ± 6.5 cm; mass 77.4 ± 9.3 kg; FTP 3.28 ± 0.47 W/kg), 14 in the Road racing category as they competed for at least the past year at British Cycling Category 2 level or higher and no experience in Time-Trial (TT) races (age 35.9 ± 12.7 yrs; height 179.1 ± 6.5 cm; mass 76.6 ± 9.0 kg; FTP 3.88 ± 0.49 W/kg), and 15 in the TT category as they considered TT racing as their main competitive aim, rode in dedicated TT positions and had recently produced a 10 or 25 mile personal best of 1) for all groups. This dominance was significantly less prominent in the TT group for flexion conditions (1.35 ± 0.18) compared to Road (1.56 ± 0.22; p = 0.031) and Novice (1.53 ± 0.19; p = 0.004) groups. Joint flexion torques showed non-significant trends; they were slightly higher in the knee and lower for the hip (1.43 & 2.08 Nm/kg respectively) in TT athletes compared to Road (1.35 & 2.14 Nm/kg) and Novices (1.36 & 2.22 Nm/kg) (p = 0.429 & 0.189). No differences were found for the angle at which peak torque occurred. The velocity effect on torque production was comparable between the groups. It decreased from its peak at 30 °/s, to 82 ± 11 % of that when tested at 270 °/s for knee flexion and to 61 ± 9 % for knee extension. Hip torque reduced to 66 ± 10 % and 79 ± 10 % for flexion and extension respectively, when tested at 210 °/s compared to 30 °/s condition. Discussion The CMJ data show that cyclists – both novice and competitive – perform poorly on vertical jumping (29 ± 6 cm) compared to strength trained (48.2 ± 2.8 cm) and even untrained individuals (33.7 ± 2.3 cm) (McBride et al., 1999). This is in line with previous research on endurance type athletes showing long-distance runners to perform inferiorly on jump tasks compared to an untrained population (27.8 ± 4.3 cm vs 37.3 ± 3.1 cm; Kubo et al., 2000). In contrast to the findings by Kubo et al. (2000), the tested competitive cyclists showed lower SJ/CMJ ratios compared to the untrained controls indicating a relatively large utilisation of elastic energy storage compared to muscular power in jump performance. Based on the dynamometry testing, it seems most plausible to suggest that the reduced hip flexion capacity in TT riders results from these muscles being disused during cycling due to the extreme hip flexion angles common in their riding positions. It could be suggested that an attempt is made to compensate for this loss in hip flexion capacity through increased knee flexors’ strength. An increased knee flexor torque in TT riders could also indicate a mechanically more effective pedalling technique on the bike, as previous literature has linked hamstring activity with increases in Index of Force Effectiveness on the bike (Bini et al., 2013). Greater separation between tested groups might have been masked due to variation in preferred bike setup within the groups, TT riders also training in road setups and novice cyclists having undergone minor adaptations through recreational cycling activities. Based on these results, it seems appropriate to advise strength training to be tailored to the type of competition a cyclist is aiming to perform on. TT riders should focus on knee flexor strength, while road cyclists could benefit from a more balanced approach between hip and knee strength. Currently ongoing research is investigating how these strength characteristics relate to determinants of cycling performance in order to further help optimising training protocols and talent identification strategies. References Bini RR, Hume P, Croft J, Kilding A. (2013) J Sci Cycl, 2(1), 11-24. Kubo K, Kanehisa H, Kawakami Y, Fukunaga T. (2000). EJAP, 81(3), 181-187. McBride JM, Triplett-McBride T, Davie H, Newton RU. (1999). J Biom, 32(10), 1021-1026. Cuk I, Markovic M, Nedeljkovic A, Ugarkovic D, Kukolj M, Jaric S. (2014). EJAP, 114(8), 1703-1714.