Quantifying body composition is central to monitoring performance and training in athletes, however limited sportspecific anthropometric reference data, assessed and reported in a standardised manner, is available. This study provides anthropometric profiles in elite male athletes from different sports. Elite male athletes (n = 73) from National squads of boxing (n = 10), cricket (n = 21), swimming (n = 23), hockey (n = 10) and eventing (n = 9) were assessed for body mass, height, eight skinfolds (triceps, subscapular, biceps, iliac crest, supraspinal, abdominal, thigh and medial calf), body circumferences (arm, waist, hip, thigh and calf) and muscle circumferences (arm, thigh, calf) using ISAK standardised guidelines. For all athletes, large variability exists for measures of skinfold thickness at each skinfold site. Swimming (64.6 ± 16.1 mm) and boxing (63.5 ± 16.1 mm) were similar for the sum of eight skinfolds (∑8SKF) but swimming had lower ∑8SKF compared to cricket (86.1 ± 21.3 mm; p = .011) and eventing (89.9 ± 30.7 mm; p = .028). Hockey (81.9 ± 26.3 mm) and eventing had the most varied ∑8SKF. Thigh body (p=.006) and muscle circumferences (p = .005) were significantly reduced in boxing compared to hockey. No differences were seen between sports for arm (p = .346; ES = .06) and calf (p = .382; ES = .06) muscle circumferences. The anthropometric profiles for elite athletes from various sports during pre-season training will be a useful resource for sports professionals when monitoring and interpreting body composition data. Large variation exists in anthropometric profiles between the different athletes and different sports, highlighting the necessity to have sport-specific normative ranges available to allow optimal monitoring of individual athletes particularly varying across sports as well as age, training status and position.

Acute pancreatitis is an increasingly frequent cause of hospital admission while the clinical significance of each incident remains high. The presentation of acute pancreatitis to the emergency department as an early declaration of symptomatic cholelithiasis is especially worrisome as it suggests a failing of recognition and/or effective referral of premonitory biliary complaints.

Transformer-based language models benefit from conditioning on contexts of hundreds to thousands of previous tokens. What aspects of these contexts contribute to accurate model prediction? We describe a series of experiments that measure usable information by selectively ablating lexical and structural information in transformer language models trained on English Wikipedia. In both mid-and longrange contexts, we find that several extremely destructive context manipulations-including shuffling word order within sentences and deleting all words other than nouns-remove less than 15% of the usable information. Our results suggest that long contexts, but not their detailed syntactic and propositional content, are important for the low perplexity of current transformer language models. 1

No abstract

No abstract