TacticAI: an AI assistant for football tactics.

Identifying key patterns of tactics implemented by rival teams, and developing effective responses, lies at the heart of modern football. However, doing so algorithmically remains an open research challenge. To address this unmet need, we propose TacticAI, an AI football tactics assistant developed and evaluated in close collaboration with domain experts from Liverpool FC. We focus on analysing corner kicks, as they offer coaches the most direct opportunities for interventions and improvements. TacticAI incorporates both a predictive and a generative component, allowing the coaches to effectively sample and explore alternative player setups for each corner kick routine and to select those with the highest predicted likelihood of success. We validate TacticAI on a number of relevant benchmark tasks: predicting receivers and shot attempts and recommending player position adjustments. The utility of TacticAI is validated by a qualitative study conducted with football domain experts at Liverpool FC. We show that TacticAI's model suggestions are not only indistinguishable from real tactics, but also favoured over existing tactics 90% of the time, and that TacticAI offers an effective corner kick retrieval system. TacticAI achieves these results despite the limited availability of gold-standard data, achieving data efficiency through geometric deep learning.

Student of Games: A unified learning algorithm for both perfect and imperfect information games.

Games have a long history as benchmarks for progress in artificial intelligence. Approaches using search and learning produced strong performance across many perfect information games, and approaches using game-theoretic reasoning and learning demonstrated strong performance for specific imperfect information poker variants. We introduce Student of Games, a general-purpose algorithm that unifies previous approaches, combining guided search, self-play learning, and game-theoretic reasoning. Student of Games achieves strong empirical performance in large perfect and imperfect information games-an important step toward truly general algorithms for arbitrary environments. We prove that Student of Games is sound, converging to perfect play as available computation and approximation capacity increases. Student of Games reaches strong performance in chess and Go, beats the strongest openly available agent in heads-up no-limit Texas hold'em poker, and defeats the state-of-the-art agent in Scotland Yard, an imperfect information game that illustrates the value of guided search, learning, and game-theoretic reasoning.

Angiotensin II type 1 receptor agonistic autoantibody blockade improves postpartum hypertension and cardiac mitochondrial function in rat model of preeclampsia.

Women with preeclampsia (PE) have a greater risk of developing hypertension, cardiovascular disease (CVD), and renal disease later in life. Angiotensin II type I receptor agonistic autoantibodies (AT1-AAs) are elevated in women with PE during pregnancy and up to 2-year postpartum (PP), and in the reduced uterine perfusion pressure (RUPP) rat model of PE. Blockade of AT1-AA with a specific 7 amino acid peptide binding sequence ('n7AAc') improves pathophysiology observed in RUPP rats; however, the long-term effects of AT1-AA inhibition in PP is unknown. Pregnant Sprague Dawley rats were divided into three groups: normal pregnant (NP) (n = 16), RUPP (n = 15), and RUPP + 'n7AAc' (n = 16). Gestational day 14, RUPP surgery was performed and 'n7AAc' (144 µg/day) administered via osmotic minipump. At 10-week PP, mean arterial pressure (MAP), renal glomerular filtration rate (GFR) and cardiac functions, and cardiac mitochondria function were assessed. MAP was elevated PP in RUPP vs. NP (126 ± 4 vs. 116 ± 3 mmHg, p < 0.05), but was normalized in in RUPP + 'n7AAc' (109 ± 3 mmHg) vs. RUPP (p < 0.05). PP heart size was reduced by RUPP + 'n7AAc' vs. RUPP rats (p < 0.05). Complex IV protein abundance and enzymatic activity, along with glutamate/malate-driven respiration (complexes I, III, and IV), were reduced in the heart of RUPP vs. NP rats which was prevented with 'n7AAc'. AT1-AA inhibition during pregnancy not only improves blood pressure and pathophysiology of PE in rats during pregnancy, but also long-term changes in blood pressure, cardiac hypertrophy, and cardiac mitochondrial function PP.

Physical therapy assessment and whole-body magnetic resonance imaging findings in children with glycogen storage disease type IIIa: A clinical study and review of the literature.

INTRODUCTION: Early recognized manifestations of GSD III include hypoglycemia, hepatomegaly, and elevated liver enzymes. Motor symptoms such as fatigue, muscle weakness, functional impairments, and muscle wasting are typically reported in the 3rd to 4th decade of life. OBJECTIVE: In this study, we investigated the early musculoskeletal findings in children with GSD IIIa, compared to a cohort of adults with GSD IIIa. METHODS: We utilized a comprehensive number of physical therapy outcome measures to cross-sectionally assess strength and gross motor function including the modified Medical Research Council (mMRC) scale, grip and lateral/key pinch, Gross Motor Function Measure (GMFM), Gait, Stairs, Gowers, Chair (GSGC) test, 6 Minute Walk Test (6MWT), and Bruininks-Oseretsky Test of Motor Proficiency Ed. 2 (BOT-2). We also assessed laboratory biomarkers (AST, ALT, CK and urine Glc4) and conducted whole-body magnetic resonance imaging (WBMRI) to evaluate for proton density fat fraction (PDFF) in children with GSD IIIa. Nerve Conduction Studies and Electromyography results were analyzed where available and a thorough literature review was conducted. RESULTS: There were a total of 22 individuals with GSD IIIa evaluated in our study, 17 pediatric patients and 5 adult patients. These pediatric patients demonstrated weakness on manual muscle testing, decreased grip and lateral/key pinch strength, and decreased functional ability compared to non-disease peers on the GMFM, 6MWT, BOT-2, and GSGC. Additionally, all laboratory biomarkers analyzed and PDFF obtained from WBMRI were increased in comparison to non-diseased peers. In comparison to the pediatric cohort, adults demonstrated worse overall performance on functional assessments demonstrating the expected progression of disease phenotype with age. CONCLUSION: These results demonstrate the presence of early musculoskeletal involvement in children with GSD IIIa, most evident on physical therapy assessments, in addition to the more commonly reported hepatic symptoms. Muscular weakness in both children and adults was most significant in proximal and trunk musculature, and intrinsic musculature of the hands. These findings indicate the importance of early assessment of patients with GSD IIIa for detection of muscular weakness and development of treatment approaches that target both the liver and muscle.

Quantitative whole-body magnetic resonance imaging in children with Pompe disease: Clinical tools to evaluate severity of muscle disease.

OBJECTIVE: Since the introduction of enzyme replacement therapy (ERT) with alglucosidase alfa, there has been increased survival in patients with Pompe disease. It is essential to characterize and quantify the burden of disease in these patients. Here, we report a measure of muscle fat infiltration in children with infantile and pediatric late-onset Pompe disease (IPD and LOPD, respectively) to better understand the extent of muscle involvement. METHODS: Eleven pediatric patients with Pompe disease (five IPD, six LOPD), ages 7-17 years, received whole-body magnetic resonance imaging (WBMRI), muscle strength testing using the modified Medical Research Council (mMRC) scale, functional assessment using gait, stairs, gowers, chair (GSGC), and urine glucose tetrasaccharide (Glc4) testing. The intramuscular fat seen on WBMRI was quantified using proton density fat fraction (PDFF) and correlated to appropriate muscle strength and functional tests, and urine Glc4. RESULTS: Patients with IPD, although younger, had higher mean PDFF values than LOPD patients (11.61% vs 8.52%). Significant correlation existed between PDFF and the GSGC assessment (r = .9273, P = .0003). Moderate correlation existed between PDFF and mMRC (r = -.667, P = .0831), and PDFF and urine Glc4 (r = .6121, P = .0667). Anterior tibialis was in the top quartile of muscle involvement for patients with LOPD. CONCLUSION: In the past, physical therapy assessments alone have been used to track disease progression. Here, we show the clinical utility of WBMRI in quantifying muscle involvement in children with Pompe disease, especially regarding the novel involvement of anterior tibialis in children with LOPD, to better assess baseline muscle burden and mapping disease progression in children treated with ERT.

Whole-body magnetic resonance imaging in late-onset Pompe disease: Clinical utility and correlation with functional measures.

Whole-body magnetic resonance imaging (WBMRI) has clinical utility in measuring the amount of fatty infiltration in late-onset Pompe disease (LOPD). Muscle strength and function testing also provide valuable insight to the progression of myopathy seen in these patients. The main purpose of this study was to determine how closely muscle strength and functional testing correlate to the amount of fatty infiltration seen on WBMRI. LOPD patients were followed longitudinally and WBMRI, muscle strength testing using the modified Medical Research Council (mMRC) scale, muscle function testing using the Gait, Stairs, Gowers, Chair (GSGC) score, and labs including urinary glucose tetrasaccharide (Glc4) were performed at each visit. The amount of fat seen on WBMRI was quantified using proton density fat fraction (PDFF) and correlated to appropriate muscle strength and functional tests. Nineteen patients with LOPD aged 10 to 67 years were followed for a 1 to 2 year duration. There was a small increase of 1.26% (±2.57%) in overall PDFF per year in patients on ERT. Muscle strength (mMRC) and functional testing (GSGC) correlated highly with PDFF (r = -.7596, P < .0001 and r = .8267, P < .0001, respectively). Time to carry out individual tasks of the GSGC also correlated highly with PDFF of the muscles involved. Glc4 levels were normal on most visits (27/39) despite varying severity of muscle weakness in patients. Muscle strength and GSGC scores correlate highly with PDFF values from WBMRI. They may be used in assessing severity of muscle disease and to follow LOPD patients over time.

Water, sanitation, and hygiene in schools: Status and implications of low coverage in Ethiopia, Kenya, Mozambique, Rwanda, Uganda, and Zambia.

Adequate access to water, sanitation, and hygiene (WaSH) in schools impacts health, educational outcomes, and gender disparities. Little multi-country research has been published on WaSH in rural schools in Sub-Saharan Africa. In this multi-national cross-sectional WaSH study, we document WaSH access, continuity, quality, quantity, and reliability in 2270 schools that were randomly sampled in rural regions of six Sub-Saharan African countries: Ethiopia, Kenya, Mozambique, Rwanda, Uganda, and Zambia. Data collection included: school WaSH surveys containing internationally established WaSH indicators, direct observation, and field- and laboratory-based microbiological water quality testing. We found 1% of rural schools in Ethiopia and Mozambique to 23% of rural schools in Rwanda had improved water sources on premises, improved sanitation, and water and soap for handwashing. Fewer than 23% of rural schools in the six countries studied met the World Health Organization's recommended student-to-latrine ratios for boys and for girls. Fewer than 20% were observed to have at least four of five recommended menstrual hygiene services (separate-sex latrines with doors and locks, water for use, waste bin). The low access to safe and adequate WaSH services in rural schools suggest opportunities for WaSH interventions that could have substantive impact on health, education, and gender disparities.

Water quality, compliance, and health outcomes among utilities implementing Water Safety Plans in France and Spain.

Water Safety Plans (WSPs), recommended by the World Health Organization since 2004, seek to proactively identify potential risks to drinking water supplies and implement preventive barriers that improve safety. To evaluate the outcomes of WSP application in large drinking water systems in France and Spain, we undertook analysis of water quality and compliance indicators between 2003 and 2015, in conjunction with an observational retrospective cohort study of acute gastroenteritis incidence, before and after WSPs were implemented at five locations. Measured water quality indicators included bacteria (E. coli, fecal streptococci, total coliform, heterotrophic plate count), disinfectants (residual free and total chlorine), disinfection by-products (trihalomethanes, bromate), aluminum, pH, turbidity, and total organic carbon, comprising about 240K manual samples and 1.2M automated sensor readings. We used multiple, Poisson, or Tobit regression models to evaluate water quality before and after the WSP intervention. The compliance assessment analyzed exceedances of regulated, recommended, or operational water quality thresholds using chi-squared or Fisher's exact tests. Poisson regression was used to examine acute gastroenteritis incidence rates in WSP-affected drinking water service areas relative to a comparison area. Implementation of a WSP generally resulted in unchanged or improved water quality, while compliance improved at most locations. Evidence for reduced acute gastroenteritis incidence following WSP implementation was found at only one of the three locations examined. Outcomes of WSPs should be expected to vary across large water utilities in developed nations, as the intervention itself is adapted to the needs of each location. The approach may translate to diverse water quality, compliance, and health outcomes.

DeepStack: Expert-level artificial intelligence in heads-up no-limit poker.

Artificial intelligence has seen several breakthroughs in recent years, with games often serving as milestones. A common feature of these games is that players have perfect information. Poker, the quintessential game of imperfect information, is a long-standing challenge problem in artificial intelligence. We introduce DeepStack, an algorithm for imperfect-information settings. It combines recursive reasoning to handle information asymmetry, decomposition to focus computation on the relevant decision, and a form of intuition that is automatically learned from self-play using deep learning. In a study involving 44,000 hands of poker, DeepStack defeated, with statistical significance, professional poker players in heads-up no-limit Texas hold'em. The approach is theoretically sound and is shown to produce strategies that are more difficult to exploit than prior approaches.

Computer science. Heads-up limit hold'em poker is solved.

Poker is a family of games that exhibit imperfect information, where players do not have full knowledge of past events. Whereas many perfect-information games have been solved (e.g., Connect Four and checkers), no nontrivial imperfect-information game played competitively by humans has previously been solved. Here, we announce that heads-up limit Texas hold'em is now essentially weakly solved. Furthermore, this computation formally proves the common wisdom that the dealer in the game holds a substantial advantage. This result was enabled by a new algorithm, CFR(+), which is capable of solving extensive-form games orders of magnitude larger than previously possible.

Effectiveness of public health quality improvement training approaches: application, application, application.

CONTEXT: Quality improvement (QI) has been identified as a key strategy to improve the performance of state and local public health agencies. Quality improvement training effectiveness has received little attention in the literature. OBJECTIVES: To evaluate the effectiveness of 3 QI training types: webinars, workshops, and demonstration site activities on improving participant knowledge, skill, and ability to conduct QI through a questionnaire conducted after training participation. DESIGN: We used a natural experimental design hypothesizing that demonstration site participants would have the greatest gains on outcomes of interest compared with webinar and workshop participants. Bivariate and multivariate models were used to examine outcome differences between questionnaire respondents who participated in various training types. PARTICIPANTS: Local health department employees who participated in the 3 training strategies. MAIN OUTCOME MEASURES: Measures included knowledge and skill gain, skill application, QI receptivity, and ability to successfully participate in a QI project. RESULTS: Two hundred eighty-four unique individuals who work in 143 health departments completed the questionnaire for a 59% response rate. The majority of these health departments serve midsize populations. Demonstration site respondents had significantly greater gains in knowledge and skills, skill application, and ability to successfully participate in a QI project. Webcast training participants had significantly higher QI receptivity in adjusted models. Respondents who participated in both webcast and demonstration site trainings had higher mean scores on all outcomes when compared with demonstration site single training participants, these differences were significant in unadjusted models. CONCLUSION: Our findings suggest that QI training for public health agency employees should include both didactic training on QI content and opportunities for QI application. Future research should examine if this approach can effectively increase successful participation in QI projects for staff in LHDs of all sizes.

Linear Fitted-Q Iteration with Multiple Reward Functions.

We present a general and detailed development of an algorithm for finite-horizon fitted-Q iteration with an arbitrary number of reward signals and linear value function approximation using an arbitrary number of state features. This includes a detailed treatment of the 3-reward function case using triangulation primitives from computational geometry and a method for identifying globally dominated actions. We also present an example of how our methods can be used to construct a real-world decision aid by considering symptom reduction, weight gain, and quality of life in sequential treatments for schizophrenia. Finally, we discuss future directions in which to take this work that will further enable our methods to make a positive impact on the field of evidence-based clinical decision support.

Intensive case management before and after prison release is no more effective than comprehensive pre-release discharge planning in linking HIV-infected prisoners to care: a randomized trial.

Imprisonment provides opportunities for the diagnosis and successful treatment of HIV, however, the benefits of antiretroviral therapy are frequently lost following release due to suboptimal access and utilization of health care and services. In response, some have advocated for development of intensive case-management interventions spanning incarceration and release to support treatment adherence and community re-entry for HIV-infected releasees. We conducted a randomized controlled trial of a motivational Strengths Model bridging case management intervention (BCM) beginning approximately 3 months prior to and continuing 6 months after release versus a standard of care prison-administered discharge planning program (SOC) for HIV-infected state prison inmates. The primary outcome variable was self-reported access to post-release medical care. Of the 104 inmates enrolled, 89 had at least 1 post-release study visit. Of these, 65.1% of BCM and 54.4% of SOC assigned participants attended a routine medical appointment within 4 weeks of release (P > 0.3). By week 12 post-release, 88.4% of the BCM arm and 78.3% of the SOC arm had at attended at least one medical appointment (P = 0.2), increasing in both arms at week 24-90.7% with BCM and 89.1% with SOC (P > 0.5). No participant without a routine medical visit by week 24 attended an appointment from weeks 24 to 48. The mean number of clinic visits during the 48 weeks post release was 5.23 (SD = 3.14) for BCM and 4.07 (SD = 3.20) for SOC (P > 0.5). There were no significant differences between arms in social service utilization and re-incarceration rates were also similar. We found that a case management intervention bridging incarceration and release was no more effective than a less intensive pre-release discharge planning program in supporting health and social service utilization for HIV-infected individuals released from prison.

Risk factors for injury among high school football players.

BACKGROUND: Football is the most popular interscholastic high school sport in the United States. Prior research has described a higher rate of injury among high school football players than in other sports, but few studies have examined potential predictors while controlling for other risk factors. METHODS: Using a 2-stage cluster sampling technique, we conducted a prospective cohort study from 1996 to 1999 among varsity athletes from 12 sports in 100 North Carolina high schools. A total of 3323 football players participated. Injury exposure and risk factor data were collected by trained school personnel. Incidence rates, rate ratios, and odds ratios (ORs) were estimated using Poisson and logistic regression. RESULTS: There were 1064 injured athletes and 1238 injuries; 106 injuries resulted in greater than 3 weeks lost from participation. The overall incidence rate was 3.54 per 1000 athlete-exposures (95% confidence interval [CI] = 3.31-3.78). The rate of game injury was 9 times that of practice injury (OR = 9.2; 95% CI = 6.6-11). Athletes with a prior injury had twice the injury rate of those without (1.9; 1.5-2.4). Among those injured, having a coach with more experience, qualifications, and training was associated with half the odds of severe injury (0.49; 0.27-0.92). CONCLUSIONS: Prior injury, additional years of playing experience, and older age were predictors of injury incidence after controlling for multiple risk factors. A high level of coaching skills did not reduce the injury rate, but was protective against severe injury.