Toward characterizing cardiovascular fitness using machine learning based on unobtrusive data.

Cardiopulmonary exercise testing (CPET) is a non-invasive approach to measure the maximum oxygen uptake ([Formula: see text]), which is an index to assess cardiovascular fitness (CF). However, CPET is not available to all populations and cannot be obtained continuously. Thus, wearable sensors are associated with machine learning (ML) algorithms to investigate CF. Therefore, this study aimed to predict CF by using ML algorithms using data obtained by wearable technologies. For this purpose, 43 volunteers with different levels of aerobic power, who wore a wearable device to collect unobtrusive data for 7 days, were evaluated by CPET. Eleven inputs (sex, age, weight, height, and body mass index, breathing rate, minute ventilation, total hip acceleration, walking cadence, heart rate, and tidal volume) were used to predict the [Formula: see text] by support vector regression (SVR). Afterward, the SHapley Additive exPlanations (SHAP) method was used to explain their results. SVR was able to predict the CF, and the SHAP method showed that the inputs related to hemodynamic and anthropometric domains were the most important ones to predict the CF. Therefore, we conclude that the cardiovascular fitness can be predicted by wearable technologies associated with machine learning during unsupervised activities of daily living.

Using machine learning pipeline to predict entry into the attack zone in football.

Sports sciences are increasingly data-intensive nowadays since computational tools can extract information from large amounts of data and derive insights from athlete performances during the competition. This paper addresses a performance prediction problem in soccer, a popular collective sport modality played by two teams competing against each other in the same field. In a soccer game, teams score points by placing the ball into the opponent's goal and the winner is the team with the highest count of goals. Retaining possession of the ball is one key to success, but it is not enough since a team needs to score to achieve victory, which requires an offensive toward the opponent's goal. The focus of this work is to determine if analyzing the first five seconds after the control of the ball is taken by one of the teams provides enough information to determine whether the ball will reach the final quarter of the soccer field, therefore creating a goal-scoring chance. By doing so, we can further investigate which conditions increase strategic leverage. Our approach comprises modeling players' interactions as graph structures and extracting metrics from these structures. These metrics, when combined, form time series that we encode in two-dimensional representations of visual rhythms, allowing feature extraction through deep convolutional networks, coupled with a classifier to predict the outcome (whether the final quarter of the field is reached). The results indicate that offensive play near the adversary penalty area can be predicted by looking at the first five seconds. Finally, the explainability of our models reveals the main metrics along with its contributions for the final inference result, which corroborates other studies found in the literature for soccer match analysis.

Penalty feet positioning rule modification and laterality effect on soccer goalkeepers' diving kinematics.

In 2019, a new rule was applied in soccer. It allows the goalkeeper to have only one foot or part of it on the goal line when the kicker hits the ball, unlike the previous rule that determined the goalkeeper should have both feet on the line. The purpose of the present study was to analyze how the change in the rule and the lower limbs laterality influences on the diving save kinematic performance in penalties. Six goalkeepers, two professionals and four amateurs, performed a total of 20 dives in the laboratory and had their force and impulse exerted by the lower limb and displacement/velocity data from the center of body mass collected through force plates and kinematic analysis. The side preference was collected through an inventory. The results showed that goalkeepers dive further (p < 0.001) and faster (p < 0.001) when diving according to the new rule. Dives for the non-dominant side presented higher values than the trials for the dominant side in mediolateral (p = 0.02) and resultant (p = 0.03) displacements. Concluding, the goalkeepers performed better with the new rule in the analyzed variables and the lower limb preference has influenced only the mediolateral and resultant displacement.

Football player dominant region determined by a novel model based on instantaneous kinematics variables.

Dominant regions are defined as regions of the pitch where a player can reach before any other and are commonly determined without considering the free-spaces in the pitch. We presented an approach to football players' dominant regions analysis, based on movement models created from players' positions, displacement, velocity, and acceleration vectors. 109 Brazilian male professional football players were analysed during official matches, computing over 15 million positional data obtained by video-based tracking system. Movement models were created based on players' instantaneous vectorial kinematics variables, then probabilities models and dominant regions were determined. Accuracy in determining dominant regions by the proposed model was tested for different time-lag windows. We calculated the areas of dominant, free-spaces, and Voronoi regions. Mean correct predictions of dominant region were 96.56%, 88.64%, and 72.31% for one, two, and three seconds, respectively. Dominant regions areas were lower than the ones computed by Voronoi, with median values of 73 and 171 m2, respectively. A median value of 5537 m2 was presented for free-space regions, representing a large part of the pitch. The proposed movement model proved to be more realistic, representing the match dynamics and can be a useful method to evaluate the players' tactical behaviours during matches.

Multiscale fractal dimension applied to tactical analysis in football: A novel approach to evaluate the shapes of team organization on the pitch.

The aim of this study was to evaluate different shape descriptors applied to images of polygons that represent the organization of football teams on the pitch. The effectiveness of different shape descriptors (area/perimeter, fractal area, circularity, maximum fractal, rectangularity, multiscale fractal curve-MFC), and the concatenation of all shape descriptors (except MFC), denominated Alldescriptors (AllD)) was evaluated and applied to polygons corresponding to the shapes represented by the convex hull obtained from players' 2D coordinates. A content-based image retrieval system (CBIR) was applied for 25 users (mean age of 31.9 ± 8.4 years) to evaluate the relevant images. Measures of effectiveness were used to evaluate the shape descriptors (P@n and R@n). The MFD (P@5, 0.46±0.37 and P@10, 0.40±0.31, p < 0.001; R@5, 0.14±0.13 and R@10, 0.24±0.19, p < 0.001) and AllD (P@5 = 0.43±0.36 and P@10 = 0.39±0.32, p < 0.001; R@5 = 0.13±0.11 and R@10 = 0.24±0.20, p < 0.001) descriptors presented higher values of effectiveness. As a practical demonstration, the best evaluated shape descriptor (MFC) was applied for tactical analysis of an official match. K-means clustering technique was applied, and different shapes of organization could be identified throughout the match. The MFC was the most effective shape descriptor in relation to all others, making it possible to apply this descriptor in the analysis of professional football matches.

KGen: a knowledge graph generator from biomedical scientific literature.

BACKGROUND: Knowledge is often produced from data generated in scientific investigations. An ever-growing number of scientific studies in several domains result into a massive amount of data, from which obtaining new knowledge requires computational help. For example, Alzheimer's Disease, a life-threatening degenerative disease that is not yet curable. As the scientific community strives to better understand it and find a cure, great amounts of data have been generated, and new knowledge can be produced. A proper representation of such knowledge brings great benefits to researchers, to the scientific community, and consequently, to society. METHODS: In this article, we study and evaluate a semi-automatic method that generates knowledge graphs (KGs) from biomedical texts in the scientific literature. Our solution explores natural language processing techniques with the aim of extracting and representing scientific literature knowledge encoded in KGs. Our method links entities and relations represented in KGs to concepts from existing biomedical ontologies available on the Web. We demonstrate the effectiveness of our method by generating KGs from unstructured texts obtained from a set of abstracts taken from scientific papers on the Alzheimer's Disease. We involve physicians to compare our extracted triples from their manual extraction via their analysis of the abstracts. The evaluation further concerned a qualitative analysis by the physicians of the generated KGs with our software tool. RESULTS: The experimental results indicate the quality of the generated KGs. The proposed method extracts a great amount of triples, showing the effectiveness of our rule-based method employed in the identification of relations in texts. In addition, ontology links are successfully obtained, which demonstrates the effectiveness of the ontology linking method proposed in this investigation. CONCLUSIONS: We demonstrate that our proposal is effective on building ontology-linked KGs representing the knowledge obtained from biomedical scientific texts. Such representation can add value to the research in various domains, enabling researchers to compare the occurrence of concepts from different studies. The KGs generated may pave the way to potential proposal of new theories based on data analysis to advance the state of the art in their research domains.

Exploring the determinants of success in different clusters of ball possession sequences in soccer.

The purpose of this study was two-step: (1) classify ball possession (BP) according to the duration and number of passes and (2) identify which tactical variables most discriminate the different BP. We obtained 527 BPs from four official matches of the Brazilian Soccer Championship 2016. Forty-one "notational", "space occupation", and "displacement synchronization" predictor variables were used. The BPs were classified into three groups: short (11.07 ± 4.49 s, 1.93 ± 0.99 passes), medium (26.83 ± 7.33 s, 5.41 ± 1.84 passes), long (55.50 ± 14.97 s, 12.11 ± 4.61 passes). Discriminant analysis identified the five most relevant variables to describe each group: coefficient of variation (CV) of the defensive team's synchronization-Y, CV defensive team´s synchronization-X, successful pass last third, CV distance between offensive team's centroid and target, mean of the offensive team's width. The approach highlights important variables and could benefit the description of offensive and defensive game sequences to provide precise knowledge on the process.

Analysis of Match Dynamics of Different Soccer Competition Levels Based on the Player Dyads.

The purpose of this study was to analyse the dynamics of play based on dyads during soccer matches, according to the competition level, period of the matches, and playing positions. We recorded eight Brazilian soccer matches (four of the national and four of the regional level), using up to six digital cameras (30 Hz). The position information of the 204 players in the eight matches was obtained using an automatic tracking system. The Euclidean distance between the nearest opponents was calculated over time to define the dyads. The interaction between the components of dyads was assessed by the distances between players and was compared among the different positions (defender, full-back, defensive midfielder, midfielder, and forward), match periods (15, 30, 45, 60, 75, and 90 min), and competition levels. Results showed smaller distances for the national level dyads, compared to the regional matches. Greater distances between the players were found in the last 15 minutes of the matches, compared to the other periods. The full-backs were more distant from opposing players compared to players from other playing positions. Thus, coaches should consider the characteristics of each playing position and the greater proximity between opponents' players in top-level competition for the development of tactical proficiency of the players.

Sentinel plants as programmable processing units: insights from a multidisciplinary perspective about stress memory and plant signaling and their relevance at community level.

Stress memory and an effective signaling among individuals in a given community are recognized to improve plant performance under recurrent stressful conditions. As living beings with memory and signaling abilities, plants can be considered as processing units and then be trained - or programmable from a computational viewpoint - and prepared for facing biotic and abiotic stresses. Here, we propose that sentinel plants could improve the resilience of agricultural and natural communities by reducing the impact of biotic or abiotic stressors on their neighbors. Modeling plants as programmable (or trainable) processing units compels us to think about a multidisciplinary perspective for integrating stress memory, signaling, and resilience of biological systems into executable programs, fostering the creation of applications and technologies that would benefit from the spatiotemporal dynamics related to plant-plant and plant-environment interactions.