A Meta-Analytic Approach to Swimming Performance Prediction: Reviewing Methods, Datasets, and Research Trends

Indonesian Journal of Kinanthropology (IJOK) Volume 5 Number 2, July 2025 https://doi.org/10.26740/ijok.v5n2.p52-71 E-ISSN: 2775-2178 Indonesian Journal of Kinanthropology (IJOK) Open Access A Meta-Analytic Approach to Swimming Performance Prediction: Reviewing Methods, Datasets, and Research Trends Ari Tri Fitrianto1, Muhammad Habibie1, Parveen Kumar2 1 Universitas 2 Islam Kalimantan Muhammad Arsyad Al Banjari, Kota Banjarmasin, Kalimantan Selatan, Indonesia 70123 Chaudhary Ranbir Singh University, Jind, Haryana, India, 126102 Correspondence: (Received: 28 July 2025 | Revised: 4 September 2025 | Accepted: 6 November 2025) ABSTRACT Background: The PICO (Population, Intervention, Comparison, Outcomes) framework is widely applied to guide systematic reviews. In this study, we extend it using the PICOC variant (Population, Intervention, Comparison, Outcomes, Context) to frame the research on swimming performance prediction. Methods: A Systematic Literature Review (SLR) and Meta-Analysis were conducted following PRISMA guidelines. Articles were retrieved from five major databases—ScienceDirect, Springer, Taylor & Francis, PubMed, and Google Scholar—covering the years 2014–2024. Twenty-one studies were included for analysis. Results: Research trends show increased attention to freestyle performance, with most studies relying on private datasets (16 studies) and fewer on public datasets (5 studies, primarily Olympic and FINA records). Across studies, predictive mathematical models and linear regression were most commonly applied. The meta-analysis revealed moderate heterogeneity (I² = 31%) but generally consistent findings across studies. Conclusions: Swimming performance prediction is an emerging research area that provides value for athlete training, talent identification, and competition preparation. Continued development of hybrid models and expanded use of standardized public datasets are recommended. Keywords: XAI; PRISMA; PICOC; statistics 1. Background Swimming performance is a complex phenomenon shaped by multiple interacting factors, including anthropometric, hydrodynamic, kinematic, and energetic aspects (de Anda Martín et al., 2024; Morais et al., 2022). These domains jointly determine a swimmer’s speed and efficiency, which ultimately affect race outcomes. Previous studies highlight that body composition (e.g., limb length, body ratio), water resistance, movement patterns, and physiological capacity serve as key indicators for predicting performance and identifying young athletic talent (Lobato et al., 2023; Marinho et al., 2020). Among biomechanical determinants, stroke rate and stroke length are strongly associated with swimming velocity, where faster and longer arm strokes contribute to improved performance (Nurmukhanbetova et al., 2023). Training frequency, duration, and intensity also play essential roles, as higher workloads demand structured training volumes (Armen et al., 2024). Indonesian Journal of Kinanthropology (IJOK) | Volume 5 | Number 2 | 2025 | 52-71 52 Ari Tri Fitrianto, Muhammad Habibie, Parveen Kumar A Meta-Analytic Approach to Swimming Performance Prediction: Reviewing Methods, Datasets, and Research Trends Another fundamental determinant is muscular strength, which underpins efficient propulsive force in the water (Apriyano et al., 2025; Sadewa et al., 2024). Strength development is closely tied to physiological adaptations that emerge through systematic training, including enhanced aerobic capacity, metabolic efficiency, and neuromuscular coordination (Nugent et al., 2019). Together, these adaptations reinforce overall swimming performance. Research on swimming performance prediction has progressed significantly over the past decades. Early studies often relied on linear models, such as differential equations and regression analyses, to capture relationships between training load and performance outcomes (Banister & Calvert, 1980; Busso et al., 1990, 1997; Chatard & Stewart, 2011; Fitz-Clarke JR et al., 1991; Hohmann, 1992) . However, biological systems are dynamic and adaptive, making linear approaches insufficient for modeling the complexity of training responses (Edelmann-Nusser et al., 2002). As a result, recent studies increasingly employ machine learning and non-linear algorithms to capture multidimensional relationships among performance-related variables (Staunton et al., 2024a). These approaches offer opportunities for more accurate, individualized prediction systems and more adaptive, evidence-based training strategies. To advance this field, the present study adopts a combined Systematic Literature Review (SLR) and MetaAnalysis approach. The SLR ensures a transparent and structured evaluation of current evidence using explicit inclusion and exclusion criteria (Okoli & Schabram, 2012), while Meta-Analysis enhances validity by statistically synthesizing findings across studies (Gurevitch et al., 2018). Together, these methods provide comprehensive insights into recent developments in swimming performance prediction. This review covers publications from 2014 to 2024, classifying them into research questions (RQs) to map existing knowledge, highlight trends, and identify methodological and conceptual gaps. The primary aim of this study is to systematically examine and synthesize the body of research on swimming performance prediction from 2014 to 2024. Specifically, the study seeks to (1) identify and classify the main determinants of swimming performance, (2) evaluate methodological and analytical approaches used in prior studies, and (3) highlight knowledge gaps to guide future research and inform evidence-based training practices. 2. Methods This research on swimming performance prediction was conducted using Meta-Analysis and Systematic Literature Review (SLR) approaches. The SLR method is used to identify, evaluate, and interpret thoroughly the results of previous research relevant to the topic or research question, in order to obtain measurable and evidence-based answers to the formulation of the problem proposed (Okoli & Schabram, 2012) . Based on feedback from reviewers during the revision process of this manuscript, in general, SLR consists of two main stages, namely: the planning stage and the implementation and reporting stages. In the planning stage, there are three important steps taken: Identifying the need for a systematic review, Developing a review protocol, and Conducting an initial evaluation of the study to be reviewed. Furthermore, in the implementation and reporting stage, there are four main steps: Conducting a primary source search, Selecting relevant primary studies, Extracting data from selected primary studies, and Compiling and disseminating review results (dissemination of results). The overall SLR process flow in this study is visualized in Figure 1, which illustrates the systematic steps taken to ensure the validity and transparency of the literature synthesis conducted. Table 1. PICOC (...truncated)