The Temporal Structure of the Running Cycle, an Essential Element in the Analysis: A Critical Review

Abstract

The running cycle is distinguished from the gait cycle by the presence of a flight phase and distinct biomechanical characteristics. Despite existing frameworks for the temporal segmentation of running, these models remain underutilized in comprehensive biomechanical analyses, particularly for delineating phases, subphases, and key events. This study aims to provide a review of historical and contemporary temporal models of the running cycle and to introduce a unified structure designed to enhance analytical precision. The proposed framework divides the running cycle into two primary phases: (a) contact (subdivided into braking and propulsion subphases) and (b) flight, together with three critical events: (1) initial contact, (2) transition of braking–propulsion, (3) toe-off. While leg swing is not considered a phase in this framework due to temporal overlap with other phases, its recognized importance in running mechanics warrants its integrated analysis under the proposed temporal phase delineation. Additionally, methodologies for identifying these events through dynamometry and motion capture are evaluated, emphasizing their role in contextualizing kinetic and kinematic data. By integrating this temporal structure, the study aims to standardize biomechanical assessments of running technique, fostering more consistent comparisons across studies. Such integration has the potential to not only refine interpretations of running mechanics but also to enable practical advancements in athletic training, injury mitigation, and performance optimization.

1. Introduction

Human locomotion has been extensively studied, with walking and running emerging as predominant movement patterns. Central to their analysis is the gait cycle, a fundamental unit defined by the period from one foot-ground contact to the next repetition by the same foot [1,2]. While this cycle underpins both walking and running, critical distinctions exist: (a) running is characterized by a flight phase (absent in walking) and (b) lacks the double contact phase inherent to walking [3,4]. These biomechanical differences necessitate specialized frameworks for analyzing running mechanics.

Despite proposals for the temporal segmentation of the running cycle into phases and key events, such models remain underutilized in biomechanical research. Current analyses often reduce running to isolated events like foot strike and toe-off, neglecting finer subdivisions (e.g., braking, propulsion). This oversimplification hinders a comprehensive interpretation of kinetic and kinematic data, complicating cross-study comparisons and hypothesis generation [5,6]. A unified temporal structure is therefore essential to contextualize findings and advance mechanistic insights into running technique.

Sports technique, defined as an athlete’s structured movement sequence to solve motor problems efficiently [7], must adapt to discipline-specific demands. For instance, sprinting and marathon running impose distinct biomechanical requirements, underscoring the need for tailored temporal models [8]. This review aims to (1) synthesize existing temporal frameworks for the running cycle, proposing a cohesive structure for biomechanical analysis for steady-state running, and (2) identify methodologies for identifying key events (e.g., via dynamometry, motion capture).

1.1. The Phenomenon of Running from the Point of View of Sports Biomechanics

The study of running biomechanics encompasses diverse frameworks, including pedagogical, sports performance, and clinical perspectives [8]. Pedagogically, the focus lies on optimizing motor pattern development throughout biological maturation to mature running mechanics [9]. Clinically, research prioritizes injury prevention and rehabilitation strategies [10,11]. In contrast, sports biomechanics centers on technique analysis to enhance athletic performance and competitive outcomes [12,13,14]. While these approaches share foundational knowledge, their analytical objectives diverge significantly, a distinction critical to contextualizing research methodologies and interpretations.

Qualitative sports technique analysis relies on systematic observation and introspective evaluation of movement [15]. Central to this process is phase analysis, which dissects techniques into sequential, purpose-driven phases bounded by key events (e.g., foot strike) [16,17]. Temporal structuring of these phases enables a precise examination of their biomechanical contributions to overall performance [18]. Morante and Izquierdo [19] advanced this paradigm by proposing a three-tier analytical model, as follows: (1) define the overall performance objective of the technique, (2) identify phase-specific mechanical purposes, and (3) select measurable biomechanics indicators (kinetic/kinematic) aligned with these goals. Currently, biomechanical analysis models have been performed for various sports techniques [20,21,22], and this framework remains conspicuously absent in running biomechanics—a gap hindering standardized technique evaluation and innovation.

1.2. History and Evolution of the Running Cycle

The temporal structure of the running cycle has been conceptualized through diverse models, reflecting evolving methodologies and analytical priorities. This review chronologically examines these frameworks, highlighting their contributions and limitations. To be as faithful as possible to the models proposed in the literature, the authors’ names of the phases and events were cited verbatim, although they sometimes refer to the same phenomenon. Figure 1 illustrates four models that have been relevant to running cycle compression.

The first formal definition of the running cycle was the one proposed by Slocum and James [3], who introduced a two-phase model: support phase and forward recovery phase (Figure 1A). The support phase was subdivided into foot-strike (initial contact to foot fixation), mid-support (fixed foot to heel lift), and takeoff (heel lift to toe-off). The recovery phase included follow-through (limb deceleration), forward swing (thigh flexion), and foot descent (hip flexion to landing). His anatomical–functional analysis compared running with coordinated jumps, and although an aerial state is mentioned, it is omitted as a formal component of the cycle. In the decade of the 70s, Elliott and Blanksby [23] focused on analyzing discrete measurement points; however, these key events did not delimit concrete phases that could serve a real mechanistic purpose in the running cycle, so they were not replicated in future research.

At the end of the 80s, Munro, et al. [24] determined a transcendental breakthrough, utilizing the ground reaction force (GRF), to delineate two distinct subphases. This segmentation, based on the anteroposterior GRF, established the braking subphase, characterized by a force opposing forward motion, and the propulsion subphase, where a force aligns with forward movement. This classification, grounded in measurable dynamometric indicators, provided a quantifiable framework for subphase determination, a departure from purely observational models. In the same decade, a paradigm shift was introduced by Blickhan [25] with the spring-mass model, a conceptualization of running and jumping as an active, nonlinear, multicomponent spring-mass system. Although this model did not extend to the delimitation of the subphases nor to the identification of key events within the running cycle, it has allowed several advances in the field of running biomechanics thanks to the successful prediction of mechanical parameters (e.g., leg stiffness).

Between the 1990s and 2000s, Dugan and Bhat [26] and Thordarson [27] proposed a model, currently among the most widely used, which builds upon the framework established by Ounpuu [28] (Figure 1B). This model delineates three primary phases: stance, swing, and float. The stance phase is segmented into absorption (from the initial contact to mid-stance) and propulsion (from the mid-stance to toe-off), while the swing phase is divided into initial swing (from the toe-off to mid-swing) and terminal swing (from the mid-swing to new initial contact). Notably, this model introduces overlapping swing and float phases, particularly at the onset and conclusion of the swing phase, thereby disrupting the strict sequential progression of gait events. In the same decade, Novacheck [2] proposed a four-phase model: stance absorption, stance generation, swing generation, and swing absorption (Figure 1C). These phases were characterized by the deceleration and acceleration of the center of mass, respectively. Key events, including initial contact, toe-off, stance reversal, and swing reversal, were identified as markers for transitions between the absorption and generation subphases within both stance and swing. While acknowledging a period of flotation resulting from limb swing, Novacheck did not incorporate it as a formal phase of the running gait. However, neither of the two proposals provided explicit definitions or delineations for these phases, subphases, or events, which impedes the standardization of criteria for the running cycle.

In a more recent approach to maximal sprint running analysis, Mattes et al. [29] introduced the swing-pull technique (Figure 1D), which delineates the running cycle into two primary phases: swing and pull. The swing phase is subdivided into swing off and forward swing; the pull phase is subdivided into pre-support and support. Notably, this model incorporates a tertiary level of division corresponding to the actions within the sub-phase, the forward swing is segmented into heel-to-buttocks and knee-lift, and pre-support is divided into swing-down and backswing. While the support subphase is not formally subdivided, an intermediate event, full-support, is identified. While this model offers a potentially valuable perspective on sprint mechanics, further refinement is necessary to ensure accurate interpretation and to fully elucidate the contribution of the proposed phases.

Figure 1. Temporal structure of relevant models in the literature. (A): Model presented by Slocum and James [3]; (B): model presented by Ounpuu [28], Dugan and Bhat [26], and Thordarson [27]; (C): model presented by Novacheck [2]; (D): model presented by Mattes, Wolff, and Alizadeh [29].

Figure 1. Temporal structure of relevant models in the literature. (A): Model presented by Slocum and James [3]; (B): model presented by Ounpuu [28], Dugan and Bhat [26], and Thordarson [27]; (C): model presented by Novacheck [2]; (D): model presented by Mattes, Wolff, and Alizadeh [29].

While these methodologies represent advancements, a standardized consensus regarding the temporal structure of running remains elusive despite nearly a century of research. The presence of inconsistent definitions, such as the conflation of ‘phases’ with ‘events’, overlapping classification systems, and the frequent exclusion of the flight phase, significantly impedes cross-study comparative analyses. Consequently, the diversity and imprecision inherent in current conceptual frameworks are likely to pose substantial obstacles to integrating temporal structure into biomechanical analyses of running technique. Therefore, the establishment of harmonized terminology is crucial to facilitate both research progression and practical application.

1.3. The Underuse of Temporal Structure in Biomechanical Analysis

Despite the collective importance attributed to the phases and events of running techniques, these models are underutilized as a complement to biomechanical analysis techniques. While kinetic analysis techniques such as dynamometry are limited to analyzing the contact phase, time-structure models can be valuable for kinematic and electromyographic analysis. Several reviews and meta-analyses have highlighted the difficulties caused by the high heterogeneity of methods, analysis techniques, and data presentation [30,31,32].

2. Materials and Methods

The present research was conducted based on the criteria described by Grant and Booth [33], considering the strengths and limitations of a critical review.

An initial search was conducted in the PubMed/MEDLINE and Web of Science databases, using different combinations of the following keywords: “running”, “biomechanics”, “mechanics”, “technique”, “cycle”, “gait”, “phase”, and “event”. Titles and abstracts served as initial selection criteria, and the full texts were subsequently reviewed to assess the results’ relevance to the research objectives. In addition, a supplementary search using citation chasing was conducted to identify original articles proposing the phase models cited in the literature. There were no restrictions based on the study design, language, or year of publication. Only the full text of one article could not be retrieved for this review. (Ounpuu [28]).

3. Results

3.1. Model of Temporal Structure for Running Cycle in Sport

To establish a robust temporal framework for the running cycle, we first define its structural components. Drawing on vocabulary references, a cycle constitutes a recurring period in which phenomena repeat in a fixed sequence [34]. Within this cycle, a phase represents a distinct stage demarcated by temporal or positional criteria [35], while an event denotes a discrete moment or spatial marker [36]. Building on these definitions, we derive four core criteria for constructing a biomechanically rigorous running cycle model, as follows: (1) Explicit Characterization: phases and key events must be unambiguously defined, with biomechanical or functional rationales. (2) Sequential Logic: phases must be mutually exclusive. (3) Temporal Precision: phase boundaries must be delimited to measurable key events (e.g., foot strike, toe-off). (4) Quantitative Delimited: events must be identifiable via objective methods.

None of the previously reviewed models fulfill all the criteria, primarily due to inconsistencies in the definition of key events, which, in turn, hinders their identification through qualitative methods. Furthermore, some models are not adaptable to certain running styles (which will be addressed subsequently), or they present an overlap of phases during the cycle. To address these gaps, we propose a running cycle structure designed to integrate with sports technical analysis. The conceptual definitions of each phase and key event are detailed in

Table 1. Definition of phases, subphases, and key events of the temporal structure to running cycle.

	Definition
Initial contact	The instant in which the foot contacts the ground, regardless of the portion of the foot that makes contact [2,26,27]
Transition of braking–propulsion (TB–P)	Moment of transition from the braking subphase to the propulsion subphase by a change in sense of the anterior-posterior force vector [2,26,37]
Toe-off	Moment in which the foot leaves the ground [2,26,27]
Contact phase	Period of time in which any part of the foot rests unilaterally on the ground. It starts with the initial contact and ends with the toe-off [3,25,27,38,39]
Braking subphase	The period in which the foot is in contact with the ground and the GRF opposes the forward movement is characterized by the deceleration and descent of the center of mass and the accumulation of elastic energy in the lower limb. It begins with the initial contact and ends with the toe-off [2,24,25]
Propulsion subphase	The period in which the foot is in contact with the ground and the GRF is coherent to the forward movement is characterized by the acceleration and rise in the center of mass, in addition to the release of elastic energy of the lower limb. It begins with and ends with the toe-off [2,24,25]
Flight phase	Period of time in which there is no contact between the ground and the person, starting with the toe-off and ending with the initial contact of the opposite limb [11,26,27,38,40]

GRF: Ground reaction force.

, and the time structure of the race cycle is illustrated in Figure 2.

3.1.1. Definition and Characterization of Phases and Subphases

The contact phase (universally acknowledged across the running biomechanics literature) is defined as follows: that period in which any part of the foot rests unilaterally on the ground [3,25,27,38,39]. Early subdivisions of this phase into foot strike, mid-support, and takeoff [3] provided foundational insights but lacked precision and generalizability. For instance, these subdivisions fail to account for forefoot strikers [41] or elite sprinters, who often avoid heel contact entirely [27]. This is also generalizable to the “full-support” event in the “swing-pull technique” model of Mattes, Wolff, and Alizadeh [29]. Such limitations underscore the need for a model adaptable to diverse running patterns.

A paradigm shift emerged with Munro, Miller, and Fuglevand [24], who subdivided the contact phase into braking and propulsion based on GRF dynamics. While absorption is often used synonymously with braking, emphasizing shock dissipation via eccentric muscle action [26], the term braking better aligns with performance contexts, where athletes aim to minimize deceleration. Therefore, the braking phase is defined as follows: that period in which the foot is in contact with the ground and the GRF opposes the forward movement; it is characterized by the deceleration and descent of the center of mass, in addition to the accumulation of elastic energy in the lower limb [2,24,25]. Conversely the propulsion phase is to be defined as follows: the period in which the foot is in contact with the ground and the GRF is coherent with the forward movement; it is characterized by the acceleration and rise in the center of mass, in addition to the release of elastic energy in the lower limb [2,24,25].

In contrast to the stance phase, defined by foot–ground interactions, the flight and swing phases exhibit marked inconsistency across the proposed models. The flight phase, characterized by a period of complete non-contact between the runner and the ground, has been described in various studies [11,26,27,38,40]. Conversely, the swing phase, representing a non-weight-bearing limb’s forward motion in preparation for subsequent ground contact, is similarly defined [2,3,27,39,42,43]. The distinction between these phases is subtle and lies in limb involvement: the swing phase denotes single-limb non-contact, while the flight phase signifies bilateral non-contact. This inherent simultaneity creates a cyclical paradox. In particular, the swing phase has been extrapolated from the clinical compression of gait to race technique [11], and although the assessment of individual body segment contributions is fundamental to understanding running biomechanics [44], the categorization of leg swing as a distinct phase within the running cycle presents a conceptual challenge. Specifically, following this logic, arm swing could also be considered a phase. This necessitates a critical examination of how segmental motion is integrated into the broader framework of running phases, ensuring a cohesive and biomechanically sound model.

The interpretation of running as a sequence of coordinated jumps [3] emphasizes the fundamental importance of contact and flight phases, drawing parallels with jumping biomechanics, while gait analysis typically predominates the swing phase [11]. The performance-oriented nature of running necessitates a heightened focus on the flight phase, a component often overlooked despite its critical influence on stride optimization [45]. By excluding swing as a formal phase, this paradigm aims to ensure that the temporal structure of running analysis is directly aligned with ground interaction—the key distinguishing feature between running and walking.

It is crucial to emphasize that the exclusion of the swing phase from the proposed model does not diminish its significance in running mechanics. Numerous studies have analyzed swing limb kinematics and kinetics, demonstrating its relevance to performance [46,47,48], running economy [13,49], and injury prevention [45,50]. Consequently, the analysis of leg swing as a component of the running mechanics is expected to persist, and to be complemented by the phase model presented in this review. In the framework of the biomechanical analysis proposed by Morante and Izquierdo [19], leg swing would have the role of a biomechanical indicator, contributing to the achievement of a higher purpose, such as the mechanical purposes of each phase and the overall performance goal of the sport technique.

Regarding the enhancement of performance, the findings indicate the relevance of the leg swing in increasing the running speed. Comparative analysis of the leg swing between elite and sub-elite athletes revealed that the elite athlete’s technical differences during the flight phase and prior to ground contact facilitate improved positioning of the stance leg, resulting in greater [48] joint torque and increased vertical ground reaction force at the onset of the contact phase. Furthermore, the significance of the iliopsoas, gluteus maximus, and hamstring muscles during the leg swing was substantiated through two mechanisms: firstly, these muscles contribute to angular acceleration in hip and knee flexion during the swing phase; and secondly, the force generated by the hamstrings of the swinging leg augments the angular acceleration of knee extension in the contralateral leg during the contact phase [46]. In these instances, the leg swing serves as a biomechanical indicator, potentially contributing to the mechanical objectives of both the flight and contact phases. Specifically, it aids in the optimal positioning of the leg at the termination of the flight phase, as well as in the production of the angular acceleration of the contralateral leg during the contact phase [19].

3.1.2. Definition and Characterization of Key Events

As previously stated, key events are defined as discrete instants within the movement cycle that delineate phases. Typically, three key events are identified for the contact phase, although their nomenclature varies among authors.

The onset of the contact phase and, consequently, the braking subphase is marked by initial contact, a key event also referred to as foot strike [51,52], touchdown [29,53], or ground contact [54]. Despite this nomenclatural variation, authors consistently interpret the concept. For the purposes of this review, initial contact will be used, defined as follows: that instant in which the foot contacts the ground, regardless of the portion of the foot that makes such contact [2,26,27]. Conversely, the termination of the contact phase and the propulsion subphase is named by toe-off [2], also termed take-off [25,52] or foot-off [55]. In this review, toe-off will be used defined as the following: that instant in which the foot leaves the ground [2,26,27].

The event exhibiting the greatest variability in interpretation is midstance, also referred to as midsupport. Unlike initial and terminal contact events, which are consistently interpreted across gait and running analyses, midstance is traditionally considered a phase within the gait cycle [56]. In contrast, within the running cycle, it was initially defined as a phase [3,57] but later reclassified as a key event [26,27]. Furthermore, the biomechanical characteristics of midstance differ significantly between gait and running. In gait, it corresponds to the maximum vertical displacement of the center of mass, whereas in running, it marks the point of minimum vertical displacement. This disparity underscores the necessity for distinct temporal analyses of these locomotor cycles [58]. An alternative and relevant interpretation of this event is its definition as the transition from braking to propulsion (T_B–P) based on ground reaction forces [37], which aligns more closely with the current proposal. Therefore, T_B–P is defined as follows: that instant in which one transits from the braking subphase to the propulsion subphase [2,26,37].

3.2. Methods for Identifying Key Events

Technological advancements have facilitated the development of increasingly sophisticated methodologies for the detection of key events, with kinetic and kinematic techniques being the most prevalent. However, the application of these techniques is contingent upon factors such as the study objective, evaluation environment, and availability of measurement instruments.

3.2.1. Dynamometry

Kinetic analysis using force platforms is often regarded as the gold standard for key event identification within the running cycle. However, its application is typically constrained to laboratory settings employing floor-mounted force plates or instrumented treadmills [59,60,61]. Key event identification is based on GRF variations relative to predefined thresholds. These thresholds are frequently determined empirically, influenced by evaluation conditions and data processing methodologies. GRF profiles exhibit differences between overground and treadmill running, necessitating higher thresholds for treadmill-based analyses due to increased signal noise from device vibrations [62,63]. Furthermore, threshold values may vary depending on the specific event being identified. For instance, initial contact thresholds are generally higher, reflecting the abrupt force increase, compared to toe-off, which is characterized by a gradual force reduction [59,64]. Table 2 presents a compilation of the vertical GRF thresholds utilized in the literature for initial contact and toe-off.

For the T_B–P, changes in the horizontal GRF along the anterior–posterior axis have been identified as the determining indicator. Consequently, a threshold value of 0 N is typically used, with negative horizontal GRF values indicating the braking subphase and positive values indicating the propulsion subphase [24,65,66,67].

Table 2. Identification of initial contact and toe-off through dynamometry.

Reference	Initial Contact		Toe-Off
	Over Ground	Treadmill	Over Ground	Treadmill
[68]	A positive change in the vertical ground reaction force greater than 1 N ms−1 (or 1000 Ns−1), below a threshold of 100 N
[69]	Above 10–20% of body mass		Under 10–20% of body mass
[70]	The first data point in which the vertical force value (N) increased to and remained greater than two standard deviations above the zero-load level
[59]	Above 10 N	Not available	Under 5 N	Not available
[71]	Above 20 N	Above 20 N	Under 20 N	Under 20 N
[63]	Above 10 N	Above 40 N	Under 10 N	Under 40 N
[72]	Above 30 N	Above 30 N	Under 30 N	Under 30 N

N: Newton; ms: millisecond.

Table 2. Identification of initial contact and toe-off through dynamometry.

Reference	Initial Contact		Toe-Off
	Over Ground	Treadmill	Over Ground	Treadmill
[68]	A positive change in the vertical ground reaction force greater than 1 N ms−1 (or 1000 Ns−1), below a threshold of 100 N
[69]	Above 10–20% of body mass		Under 10–20% of body mass
[70]	The first data point in which the vertical force value (N) increased to and remained greater than two standard deviations above the zero-load level
[59]	Above 10 N	Not available	Under 5 N	Not available
[71]	Above 20 N	Above 20 N	Under 20 N	Under 20 N
[63]	Above 10 N	Above 40 N	Under 10 N	Under 40 N
[72]	Above 30 N	Above 30 N	Under 30 N	Under 30 N

N: Newton; ms: millisecond.

While threshold values are commonly employed, they are susceptible to a wider range of influencing factors. Consequently, a seemingly more robust method is that utilized by Bezodis, Kerwin, and Salo [70], which establishes a threshold calculated as the sum of the mean baseline data plus two standard deviations. This approach defines initial contact as the first data point equaling or exceeding this threshold, and toe-off as the first data point equaling or falling below it. This methodology aligns with current methods for determining events and phases in jumping actions, differing primarily in the number of standard deviations added (two versus five) [73]. This method mitigates the impact of noise variations between devices, potentially enhancing the standardization of detection. Furthermore, normalizing force values by the subject’s body mass would be advisable to neutralize substantial differences in the initial force due to variations in body mass [69]. Given that during running the foot impacts the force platform (in contrast to jumping where the subject is static on the platforms), and considering that running speed influences the production of vertical and horizontal GRF [74], it is likely that more than two standard deviations may be required to improve detection sensitivity. Future research should consider contrasting data processing methods (number of standard deviations added and different digital filtering cutoff frequencies) [75] for event detection across various speeds of running.

3.2.2. Motion Capture Systems

The three-dimensional kinematic motion analysis has been extensively employed to assess running techniques. In scenarios where force platforms are unavailable, discrete kinematic indicators have been proposed as alternatives for key event identification. Additionally, predictive algorithms, considered gold standard alternatives, have been developed, primarily focusing on the detection of initial contact and toe-off. These algorithms have demonstrated progressively reduced error rates with advancements in instrumentation and analytical methodologies. However, their predictive accuracy can be influenced by factors such as runner-specific stride patterns, footwear, surface type, slope, and running speed.

Table 3. Identification of initial contact and toe-off through kinematics algorithms.

Reference	Conditions	Initial Contact		Toe-Off
		In-Puts	Outcome (ms/Frames)	In-Puts	Outcome (ms/Frames)
[76]	Overground Shoes RF Self-selected speed 180 Hz	Foot angular acceleration	RF: RMSE = 4.5/0.81	Foot angular acceleration	RF: RMSE = 6.9/1.24
[63]	Overground Barefoot RF and MF Self-selected speed 200 Hz	Position and acceleration of hallux and posterior heel distal	AP: ER = −55/−11 RF: ER = −55/−11 MD: ER = −55/−11	Position and acceleration of hallux and posterior heel distal.	AP: ER = −27.5/5.5 RF: ER = −27.5/5.5 MD: ER = −22.5/4.5
[77]	Overground Shoes RF, MF, and FF At 5.6 m/s 240 Hz	Vertical displacement between the heel and posterior superior iliac spines	AP: RMSE = 14.1/3.4 RF: RMSE = 5.9/1.4 MD: RMSE = 14/3.4 FF: RMSE = 22.4/5.4	Vertical displacement between the second metatarsal head and posterior superior iliac spines	AP: RMSE = 9.2/2.2 RF: RMSE = 6.8/1.6 MD: RMSE = 5.3/1.3 FF: RMSE = 17/4.1
[78]	Overground Shoes RF, MF, and FF At 3.7 m/s 200 Hz	Velocity of the pelvis center of mass	AP: RMSE = 6.5/1.3	Not applicable	Not applicable
[79]	Overground Shoes RF, MF and FF Self-selected speed (range 2–6 m/s) 250 Hz	Acceleration of the posterior aspect of the calcaneus and 1st metatarsal head	AP: RMSE = 8.3/2.1	Vertical jerk peak of the distal end of the hallux	AP: RMSE = 5.6/1.4
[80]	Treadmill Shoes RF, MF, and FF At 2.5, 3, and 3.6 m/s 200 Hz	Vertical acceleration of foot calcaneus and 3er metatarsal	AP 2.5 m/s: RMSE = 8.4/1.7 AP 3 m/s: RMSE = 8.2/1.6 AP 3.6 m/s: RMSE = 7.8/1.	Position of the 3er metatarsal	2.5 m/s: RMSE = 8.6/1.7 3 m/s: RMSE = 6.6/1.3 3.6 m/s: RMSE = 6.9/1.4
[61]	Treadmill (slope −9° to +9°) Shoes RF, MF, and FF Range of 2.5–5.0 m/s and self-selected 150–300 Hz	Anteroposterior velocity of the distal tibia, ankle angle, anteroposterior and vertical velocity of the foot center mass	AP: RMSE = 5/0.75–1.5	Anteroposterior velocity of the distal tibia, ankle angle, anteroposterior, and vertical velocity of the foot center mass	AP: RMSE = 6/0.9–1.8

RF: Rearfoot; MF: Midfoot; FF: Forefoot; AP: All patterns; ms: milliseconds; TE: True error; ER: Error residual; RMSE: Root mean square error.

presents algorithms that have been used for the identification of initial contact and toe-off.

Considering the error of the algorithms used, added to the sampling frequency used, the error level is low. Most of the methods present an error of less than three frames, being highly recommended for use in this area. Only the algorithm of Leitch, Stebbins, Paolini, and Zavatsky [63] demonstrates a greater error, 11 frames, compared to the others. Although when selecting an algorithm, the degree of error is an important factor to consider, along with the conditions under which this algorithm was validated, there are elements such as the running surface, running speed, use of footwear, foot strike patterns, as well as the inputs required for data processing.

Regarding the five algorithms validated for overground running, the algorithm proposed by Hreljac and Stergiou [76] is validated only for the rearfoot pattern, and by not declaring the running speed of the subjects, it is not easy to know the accuracy of the method at different speeds. Similarly, Leitch et al. [63] do not declare a self-selected velocity value, and it was also validated for barefoot conditions, so its application in subjects with shoes could alter the estimate. The algorithm is based on the velocity of the pelvic center of mass [78], and although it presents a low error, it can only be used for IC, limiting a complete analysis of the running cycle. Regarding the methods described by Smith et al. [77] and Handsaker et al. [79], both prove to be suitable for all gait patterns, so it is advisable to select one or the other based on the running speed to be analyzed; the first [77] would be more relevant for high speed races and the second [79] is more flexible, being suitable for slow, moderate, and fast speeds.

Regarding the two treadmill algorithms, the method presented by Patoz, Lussiana, Gindre, and Malatesta [80] is valid for all three footfall patterns at low to moderate speeds. However, the FootNet algorithm [61] has a superior adaptive capacity, complying with all three footfall patterns, a wide speed range, and even positive and negative slopes of almost ten degrees.

Based on the reviewed literature, no specific algorithm for the direct prediction of the T_B–P was identified. However, discrete kinematic indicators have been proposed for its estimation. An initial intuitive approach involved defining T_B–P as occurring at 50% of the total contact phase duration [25,81]. Nevertheless, evidence suggests that the temporal ratio between the braking and propulsion subphases averages approximately 40% and 60%, respectively [37,82]. Furthermore, during the initial steps of a sprint, the braking subphase constitutes a larger proportion of the total contact time during the acceleration phase, reaching 12.9% for the first step [83]. Other established indicators for T_B–P include the maximum vertical descent of the center of mass [2,84,85,86] and maximum knee flexion [87,88,89]. One study comparing maximum knee flexion and the horizontal and vertical center of mass displacements in five sprinters [37] revealed discrepancies in T_B–P identification, locating it at 51.7%, 45.8%, and 27.7% of the total cycle, respectively. Although future research is still required to empirically prove which is the most appropriate kinematic indicator to determine T_B–P, it is recommended that provisionally the maximum knee flexion and the maximum vertical descent of the center of mass be used as indicators for 2D and 3D motion analysis, respectively. Mainly due to the feasibility of each indicator in relation to the method of capture, since the peak knee flexion is easy to measure and has proven to be valid and reliable in 2D analysis [90], while the measurement of the center of mass can be difficult to identify by 2D analysis and from a sagittal plane, it represents a more accessible and reliable indicator for 3D analysis.

Despite this recommendation, again the need for future research is valid in relation to anteroposterior reaction forces (gold standard) for 2D and 3D motion analysis. Currently, indicators based primarily on marker position or item angles have been proposed; however, we recommend that indicators that consider velocity or acceleration be explored, as these may present greater consistency in terms of the subphases of braking and propulsion during the contact phase.

It is pertinent to note the existence of alternative techniques and equipment for key event identification, such as accelerometers and plantar pressure insoles, which demonstrate significant potential for precise application in future years. Similar to dynamometry and motion capture systems, these methodologies warrant their own comprehensive review due to the depth and specificity of the subject matter, which falls outside the scope of the present investigation. Nevertheless, a clear need persists for accurate key event identification using cost-effective equipment applicable in authentic training and competitive settings.

4. Practical Applications

The temporal structuring of phases and events in the running cycle can be understood as a theory within the field of human movement studies, serving as a means of structuring, constructing, and sharing knowledge about the phenomenon and study of running [91]. Theories often serve various roles in the study design, including their practical application. The main practical contributions are aimed at sports science centers, biomechanics laboratories, and technical and sports scientific teams, which can be used as an assessment tool to understand the phenomenon analytically [91].

The temporal structure model of the running cycle allows for the grouping of information by phases and events and can provide continuous or discrete variables in temporal terms, which increases the sensitivity of the analyses. On the other hand, precise definitions of events and phases provide important information that users can replicate, facilitating comparisons between the subjects tested and improving criteria for identifying the transition between phases. An example of this is the transition between braking and propulsion since the techniques used are diverse: (a) maximum knee flexion [92], (b) full foot support [93], (c) projection of the pelvis in front of the midfoot [94], or (d) maximum lowering of the center of mass or sacrum, indistinctly [95]. In addition, Ciacci, Di Michele, and Merni [37] demonstrated an 18% difference between identifying the braking–propulsion transition through the knee flexion technique and the maximum center of mass lowering technique.

Similarly, the analytical use of the temporal structure of the running cycle provides an interpretation of the results consistent with the proposed theory, where the analyses could achieve greater quality and clarity in the findings. The quality of the findings is related to using the proposed analytical model since the difficulties or deficiencies detected could be improved by prescribing physical exercise that fulfills the mechanical objectives described in each phase. This clarity in the events and phases could correct inconsistencies in some current analyses in some phases.

Example for the Application of the Model

In the context of high performance, the improvement potential of athletes is increasingly limited and high sensitivity in the analysis is required to detect real changes in their performance. The presented model allows a rigorous analysis of running technique, which would help to decrease the error associated with the evaluation of biomechanical parameters. A concrete example of this is the measurement of vertical, horizontal, braking, and propulsive relative impulse to analyze the performance of a sprinter. It has been shown that braking and propulsive relative impulse are performance indicators for the athlete’s speed [65]. Therefore, based on the definition provided for the key events and detection methods using dynamometry, initial contact, T_B–P, and toe-off can be accurately identified using the method proposed by Bezodis, Kerwin, and Salo [70]. Furthermore, as a result of this rigorous determination of events, the synchronized analysis of other techniques, such as electromyography, will complement the neuromuscular needs of the braking and propulsion subphase. Such precision in the analyses will allow identifying with greater sensitivity the improvements resulting from training interventions on relative impulse production and neuromuscular function, for the athlete’s performance.

5. Conclusions

Despite the acknowledged relevance of temporal structure concepts in running research, the integration of formalized models for phase/event identification with biomechanical analysis remains infrequent. This review succeeded in (i) reviewing the most relevant temporal models of the running cycle and their respective evolution; (ii) presenting a rigorous definition and determination of phases and key events based on the logical criteria that a motion analysis should possess; and (iii) reviewing and comparing the use of dynamometry and motion capture systems for the detection of key events.

The model presented delineates the running cycle into two primary phases: contact (subdivided into braking and propulsion subphases) and flight, demarcated by three key events: initial contact, braking–propulsion transition, and toe-off. Each phase, sub-phase, and key event is rigorously defined, emphasizing their respective contributions and interdependence within the running cycle. While leg swing is not considered a phase in this framework due to temporal overlap with other phases, its recognized importance in running mechanics warrants its integrated analysis under the proposed temporal phase delineation. The integration of kinetic, kinematic, and electromyographic data with the present model will facilitate the systematization of analysis, thereby improving the communication between theoretical and applied contexts. Such integration has the potential to not only refine interpretations of running mechanics but also to enable practical advancements in athletic training, injury mitigation, and performance optimization.

The precise delineation of phases and subphases in running analysis is critically dependent on the objective identification of key events. This process is significantly enhanced by the application of advanced technologies such as motion capture and dynamometric analysis. However, future research should prioritize the development of methodologies for key event identification using more accessible technology in authentic sporting environments. In particular, more research is needed to determine a valid and reliable kinematic method to identify the transition between braking and propulsion. Future research should delve into the systematization of frameworks, considering the specific sports context, available equipment, and various running modalities, and also explore key biomechanical performance indicators related to the running cycle.