Technological Solutions for Human Movement Analysis in Obese Subjects: A Systematic Review

Obesity has a critical impact on musculoskeletal systems, and excessive weight directly affects the ability of subjects to realize movements. It is important to monitor the activities of obese subjects, their functional limitations, and the overall risks related to specific motor tasks. From this perspective, this systematic review identified and summarized the main technologies specifically used to acquire and quantify movements in scientific studies involving obese subjects. The search for articles was carried out on electronic databases, i.e., PubMed, Scopus, and Web of Science. We included observational studies performed on adult obese subjects whenever reporting quantitative information concerning their movement. The articles must have been written in English, published after 2010, and concerned subjects who were primarily diagnosed with obesity, thus excluding confounding diseases. Marker-based optoelectronic stereophotogrammetric systems resulted to be the most adopted solution for movement analysis focused on obesity; indeed, wearable technologies based on magneto-inertial measurement units (MIMUs) were recently adopted for analyzing obese subjects. Further, these systems are usually integrated with force platforms, so as to have information about the ground reaction forces. However, few studies specifically reported the reliability and limitations of these approaches due to soft tissue artifacts and crosstalk, which turned out to be the most relevant problems to deal with in this context. In this perspective, in spite of their inherent limitations, medical imaging techniques—such as Magnetic Resonance Imaging (MRI) and biplane radiography—should be used to improve the accuracy of biomechanical evaluations in obese people, and to systematically validate less-invasive approaches.


Introduction
Obesity represents one of the most prevalent pathologies in developed countries [1], and the global prevalence of obesity has increased by nearly 80% since the 1980s [2].
Obesity has been shown to be a condition that can affect the ability to perform dailylife activities, such as walking [3][4][5][6], lumbopelvic movements [7], standing up [8,9], and tasks involving upper limb movements [10], thus significantly impacting the overall quality of life. In obese individuals, excessive weight can limit and alter their overall capacity for movement, leading to a range of functional limitations [11].
Hence, it is important to quantify and monitor how the obese population carries out specific movements. This information can help to characterize their functional limitations, and prevent further issues impacting their musculoskeletal system, due-for instance-to high compressive forces on the anti-gravitational joints and the muscular fatigue [12]. Therefore, there is growing interest in developing tools that are able to monitor the physical activity and movement of the obese population, so as to evaluate the limitations affecting their daily life activities and quantify their clinical conditions.
In this context, it is essential to introduce technological and methodological approaches able to reliably and safely provide quantitative information about the movements of obese subjects. Indeed, the analysis of human motion is widely adopted in clinical and research fields to investigate pathological conditions, including obesity, and pursue an objective and integrated assessment [13]. Specifically, human movement analysis is used to assess the biomechanical features (including both kinematics and kinetics) of obese subjects in order to better understand the most common problems affecting this population.
Currently, movement analysis-particularly motion tracking-is usually performed using marker-based stereophotogrammetric optoelectronic systems integrated with force platforms, and used within controlled experimental environments [13,14]. However, over the past decade, the availability of wearable equipment has made it possible to conduct examinations in more ecological and daily-life conditions [15][16][17][18]. For instance, magnetoinertial measurement units (MIMUs) can provide a quantitative evaluation of movement by integrating information from triaxial accelerometers, gyroscopes, and magnetometers; these solutions are, in general, non-invasive and easy-to-use, allowing for continuous monitoring and self-assessment to evaluate and prevent risk, even at home.
Unfortunately, the use of non-invasive technological solutions that are externally applied to the skin is highly susceptible to soft tissue artifacts (STA) [19], particularly when considering the obese population. This issue can be addressed by simultaneously monitoring the movement of skin markers/sensors and the underlying bone using imaging methods such as biplane radiography [20] and Magnetic Resonance Imaging (MRI) [21].
While numerous studies investigated the application of various technologies to movement analysis in the obese population are present in the literature, a systematic review is currently absent. Therefore, this study aimed to identify and summarize the existing literature, and highlight the different technological and methodological strategies used to evaluate movement analysis approaches within the obesity context, with a particular focus on gait and functional assessment. Based on the outcomes obtained from the reviewing process, this work also aimed to report any advantages or limitations regarding each technology, whenever available and discussed.

Information Sources
This study was conducted and reported following the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) 2020 [22]. A proper workflow was created following the PRISMA for systematic review protocols (PRISMA-P) [23] which was online, registered on Open Science Framework available at https://doi.org/10.17605/OSF. IO/72UAE (accessed on 5 March 2023). The search process was started in March 2022 by considering three different electronic databases, i.e., PubMed, Scopus, and Web of Science.
Three reviewers (M.R., R.G. and S.E.) were involved in the inclusion/exclusion process, following the eligibility criteria, as hereinafter reported; a fourth reviewer helped in case of discrepancy (L.N.F.).

Eligibility Criteria and Search Strategy
The research question was defined using the Patients-Intervention-Comparison-Outocome-Study (PICOS) approach [24]; in particular, papers were included if they were focused on observational studies reporting quantitative measurements (S); they were focused on obese population (P); they performed the assessment using specific human motion analysis techniques (I). Comparisons (C) and outcomes (O) were not defined.

Study Selection
Studies were included if they met the following inclusion criteria: • published from 2010 to 2022. • written in the English language. • conducted on adults (≥18 years old). • conducted on subjects with Body Mass Index (BMI) > 30. • reporting at least one measurement parameter/metric related to movement analysis (e.g., joint angles, joint moments, etc.). • based on observational study design.
Otherwise, the exclusion criteria were: • Systematic narrative and scoping reviews; letters to editors and commentaries; book or chapters; conference proceedings; case reports or case series. • Studies including population not stratified among normal-weight, overweight, and obese subjects. • Studies focused on pre-post intervention evaluation (e.g., arthroplasty, etc.). • Studies involving patients affected by chronic pathologies and/or patients in pain.

•
Studies reporting no information about the used systems/devices/instrumentations (e.g., model or manufacturer specification). • Studies focused on posture and balance evaluation.

Data Items and Collection
Identified data were exported from electronic databases and imported into a web application for systematic reviews (Rayyan [25]). All the information concerning the studies was manually extracted from the included papers thanks to a full-text analysis, exported, and collected in a custom database, created according to the Cochrane guidelines [26].
As summarized in Table 1, the following data were included: • Author and year of publication.

Quality Assessment: Risk of Bias in Individual Studies
Included papers were assessed using the methodological index for non-randomized studies (MINORS) [27] in order to quantify the individual and overall risk of bias. The MINORS tool comprises a total of twelve items, with items from eight to twelve used only for comparative studies. The items are scored 0 when they are not reported, 1 if they are reported but inadequate, or 2 if reported and adequate. The ideal overall score is 16 for non-comparative studies and 24 for comparative studies. The data extracted from the included studies were summarized qualitatively, providing a narrative synthesis of the findings.  Passive reflective markers were placed on the lateral aspect of the pelvis to represent the anterior surface of the palpated ASIS landmark. Rigid clusters of 4 non-collinear markers were firmly affixed on the sacrum, and bilaterally on the thigh, shank, and foot segments to minimize soft-tissue artifacts [35].   Retroreflective anatomical markers were placed on bony landmarks bilaterally on the acromion process, iliac crest, greater trochanter, medial femoral epicondyle, lateral femoral epicondyle, medial malleolus, lateral malleolus, and on the shoe above the first and fifth metatarsal heads and the second toe. For the tracking markers, 4 retroreflective markers, attached to thermoplastic plates, were placed on the posterior trunk, posterior aspect of the pelvis (2 marker clusters on each side), the lateral surface of thighs and shanks, and finally on the the heel of the shoe.
The ultrasound probe was placed anteriorly over the medial and lateral femoral condyles in the transverse plane and superior to the border of the patella.
Markers were placed on the lateral side of the pelvis such that the marker represents the anterior surface of the palpated anterior superior iliac spine (ASIS) landmark [48]. Rigid clusters of four markers were affixed to the sacrum, and bilaterally to the thigh, shank, and foot.
The ultrasound probe was placed anteriorly over the medial and lateral femoral condyles in the transverse plane, and superior to the border of the patella.
Markers were placed on the lateral side of the pelvis such that the marker represents the anterior surface of the palpated anterior superior iliac spine (ASIS) landmark [48]. Rigid clusters of four markers were affixed to the sacrum, and bilaterally to the thigh, shank, and foot. The exoskeleton was fixed on one of the subjects' lower limbs and was calibrated to define the anatomical frames of the femur and tibia relative to the technical frames of the exoskeleton. The anatomical frames were built using the functional approach developed by [52].
• Kinematics: movement of the femoral harness and tibial plate relative to the femur and tibia; knee flexion angle ( • ). Retro-reflective anatomical markers were placed bilaterally on the 1st and 5th metatarsal heads, the distal end of 2nd toe, medial and lateral aspects of malleoli and femoral epicondyles, greater trochanters, iliac crests, and acromion processes. A semi-rigid thermoplastic shell with four reflective tracking markers was placed on postero-lateral aspects of the posterior trunk, shanks and thighs, and mid-dorsal aspect of shoes. Four tracking markers placed on two separate shells were placed on the left and right posterior-lateral aspects of the pelvis.
• Spatio-temporal parameters: stair ascent, descent, and level walking Step Widths (m) and Speeds (m/s). • Kinematics: knee extension and abduction angle ( • ). • Kinetics: vertical ground reaction force (N), knee extension, and knee abduction moments (Nm).  Triads of infrared-emitting diodes were placed on the pelvis and trunk and bilaterally on the thighs, legs, and feet. Markers were affixed to the lateral aspect of the foot, the shaft of the tibia, and the lateral aspect of the thigh. Femoral epicondyle motion was tracked by two markers mounted on a custom femoral tracking device [60]. Pelvic markers were affixed on the sacrum using a 5-cm extension. A similar extension was placed on the lower cervical vertebrae to track the trunk segment.  Passive reflective markers were placed over the seventh cervical vertebrae, acromion processes, right scapular inferior angle, sternoclavicular notch, xiphoid process, 10th thoracic vertebrae, posterior-superior iliac spines, medial/lateral femoral epicondyles, medial/lateral malleoli, calcanei, first metatarsal heads, second metatarsal heads, and proximal and distal heads of the fifth metatarsals. To account for adipose tis-sue around the pelvis, virtual markers were placed on the anterior superior iliac spines and iliac crests using a digitizing wand (C-Motion, Germantown, MD). Marker clusters were placed on the thighs, shanks, and sacrum to aid in 3D tracking [44]. Markers were affixed to the lateral aspect of the foot, to the shaft of the tibia, and the lateral aspect of the thigh. Femoral epicondyle motion was tracked by two markers mounted on a custom femoral tracking device [66]. Pelvic and trunk marker triads were attached to 5 cm extensions with base plates affixed over the sacrum and lower cervical vertebrae.
• Kinematics: range of motion at the hip, knee, and ankle, and trunk segment flexion angles ( • ). • Kinetics: support moment (summation of the lower limb ankle, knee, and hip extensor moments, Nm/kg).   Anatomical markers were placed bilaterally on the iliac crest, greater trochanter, medial and lateral epicondyles, medial and lateral malleoli, and the first and fifth metatarsal heads. Four non-collinear tracking markers were attached to molded thermoplastic shells [75] on the pelvis, thighs, and shanks [30], and three separate non-collinear markers on the heels [76].  Marker locations included bilateral iliac crests, greater trochanters and anterior superior iliac spines. A sacral marker was placed on L5/S1. Other markers on the right leg only included medial and lateral femoral epicondyles, medial and lateral malleoli, first and fifth metatarsal heads, and the distal end of the first metatarsal. Rigid arrays of markers secured to the right lateral thigh, lower leg, and posterior heel tracked the motion of the segments. Five markers were placed along the spine: two on the thoracic (T1 and T6), two on the lumbar vertebrae (L1 and L3), and one on the sacrum (S1). • Kinematics: sagittal plane: forward trunk inclination, anterior pelvic tilt, angle related to lordosis, lumbar movement, angle related to kyphosis, thoracic movement. Frontal plane: lateral bending, lateral trunk inclination, pelvis obliquity, lumbar curve, lumbar movement, thoracic curve, thoracic movement, and shoulders. Symmetry index of lateral trunk inclination.

Study Selection
The initial search identified a total of 3720 papers; of these, 1281 were duplicates and were removed accordingly. The obtained 2439 studies were preliminary screened for eligibility criteria, and 2207 were eliminated. A total of 232 were then screened in full, and from the residual papers: thirty-two had full text not available; one had a different study design; thirty-seven had a population not stratified according to the inclusion criteria or with BMI < 30 kg/m 2 ; thirty-nine had a population with less than 18 years; twenty-three included a population affected by other pathologies or pain; nine were focused on the surgical intervention; two papers had no precise indication about the instrumentation; twenty-four papers were focused on posture or balance, or the analysis of plantar pressure; fifteen papers had other focuses; one paper was not written in English; six papers were different publication types from the inclusion criteria. Finally, a total of 43 papers were included in this review. The complete PRISMA flowchart is reported in Figure 1.

Study Selection
The initial search identified a total of 3720 papers; of these, 1281 were duplicates and were removed accordingly. The obtained 2439 studies were preliminary screened for eligibility criteria, and 2207 were eliminated. A total of 232 were then screened in full, and from the residual papers: thirty-two had full text not available; one had a different study design; thirty-seven had a population not stratified according to the inclusion criteria or with BMI < 30 kg/m 2 ; thirty-nine had a population with less than 18 years; twenty-three included a population affected by other pathologies or pain; nine were focused on the surgical intervention; two papers had no precise indication about the instrumentation; twenty-four papers were focused on posture or balance, or the analysis of plantar pressure; fifteen papers had other focuses; one paper was not written in English; six papers were different publication types from the inclusion criteria. Finally, a total of 43 papers were included in this review. The complete PRISMA flowchart is reported in Figure 1.
In the Supplementary Materials, a detailed description of the references and the reasons for exclusions are reported.

Risk of Bias
Focusing on the evaluation of risk of bias, Table 2 shows the scores specifically associated with the included articles and estimated through the MINORS tool [37].
Items 1 (clearly stated aim), 3 (prospective collection of data), and 4 (endpoints appropriate to the aim of the study) were graded as two in all forty-three studies. Item 2 (inclusion of consecutive patients) represented a critical item since all studies were graded In the Supplementary Materials, a detailed description of the references and the reasons for exclusions are reported.

Risk of Bias
Focusing on the evaluation of risk of bias, Table 2 shows the scores specifically associated with the included articles and estimated through the MINORS tool [37]. Table 2. List of the included papers assessed using the methodological index (MINORS) tool, to quantify the individual and overall Risk of Bias. Legend of items 1: A clearly stated aim; 2: Inclusion of consecutive patients; 3: Prospective collection of data; 4: Endpoints appropriate to the aim of the study; 5: Unbiased assessment of the study endpoint; 6: Follow-up period appropriate to the aim of the study; 7: Loss to follow-up less than 5%; 8: Prospective calculation of the study size; 9: Adequate control group; 10: Contemporary groups; 11: Baseline equivalence of groups; 12: Adequate statistical analyses. The items are scored 0 (not reported), 1 (reported but inadequate), or 2 (reported and adequate). The global ideal score is 16 for non-comparative studies and 24 for comparative studies. Items 1 (clearly stated aim), 3 (prospective collection of data), and 4 (endpoints appropriate to the aim of the study) were graded as two in all forty-three studies. Item 2 (inclusion of consecutive patients) represented a critical item since all studies were graded as one. Item 5 (unbiased assessment of the study endpoint) was also considered critical, and was graded as zero in all studies because no one mentioned strategies or methods to avoid bias concerning this issue. Items 6 and 7 were graded as zero in all forty-three studies because no studies had a follow-up evaluation in their study design.
A further critical item was represented by Item 8, where twenty-one papers were graded as zero, and one paper was graded as one. Item 9 (adequate control group) was graded as two when the control group was available, and the BMI stratification was used as a reference. Item 10 (contemporary groups) was graded as two in thirty-three studies, where subjects were collected for this study; this point was graded as one in five studies, where data were already available and derived from retrospective studies.
Furthermore, Item 11 (baseline equivalence of groups) was graded as two in twenty-nine studies, where samples were balanced between the study group and the control group; this item was graded as one in nine studies because the data were reported but considered not adequate. Finally, Item 12 (adequate statistical analysis), where twenty-three of thirty-eight papers were graded as one, and fifteen papers were graded as two.

Study Characteristics
Study characteristics are summarized and displayed in Table 1. Considering the population involved, five of forty-three papers included only one group of obese subjects without a control group [33,46,48,59,77]; six studies included three groups, represented by obese, over-weight, and normal-weight subjects [21,38,58,72,74,79]. The remaining thirty-two papers involved two groups, i.e., obese and normal-weight subjects.
Passive refracting markers were usually located according to standard protocols, such as "Davis", "Plug-in Gait", or "Hellen Hayes" [31,40,46,77], while different custom markersets were also proposed according to the different objectives. The relevant anatomical markers position for the gait assessment were iliac spines, greater trochanters, medial and lateral epicondyles, medial and lateral malleoli, lateral wands over the mid-femur and mid-tibia, medial and lateral knees, medial and lateral ankles, second metatarsal heads of the toes, and heels. Lerner et al., [44] proposed a market-set specifically designed for assessing movement in obese population, which was also adopted in [43,68,69].
In this context, Camomilla et al., [21] quantified the soft tissue displacement of pelvic landmarks during hip movements integrating the data gathered from the optoelectronic system and MRI, while Clement et al., [20] evaluated the effects of STA on the estimated 3D knee kinematics using biplane radiography.
Concerning wearable technologies, in five papers, MIMUs were specifically adopted, focusing on a single sensor [33,45] or using a 7-sensors protocol for assessing only the lower limb kinematics [49,55,58]. Pamukoff et al., [62] used electromagnetic tracking sensors, while Clement et al., [20] used an exoskeleton to estimate the kinematics of the lower limbs.
Surface Electromyography (sEMG) was further adopted in [41,44,68,69] to quantify the muscle activity and estimate the residual lower-extremity muscle forces in the obese population.
Concerning sEMG data, the root means the square of muscle activation signal during the gait cycle was evaluated in [41], whereas the peak of compression tibiofemoral forces and rate of tibiofemoral loading in [68], and the individual muscle forces in [44,69].

Discussion
In this systematic review, we identified and summarized the evidence provided by 43 original studies regarding the different technological and methodological strategies used to assess the human movement in obese subjects. To our knowledge, no systematic reviews on this topic were published so far.
Upon analyzing the literature and examining available technologies for human movement analysis, it appears evident that the majority of the adopted solutions are markerbased optoelectronic stereophotogrammetric systems. However, it is worth noting that most of the studies that use motion capture systems for gait analysis-even those directly assessing the impact of obesity-use standard kinematic marker-sets or methodologies developed for non-obese individuals, thus not accounting for adiposity; these approaches were, namely, "Plug-in Gait", "Davis", and versions of the "Helen Hayes" marker-set [32,36]. Although the literature confirms that it is extremely difficult to identify anatomical landmarks in obese individuals by using a palpatory approach [81], markers are usually placed on the skin trying to guess the placement in correspondence with the underlying anatomical references of interest [82], but without quantifying the error or estimating the possible impact on the final outcomes. In fact, the kinematic data derived from a (skin) markerbased motion capture system are extremely vulnerable to soft tissue artifacts (STAs) [19], which are caused by the relative movement of the tissues over the underlying bone reference [21]. It is evident that obese subjects are more sensitive to STAs, due to the presence of adipose tissue, which can thence increase the error in estimating the overall kinematics. In order to mitigate this issue, several researchers used clusters of markers fixed to rigid plates, aiming at reducing the effect of skin movement by limiting relative marker motion [21,28,29,34,35,37,42,44,47,48,64]. Furthermore, a study presented an obesity-specific motion capture methodology that used a spring-loaded digitizing pointer to manually mark the ASIS landmarks and an additional marker on the iliac crest, which are used to define the pelvis segment [44]. However, the massive amount of STAs characterizing obese populations warrants further investigation of the overall accuracy of marker-based approaches, depending on the specific tasks and setup.
To further counteract this problem, several studies proposed medical imaging approaches oriented to quantifying STAs and their effects and impacts on the kinematics of obese individuals. Among these studies, one was specifically focused on the use of MRI for studying STAs at the pelvis [21], which indeed represents the most critical segment due to the presence of the waist adipose tissue; however, the analysis was conducted on a small cohort of subjects who were classified as obese. Another research specifically analyzed knee kinematics during squatting while the subjects wore an exoskeleton, and STAs were estimated via biplane radiography [20]. Analyzing these papers, we can affirm that medical imaging approaches indeed returned the most reliable information directly focused on bone motions; for this reason, they can be considered as the basis to validate other methodologies. On the other hand, they are in fact invasive (e.g., ionizing radiation), difficult to use due to their inherent complexities, expansive, and bulky to be used in daily inpatient/outpatient setups or during daily clinical practice. Furthermore, to provide reliable information, these approaches require biomechanical subject-specific models, which require additional resources to be defined and reliably implemented. Moreover, it is important to highlight that, at present, the studies that used medical imaging techniques for assessing obese individuals were primarily focused on the pelvis and the knee joint; therefore, it is not possible to simply generalize the approach to other body joints or conditions. Wearable sensors represent a viable solution to perform movement analysis since they are less expensive and more versatile, with respect to the marker-based optoelectronic stereophotogrammetric systems. Furthermore, they can be applied outside the laboratory in free-living conditions, and can be used in an unrestricted area even for a long acquisition time [15][16][17]. Among wearable sensors, magneto-inertial measurement units (MIMUs) are the most promising ones [83,84]. However, from this systematic review, only five papers employed MIMUs in their study protocol so far. Gait was evaluated using either a single-sensor approach, that was previously validated in healthy and pathological individuals [33,45], or by deploying a 7-sensor setup, specifically addressing lower limbs [49,55,58]. Further, Pamukoff et al. [62] adopted electromagnetic wearable sensors to compare gait biomechanics between normal-weight and obese young adults. In addition to STAs, one of the main sources of uncertainties related to the use of wearable sensors is the presence of kinematics "crosstalk". This phenomenon can also affect marker-based systems and is linked to the error that can be made in the definition of the rotation axes around which the kinematics are decomposed (e.g., for the knee, the axes of flexion/extension, adduction/abduction, internal/external rotation) [85]; therefore, all the joint angles estimated out-of-sagittal-plane should be carefully interpreted. Moreover, wearable devices are also sensitive to STAs, which were specifically addressed by fixing the sensors over bones instead of muscles and using elastic bands and medical tape [49].
Most of the selected papers investigate the effect of obesity on biomechanics outcomes compared to a control group; unfortunately, only a few studies proposed a coherent approach for the evaluation of motion in obese people including a proper validation. Specifically, on this very topic, Lerner et al. [44] proposed an obese-specific marker-set and compared the output with that obtained from a standard marker-set designed for a normal-weight population. Staying on this subject, Horsak et al. [48] investigated whether test-retest reliability for three-dimensional gait kinematics in a young obese population is affected by the identification of the hip joint center; in particular, the authors assessed either a predictive or a functional hip joint center localization approach. Finally, Agostini et al. [55] validated a gait analysis system based on magneto-inertial sensors, both in normal-weight and overweight/obese subjects; the validation was performed with respect to a reference multichannel recording system providing direct measurements of joint angles in the sagittal plane through electrogoniometers, which indeed present inherent limitations.
From a deeper analysis of the included paper, we identified two main categories of limitations; the former is related to the population specifically involved in the studies, whereas the latter concerns the use of suitable technological solutions.
Concerning the first limitation, most of the evaluated studies involved in fact both males and females introducing a source of potential variability. In fact, obesity modifies the body geometry by adding mass to different regions and dissimilar fat distribution in males and females, and this could produce gender-related effects [33]. Moreover, the current literature did not consider that different body shapes (e.g., apple-shaped vs. pearshaped) may change the biomechanics output [86]. Another limitation highlighted in the included papers is the absence of a stratification of the participants in terms of the severity of obesity. The lack of homogeneous groups reduced the generalizability of the results, making difficult the comparison among groups since it is not possible to reliably identify which one of these confounding factors directly influenced the measured parameters.
Second, in most of the analyzed articles, the authors focused on the analysis of different biomechanical properties characterizing the obese population, but they did not address the reliability of the technology specifically used to obtain those parameters/metrics. In fact, over 13 years of analyzed literature, only three papers proposed solutions for studying the movement of obese individuals and a proper validation of them; the majority of included papers only narratively referred to some specific limitations that have been noticed during the experimental procedures, but did not report quantitative information and/or possible strategies to mitigate them.
However, we are aware that these limitations could be ascribed to the lack of a noninvasive reliable method for the measurement of the movement specifically designed for obese individuals. Indeed, future studies should consider the use of optimization methods for reducing the influence of STAs, as well as suitable procedures for biomechanical modelling to properly mitigate the errors possibly associated with sensors/markers placement.

Conclusions
Obesity inherently requires the need to quantify the functional limitations due to the critical physical conditions. Indeed, the movement analysis through marker-based optoelectronic stereophotogrammetric systems is the most widespread approach in specialized laboratories; however, although it is considered reliable, this solution is particularly affected by soft-tissue artifacts. In recent years, wearable systems have been representing a viable solution to monitor the movement of obese subjects even at home during their daily activities, but the kinematics in the frontal and transversal plane should be carefully interpreted due to the presence of crosstalk. Soft tissue artifacts and kinematic crosstalk are still debated topics in human motion analysis literature, especially for obese subjects. The general problem related to the STAs issue could be addressed by employing medical imaging techniques, which can represent the basis for the validation of more agile solutions. In fact, the proposed medical imaging approaches are expensive, difficult to use due to their inherent complexities, bulky, and resource-consuming.
From this systematic review emerges the urgency for further research to account for the effects of subcutaneous adiposity on kinematic data collection and analysis, and the need for dedicated, feasible, and reliable solutions that may improve the accuracy of the measurements in the obese population.