A method and system of generating a design for a footwear component

Abstract

A computer-implemented method of generating a design for a footwear component, the method comprising: obtaining motion data derived from monitoring motion of a user; calculating, according to the motion data, a biomechanical model of at least a portion of the body of the user; and generating, according to the biomechanical model and a first criterion, a footwear component design comprising one or more footwear parameters, each footwear parameter defining a physical characteristic of the footwear component.

Description

[0001] A METHOD AND SYSTEM OF GENERATING A DESIGN FOR A FOOTWEAR COMPONENT

[0002] FIELD OF THE INVENTION

[0003] The present invention relates to a method and system generating a design for a footwear component for a user.

[0004] BACKGROUND

[0005] Footwear can have a substantial influence on foot function, mobility, quality of movement, comfort, fatigue, human performance, injury risks and other pathologies. Selecting the right footwear, footwear components or footwear modifications is of particular importance for individuals experiencing pain, those at high risk of injury or falling, or with specific disabilities or pathologies such as diabetics, Alzheimer’s, Parkinson’s, or for applications with specific performance needs such as military, hiking, mountaineering, first responders, athletes. Appropriate footwear can also be of benefit to improvements in healthy working conditions, and for day-to-day comfort and health.

[0006] For athletes and sports applications, for example running, properly performing footwear will provide a more comfortable run and can positively influence the risk of injury and improve running efficiency, speed and performance. In contrast, an ill-fitting shoe can cause a number of foot and leg injuries such as nerve impingement, tendonitis, heel pain, stress fractures and sprains, and can further inhibit running efficiency, speed and performance.

[0007] Typically, footwear is sold in a variety of pre-determined sizes and are often chosen based on the length and width of the user’s foot. Although there are standardising sizing methods in the majority of countries around the world, shoes still vary in shape and size between manufacturers and styles, and differences in materials can make footwear stiffer or more flexible or software or harder, properties that can also contribute to the feeling of fit, making it a complex process to determine the correct fit of the footwear for the user. Moreover, it is even more complex to select a shoe compatible with the user’s particular movement style, or biomechanical function, and often users are not aware of all aspects of their movement patterns or biomechanics, or which footwear design best complements these personal attributes. Even when the user is aware of their correct fit and footwear performance needs, they are often forced to compromise with a sub- optimal fit due to the categorisation of footwear products available in the market, unaccounted-for factors in the design of the footwear such as the specific volume, shape and contours of the user’s foot and the high complexity associated with compound material and mechanical effects that make overall performance highly difficult or impossible to predict or understand for the majority of individuals. Whilst combining commercially available categorised footwear types, such as combing a production shoe with an additional insole can help to adjust, personalise or customise fit, feel and performance, the compound effects of the different footwear types (e.g. adding an insole) is also very complex and difficult or impossible to accurately predict or observe. Current state of the art systems available on the market include 3D scanning technologies, based on imaging, that are able to quickly measure and render the 3D dimensions of a human foot and in some cases, additionally can combine data from scans of the internal dimensions of a shoe when it is not worn on the foot. The goal of such systems is to overcome some of the aforementioned challenges that exist in the current standardised footwear sizing approaches. In particular these systems attempt to address the issues associated with variations between different footwear manufacturers and styles, and to further account for some of the other unaccounted-for factors in the design of the footwear such as the specific volume, height, shape and contours of the user’s foot. A key limitation of this technology is that the foot is scanned whilst un-shod, and typically in a standing position in which the user is statically positioned with weight relatively evenly distributed across both feet. In practice footwear is used in many more conditions beyond static standing, and the foot shape and associated fit and performance of the footwear are greatly affected by the position, articulation, loading and further dynamics of the structures of the body and foot, and physical and mechanical properties of tissues and materials. For example by simply standing on one foot instead of distributing body weight across two feet, all forces and loads and balance of the body must be managed by the single point of contact, this will mean that the joint angles are affected as the foot needs to be positioned more centrally under the centre of mass of the body to maintain balance, potentially placing the foot in a more pronated position. Additionally, the muscles and connective tissue will be placed under different stresses, strains, forces and torques, that can result in, for example, the metatarsals becoming more widely spread, or the medial arch of the foot to become more flattened due to the higher load borne by these structures. The individual may also utilise different neural-motor control to activate muscle groups and reposition body parts, for example spreading the toes so that the large toe is further displaced creating a wider base upon which the user can balance. This simple example is further exacerbated in case where there are more dynamic motions, for example, the forces and loads will become much greater during walking, running and jumping than simply standing, and to facilitate movement, the foot can further articulate, for example, the toes can flex and bend, together with the metatarsal joints as the heel is lifted from the ground, but the forefoot remains in contact to the ground. This not only changes the shape and form of the foot, and forces acting upon it, but also the footwear that is in contact with and attached to the foot. The materials of this footwear must also bend, stretch and generally deform to allow and facilitate the movements of the body, thereby greatly influencing the form, fit, physical and mechanical properties and overall performance of the footwear. Furthermore the footwear materials that are in contact with and attached to the foot, will apply forces and pressure that can influence the shape, form, physical and mechanical properties of the foot itself, for example, a shoe with a stiffer tighter upper material, with laces fully tightened, can compress the toes and metatarsals, resulting in a much narrower, flatter, foot form than is the case in a barefoot condition required for safe scanning of the foot. These 3D state of the art scanning technologies, therefore are not able to fully address and account for critical factors that can greatly influence footwear selection, function, and design.

[0008] Currently, if a user wants to improve their performance or reduce their risk of injury, a human coach or a trainer may recommend the shoe or footwear most suited to the user based on observations of the user running. They carry out additional tests on the user by employing purpose-designed software which can track the movement of the lower leg, ankle, and foot of the user. From this, the coach may be able to understand, for example, portions of the foot which need more support and may identify a specifically shaped insole to the user which will improve their running style. However, this identification is often based on experience and intuition, and is highly difficult or impossible to identify all anatomical and biomechanical aspects of the foot with the human eye or to precisely characterise the movement, running or walking style of the user. Particularly when a user is wearing enclosed footwear the foot is not directly observable. Additionally many contributing factors are not visible to the human eye, which makes it particularly hard for a coach to identify footwear with the physical characteristics that would improve performance and reduce injury of the user. Without the use of additional sensors and software it is not possible, for example, for a coach or expert to accurately observe or estimate the contact time, flight-time, forces within the knee or hip, foot angles etc. or to accurately assess the compound influences of all contributing factors.

[0009] Even in the case that sensors are employed, it is likely that only non-invasive sensors will be easily utilised in normal day to day practices, limiting the parameters or metrics that may be measured to those that can be directly monitored using conventional non-invasive sensors such as contact time, flight time, pressure under the foot, etc. These non-invasive sensors are not able to directly measure, for example, forces in the knee, the elastic energy stored in the Achilles tendon, or any other data regarding internal mechanics of the body. This means, that in practice, even with sensors employed to gather external measurements of the body, the resulting numbers still require human expertise to make an estimate, based on extensive experience or intuition, for which footwear would bring benefit to a user

[0010] Even in a laboratory setting, in which far more detailed data can be collected, potentially including more invasive sensors, the resulting data requires extensive processing to be able to interpret into human movement KPI’s and ultimately still relies on human expertise to translate further into a footwear recommendation, which inherently takes a lot of time that may be too slow for many potential applications such as rapid manufacturing / 3D printing of footwear during a patients visit to a clinic, or to make footwear selection at a city centre sports retail store. In day to day, mobile or remote locations, a laboratory is also not available or feasible, and in many applications relevant specialised human expertise is not readily available.

[0011] Therefore, it would be desirable to have a system and method able to generate a design for a footwear component that can positively influence human performance, injury risk, foot function, mobility, quality of movement, comfort, fatigue, and mitigate complications or issues associated with other relevant pathologies such as diabetes, brain diseases, deformities etc. In the case of an athlete, movement efficiency or performance and injury risk are particularly relevant.

[0012] SUMMARY OF INVENTION

[0013] In accordance with a first aspect of the invention there is provided a computer- implemented method of generating a design for a footwear component, the method comprising: obtaining motion data derived from monitoring motion of a user; calculating, according to the motion data, a biomechanical model of at least a portion of the body of the user, wherein the biomechanical model is representative of a biomechanical load distribution through at least the portion of the body of the user and wherein the portion of the body of the user comprises one or more anatomical structures of the body other than a foot of the user; and generating, according to the biomechanical model and a first criterion, a footwear component design comprising one or more footwear parameters, each footwear parameter defining a physical characteristic of the footwear component.

[0014] Advantageously, the method includes obtaining and using motion or force data that contains an indication or measure of the motion or forces of a part, or parts of the body of a user performing motion or during movement or whilst executing specific actions or activities to infer or estimate the forces to which the body of the user is subjected while they are performing motion, and how those forces are distributed throughout the body, and in particular the musculoskeletal system, of the user. These specific actions or activities may include any one or more of locomotion, jumping, changing direction, squatting, weight training, rehabilitation, sporting activities such as football, basketball, tennis or golf or any other activity which requires human motion. As such, the term “motion” refers to the act of moving or the state of being in motion. In some examples, motion comprises a cyclic or repeated action such as locomotion including walking, running, swimming, cycling, jumping or squatting, cutting, weight transitions or kicking. In other examples, motion comprises a chaotic or sporadic action such as those exhibited by a user when playing basketball, football, golf or tennis. Motion of any one or more body part for example motion of the foot, leg, knee, hip, lower / upper back, spine, neck, head, shoulder, arm, torso or hand. Motion may include any linear, rotational, or oscillatory motion carried out by the user. Linear motion refers to movement in a straight line, while rotational motion refers to movement around an axis, and oscillatory motion refers to back-and-forth movement around a fixed point. The motion can be performed as part of an action, such as the specific movement or series of movements of a part or parts of the body.

[0015] The motion data is used to infer or estimate a biomechanical model of the user. The biomechanical model and a first criterion, which may be an improvement criterion representing or corresponding to a targeted improvement in the motion or action or activity of the user, are used to generate a design for a footwear component. The generating of the footwear component design may be performed in accordance with the first criterion only, or, in some embodiments, may be performed in accordance first and second criteria, or any number of further criteria, according to the requirements for the design. In some cases, an improvement criterion is used, as alluded to above. In this way, the improvement criterion may be a condition on which the design is generated, and the footwear component may be configured such that in use, the motion of the user is improved so as to achieve the targeted improvement criterion. For example, the improvement criterion may correspond to a primary objective of the user such as increased mobility, lower energy cost, or injury risk reduction. In one example, the improvement criterion may relate to a targeted improvement in running economy (or speed for the same effort) for the user, such that a design for the footwear component may be generated that would provide support to the appropriate muscles and joints to avoid unnecessary, potentially energy wasting directions of motion (e.g. knee or hip moving out laterally), and promote or improve the energy return effect from the footwear component such that the runner can reuse energy from initial ground contact in propulsion as the foot leaves contact with the ground in the direction of motion, similar to the action of a spring / dampener system, tuned to resonate in phase with the action of the user.

[0016] In another example, the improvement criterion may relate to a targeted improvement in jump height achievable by the user, such that the footwear component may be generated so as to support to the appropriate muscles and joints to avoid unnecessary, potentially energy wasting directions of motion (e.g. knee moving inwards medially), whilst facilitating improvements in range of motion for critical joints and promote activity and contributions from more, or stronger muscle groups (e.g. adjusting heel-toe drop to increase range of motion around the ankle, increasing the stretch and contribution of associated connective tissue and tendons, such as the Achilles, increasing the activity and contribution of the calf, (primarily gastrocnemius and soleus muscles) and promoting or improving the energy return effect from the footwear such that the user / jumper can reuse energy from the initial counter movement in propulsion as the foot leaves contact with the ground in the direction of motion, similar to the action of a spring / dampener system, tuned to resonate in phase with the action of the user.

[0017] In some examples, the improvement criterion may be related to a targeted improvement in performance or efficiency achievable by the user, such that the footwear component may be generated so as to utilise the energy of the user more efficiently. In some examples, the improvement criterion may be related to an improvement in safety or injury index, such that the footwear component may be generated to take into account factors which contribute to injury and safety of the user while running or performing activities. The footwear design comprises one or more footwear parameters, each footwear parameter defining a physical characteristic of the footwear component. The physical characteristic may be a physical or mechanical component of the footwear component with specific physical properties or mechanical properties or material properties defined by the parameter and configured such that the footwear component is adapted to according to the first criterion which, in some examples, is an improvement criterion. In some examples, the footwear component is adapted to modify or improve the motion of the user according to the first or improvement criterion.

[0018] The physical properties of the physical characteristics may refer to attributes, features or properties of the physical characteristics. Examples of physical properties include colour, density, size, shape, weight, structure, etc. Mechanical properties are a subset of physical properties that describe how a material, or combination of materials behave when subjected to external forces or loads. These may include strength, toughness, elasticity, plasticity, stiffness, Youngs modulus, shore, roughness, co-efficient of friction, and ductility of the material. Material properties are a further subset of physical properties that describe the types of materials or combination of materials that may be utilised within footwear, for example, plastic, foam, carbon fibre, glass fibre, fabric, nylon, rubber, silicon, piezo-resistive / electric, wood, cotton, algae, etc.

[0019] In some examples, the footwear component is adapted to modify motion of the user according to the first criterion.

[0020] In some examples, the modified motion of the user is an improved motion of the user. Improved motion may refer to a change to the motion of the user such that the adjustment in their motion closer meets the first criterion or improvement criterion. Improved motion can refer to improving the form of motion when the user is performing for example an action. For example, motion may refer to an activity such as walking, running, jumping, cycling, squatting, lunging, cutting, weight transitions, kicking, jumping etc. such that a change to the motion of the user relates to an adjustment in the motion of the user such that the motion of the user closer meets the improvement criterion. Preferably the footwear component may be adapted to modify at least one of: a spring coefficient according to the first criterion; a damping coefficient of the user according to the first criterion; a Q factor of the user according to the first criterion; a biomechanical loading of the user according to the first criterion; an efficiency of the user; and an injury risk of the user according to the first criterion.

[0021] In some examples, the biomechanical model is representative of a biomechanical load distribution through at least the portion of the body of the user. The term “biomechanical load distribution” may be representative of one or more of: load values, force values, positional information, directional information and acceleration information, and may be associated with at least the portion of the body of the user. It may further, or instead, represent any one or more of: a stiffness associated with at least the portion of the body of the user, a dampening value associated with at least the portion of the body of the user, and any other value which may be associated with at least the portion of the body of the user. Therefore, the biomechanical model may represent forces and / or stresses within the user body, and preferably how such forces are distributed or spread within or across the said body portion, one or more regions of the body, body parts, or various anatomical structures of the user within the plurality of the body during movement.

[0022] In this way, the motion data is used to infer or estimate the forces to which the body of the user is subjected while they are performing motion, and how those forces are distributed throughout the body, and in particular the musculoskeletal system, of the user, and how the part or parts of the body experience or are exposed to these forces and / or how much energy is lost / used / dissipated or returned from the part or parts of the body during the motion, action, activity or during locomotion. In this way, a biomechanical load distribution may comprise data representative of the distribution within the portion of the body of the user of forces exerted upon or by the body as a result of the motion of the user. In particular the biomechanical load distribution may comprise estimates of the forces, stress, strain, moments, or torques, distributed throughout the body. In some examples, the biomechanical model is representative of a stiffness associated with at least the portion of the body of the user. A stiffness of a user may represent the extent to which a part or parts of the body of a user may resist deformation in response to an applied force, and in particular the forces to which the user is subjected to while they are in motion. Therefore, the term stiffness may refer to the rigidity or malleability of the part or parts of the body of a user, determined by the composition and properties of their musculoskeletal system.

[0023] In this way, the motion data can be used to infer or estimate the forces to which the body of the user is subjected while they are performing motion and how those forces are distributed throughout the body, along with a length measurement, of for example a leg, and a centre of mass of the user, or of parts of the user, to model the stiffness of the at least the portion of the body of the user.

[0024] In some examples, the stiffness of the user may comprise data representative of a spring constant for at least a portion of the body of the user as a result of the motion of the user. A spring constant is typically a measure of the stiffness of a spring. In one example where the stiffness comprises data representing a spring constant, the stiffness is modelling the action, motion or locomotion with a springmass system.

[0025] In some examples, the stiffness of the user further comprises data representative of a damping constant for at least a portion of the body of the user as a result of the motion of the user. In some examples, the stiffness of the user may comprise a spring constant and a damping constant. In some examples, the stiffness may be modelled using one or more springs and / or one or more dampeners, such that the spring constant and damping constant are representative of the resulting spring constant of the one or more springs and the damping constant of the one or more dampeners respectively.

[0026] Therefore, the stiffness can be calculated by adding complexity to the spring model with a dashpot, creating mass-spring-damper model (MSD), so as to increase the accuracy of estimating the stiffness. In some examples, a single-body MSD is used comprising a mass attached to the ground by both a spring and a damper. An external force is applied on top of the mass, which represents the weight of the person, and any additional mass born by the person such as a weight lifting bar, a backpack, clothing, package or any other form of mass that may be worn or carried by the person.

[0027] The stiffness may be modelled using various combinations of springs and / or dampeners in series, parallel or any combination thereof.

[0028] In some examples, calculating the biomechanical model may comprise calculating values for the magnitude and direction of ferees exerted upon a plurality of parts of the body of the user, based upon the motion data and using a computational mechanical model of the body. Further, calculating the biomechanical model preferably including the biomechanical load distribution and / or stiffness may comprise calculating values for the magnitude and / or direction of force, pressure, stress, strain, motion, moments and torques exerted upon a plurality of parts of the body of the user.

[0029] The plurality of body parts may or may not include a foot of the user. However, as alluded to above, the model typically comprises information pertaining to a nonfoot body part, that is the said portion of the body typically includes body parts, anatomical structures, or portions thereof, that are not part of a foot of the user. Therefore, in some examples, the said portion or the plurality of body parts may be outside of the foot and / or separate from the foot. As such, there are examples wherein the portion or the plurality of body parts extends or is located beyond the foot or feet of the user, typically such that the biomechanical model calculates an inference, prediction, or estimate offerees distributed for the portion or the plurality of body parts beyond / outside or other than the foot.

[0030] In other words, the said portion of the body of the user is, at least partially, preferably wholly, typically outside of or distal to the foot. Thus at least some of the modelled portion may be above the ankle joint, or the point at which the foot articulates with the lower leg.

[0031] The portion of the body, or the plurality of body parts may include any one or more of head, neck, one or two shoulders, one or two arms, one or two elbows, one or two hands, one or more fingers, chest, abdomen, back, hips, buttocks, one or two legs, one or two knees, one or two ankles. In some examples, the plurality of body parts may include any one or more of a tibia bone, a fibula bone, a femur bone, a knee joint, a hip joint, an intervertebral joint, a sacroiliac joint, an upper leg, a lower leg, a lower back, an upper back, an Achilles tendon, a patella tendon, a cruciate ligament, a collateral ligament, a meniscus, a joint capsule, a cartilage, a muscletendon unit, a bone, a joint, a ligament, a connective tissue, a body segment, a bicep, a tricep, a forearm, a pectoral muscle, a pelvis, a spine, a gluteal muscle, a thigh, a hamstring muscle, a calf muscle or any other part of the body of the user. Furthermore, an inference or estimate of the forces distributed through, not only a foot of the user, but also an ankle, a calf, a knee, a thigh, lower back and / or any other part of the body of the user may be calculated so as to provide a biomechanical load distribution for this collective portion of the body based upon the motion data and using a computational mechanical model of the body. Similarly, a stiffness can be calculated for this collective portion of the body. Therefore, the biomechanical load distribution and / or stiffness may be calculated for any portion of the body of the user, comprising one or more anatomical structures.

[0032] When calculating a biomechanical model of the user, it is useful to consider both the shoe the user is wearing as well as the body of the user. This is because footwear is the contact point between the foot of the user and the ground, such that it directly influences the joint angles, contact moments, forces and pressure exerted on or by the body muscles and joints. In this way, it is advantageous to include the shoe when modelling the mechanics and biomechanics of the user to better represent the biomechanical model of the user in motion. Therefore, in some examples the biomechanical model is calculated including and / or in accordance with the footwear component design and associated mechanical properties. In some examples, the biomechanical model is a mechanical model that comprises at least the portion of the body of the user and the footwear component design. Therefore, while the “biomechanical model” may refer to a mathematical and / or computational representation of the mechanical aspects of the biological system of a user, in some examples, the biomechanical model may additionally comprise one or more mechanical aspects of an article on or worn by the user, typically such that the article affects the motion of the user and / or the forces within the user body. The article may comprise a footwear component. In such an example, the biomechanical model comprises the mechanical aspects of the biological system of the user and the mechanical aspects of the footwear component on the user. When the biomechanical model comprises the mechanical aspects of the footwear component on the user, the biomechanical model may be according to the anatomical and physiological aspects of the user and according to properties of the footwear component on the user. The biomechanical model therefore can comprise an inference, prediction, or estimate the forces distributed throughout both the body and footwear component, or forces through the combined system of a portion of the body and footwear component

[0033] In some embodiments, the method further comprises providing the footwear design to the user. Preferably, the providing the footwear design to the user comprises at least one of: displaying the generated footwear design to the user; providing an audible description of the one or more of physical characteristics of the footwear design to the user; and providing a written description of the one or more of physical characteristics of the footwear component to the user. Preferably, the step of providing the footwear design to the user comprises at least one of: displaying the generated footwear design to the user; providing an audible description of the one or more of physical characteristics of the footwear design to the user; and providing a written description of the one or more of physical characteristics of the footwear component to the user.

[0034] Preferably, displaying the generated footwear design to the user comprises displaying at least one of a two-dimensional, 2D, visualisation of the footwear design to the user; and a three-dimensional, 3D, visualisation of the footwear design to the user. The footwear design may be provided to the user when they have completed an activity, motion or session in motion. A session may comprise the user carrying out one or more gait cycles by the continuous pattern of walking or running, or alternatively by carrying out a specific and continuous or non-continuous action such as jumping, lunging, cutting / changing direction, balancing, squatting, accelerating, decelerating, etc. Activities, motions or sessions in motion may be completed under a variety of conditions and in a variety of environments such as carrying a back-pack, holding weights, wearing different clothing, etc., in warm, cold, wet, dry weather, etc., and on different terrain or surfaces, such as grass, concrete, sand, rocks, etc. In this way, the footwear design may be provided to the user virtually immediately after they have finished a session.

[0035] In some embodiments, the session may comprise multiple actions, activities, motions or locomotion types, from which specific actions can be characterised to identify and isolate relevant actions for the improvement criteria and footwear. Optionally the generated footwear design will attempt to optimise the improvement criteria for multiple actions, either by averaging, or through weighted criteria to mathematically prioritise specific actions and / or improvement criteria for one or all actions.

[0036] In one example, a basketball shoe is likely to be used for multiple actions, activities or locomotion, the design of the shoe can be optimised for an improvement criteria of increased jump height during jumping, but it may also be desirable to optimise the design for an improvement criteria of increased maximum running acceleration / speed during running, and / or for an improvement criteria of quicker or shorter changes in direction / cuts / decelerations during running. Multiple actions may share the same improvement criteria (e.g., more efficiency during running and walking), and a single action may have multiple improvement criteria (e.g. higher efficiency and lower injury risk when running). In these cases the design elements / footwear parameters will be optimised to meet the improvement criteria(s) for all actions, and in the case that a design change will compromise one improvement criteria for the benefit of another, the design will be balanced to limit the compromise / benefit equally, or will prioritise the improvement criteria or action considered to be most important, either predetermined or as selected by the user, or will weigh the design changes according to a weighted order or priority for the improvement criteria(s) or action(s). For example, increasing the stack height of a shoe may help increase stride length and subsequently running or walking speed or efficiency, whilst also increasing the cushioning of the shoe, potentially reducing loading or impacts and injury risks for the knee in both running and walking. In this case the design can be optimised to meet both an improvement criteria of running economy and injury risk for both actions / activities of running and walking, In another example, stiffening a carbon plate within a shoe, may improve the energy return and overall efficiency of an individual during running, but may increase loading / torques / forces experienced by the Achilles tendon during running. In this case the design may be optimised to compromise the positive effective of stiffening on running economy / efficiency with the negative effect on injury risk, resulting in a somewhat less stiff resulting shoe that balances the benefit of the increased stiffness on efficiency with the benefit of reduced stiffness on injury to achieve a moderated result for both.

[0037] Alternatively, a priority can be determined, pre-set or provided by the user to, for example, only generate designs that fully meet the criteria of efficiency, allowing limited adaptations to be made to the design that would benefit the improvement criteria of injury. In the carbon plate example above this would result in a generated design that would “accept” the increased injury risk in order to meet the efficiency improvement criteria by stiffening the carbon plate to optimise running efficiency fully). Alternatively a weighting can be provided to the improvement criteria priority to, for example, account for 80% of the full potential of design elements that contribute to an improvement criteria such as efficiency, to allow for some wider degree of design freedom to better meet the injury criteria (in the carbon plate example this would mean the carbon plate stiffness could be reduced to the point of providing 80% of the full potential efficiency benefit to somewhat reduce the compromise to the injury improvement criteria. In the example case of a carbon plate stiffness, some research (for example, Bone Stress Injuries in Runners Using Carbon Fiber Plate Footwear, Tenforde et al. 2023 https: / / link.springer.eom / article / 10.1007 / s40279-023-01818-z) has suggested that a stiff carbon plate may contribute to higher loading and / or injury risk in the foot and lower leg, thus by reducing the stiffness of a carbon plate within the provided range, a resulting footwear design could slightly reduce the risk of injury whilst maintaining a relatively high efficiency.

[0038] In some embodiments, the method may further comprise manufacturing at least a portion of the footwear component. For example, a physical item of footwear, or component or part of the footwear component is provided to the user, matching or emulating the generated design the footwear component. For example, a 3D printer maybe used to manufacture at least a part or portion of the footwear component that can be provided to the user directly. The at least a portion of the footwear component may comprise, for example, in part or whole, an insole, a midsole an outsole or a complete article of footwear.

[0039] In some examples, the footwear component comprises at least one of: an article of footwear, an insole, a midsole, an upper, a plate, an outsole, or an outsole spike configuration, an outsole stud configuration, an outsole lug configuration and an outsole grip configuration. The physical or mechanical properties may be determined by the durability, ductility, strength, roughness, co-efficient of friction, malleability, stiffness, shore, and density of the material of the component. Alternatively, a component may be a truss, a sock-liner, a heel counter, a heel collar, a guide rail, a plate, studs / cleats / spikes / lugs or a tongue with specific physical or mechanical properties.

[0040] In some embodiments, the method may further comprise comparing the generated footwear design to each pre-existing footwear designs from a database of pre-existing footwear designs to identify at least one pre-existing footwear design, wherein the at least one identified pre-existing footwear design comprises one or more footwear parameters most closely matched to those of the generated footwear design. In order to determine the most closely matched pre-existing footwear design, a pre-design design may be identified that optimised the improvement criteria(s) for all actions, and in the case that a pre-existing design will compromise one improvement criteria for the benefit of another, the design identification will be balanced to limit the compromise / benefit equally, or will prioritise the improvement criteria or action considered to be most important, either predetermined or as selected by the user, or will weigh the design changes according to a weighted order or priority for the improvement criteria(s) or action(s).

[0041] In some examples, the method may further comprise comparing the generated footwear design to each pre-existing footwear design in a database of pre-existing footwear designs so as to obtain an ordering of the pre-existing footwear design, the ordering being based on a matched level of the one or more footwear parameters of each of the pre-existing footwear designs to those of the generated footwear design, and identifying at least one pre-existing footwear designs based on the ordering. As such, a number of pre-existing footwear designs may be identified and, in some examples, can be provided to the user. For example, if a parameter of the pre-existing design compromises one improvement criteria for the benefit of another, while another pre-existing design has both improvement criteria at the similar level of benefit but at a lower level of benefit for one of the criteria compared to the other pre-existing design, then both pre-existing designs can be identified. In this way, more than one pre-existing design may be provided to the user, optionally together with information regarding the benefit towards the improvement criteria.

[0042] The method may further comprise providing the at least one identified pre-existing footwear design to the user. Identifying a pre-existing design reduces computational power compared to that required to provide the generated design to the user since the system is required only to select a pre-existing shoe as opposed to running a computationally heavy algorithm. Further it identifies to a user a footwear component that is available “off the shelf’, in stock or in a particular style, colour, made from specific materials (e.g., only natural materials, only nonanimal based materials, only sustainable / recyclable materials), or from a particular manufacturer. Further, it improves accessibility to a user as the combination of components may already be provided by a given manufacturer, omitting the need for custom components to be manufactured for each user. This reduces the necessary power, resources and time required to manufacture the footwear component that is adapted to improve the motion of the user.

[0043] In some examples, one or more differences between the generated footwear design and the most closely matched pre-existing footwear design are provided to the user as proposed modifications to the pre-existing footwear design. The most closely matched pre-existing footwear design may be the identified preexisting footwear design comprising one or more footwear parameters most closely matched to those of the generated footwear design.

[0044] In some examples, the first criterion is an improved biomechanical function of the user. The improved biomechanical function may be an improved gait of locomotion of the user. The term “gait” refers to the user’s manner of walking, running (or other locomotion), which includes a gait cycle, starting from when the heel of one foot strikes the ground and ending when the same heel touches the ground again. The gait cycle comprises two main phases of a stance phase and a swing phase. By “improving the gait” at least one phase of the user’s gait cycle is improved. Advantageously, improving the gait may reduce the user’s susceptibility to injury.

[0045] In some examples one or more footwear parameters may comprise one of: a support / stiffness parameter, a heel to toe drop parameter, a heel stack height parameter, a spring constant parameter, a dampening parameter, a shoe size parameter, a shoe shape parameter, a insole parameter, a weight parameter, a grip parameter, a friction parameter, a stud / spike / cleat parameter and an upper parameter. These parameters may be further distinguished to comprise, a stiffness parameter, a support parameter a heel to toe drop parameter, a stack height parameter, (outsole) a traction parameter (which may be influenced by outsole material, tread height / shape, stud / spike / cleat shape, width length, and / or configuration), a cushioning parameter, a tightness parameter and an upper flexibility parameter, a width parameter, a length parameter, a height parameter and a weight parameter.

[0046] Typically, the stiffness parameter may refer to the stiffness or flexibility of a footwear component (which may be influenced by plate size shape and / or stiffness (e.g., carbon plate, nylon plate, glass fibre plate, etc.)). The support parameter may refer to the support provided by the physical characteristic which may comprise any of the above-mentioned components and is commonly used to limit range of motion of the body or some parts of the body (which may be influenced by a heel counter, anti-pronation post, guide rails, dual density foams, material stiffness, shape, viscoelastic properties, etc.). The heel to toe drop parameter refers to the difference in height between rearfoot and forefoot of a component of the footwear component. The heel to toe drop is alternatively referred to as heel drop, shoe drop, shoe offset, heel differential, toe drop, pitch, gradient or simply, drop. The heel stack height parameter refers to the measure of the height of the component of the footwear component at its maximum height. In other words, heel stack height refers to the height, for example, of the sole of the footwear component stacked between the ground the foot of the user. The term “height” may also refer to a distance from the top of the component to the bottom of the component, in a direction parallel to a gravitational vector. The traction parameter refers to the measure of traction or co-efficient of friction between the footwear and the ground (which may be influenced by outsole material, tread height / shape, stud / spike / cleat / lug shape, width length, and / or configuration) and can directly influence the forces and torques experienced by the foot and body, and the ability to transfer forces to the ground potentially influencing efficiency of movements or actions The cushioning parameter may refer to a characteristic or property of the footwear component that affects the effectiveness of the component to cushion the foot of a user and / or to absorb shock when the user is in motion. For example, the cushioning parameter may refer to at least one of a dampening constant / coefficient / effect, compression properties, softness, or shore of the footwear component. The tightness parameter refers to the force applied to the foot when the footwear is worn or the range of movement possible for the foot or part of the foot within the upper of the footwear (both of which may be influenced by the size of the upper or of the footwear, the shape of the upper or footwear, the stiffness / flexi bil ity / elasticity of the upper or of the footwear, the fastening method (e.g. laces, clips, buckles, Velcro, etc.), at any part of the footwear or across the entire footwear, e.g. footwear may “lock-down” the arch of the foot by applying pressure across the laces increasing tightness in this area, whilst providing a wide and high toe-box, applying very little to no pressure across the forefoot and toes, allowing a great degree of movement and low tightness in this area of the foot).

[0047] In some examples, the step of calculating a biomechanical model may comprise calculating a target biomechanical model. Calculating the biomechanical model preferably comprises calculating a target adjustment to any one or more of magnitude of a force, direction of a force exerted, during motion, to, within or through one or more user body parts that are modelled. In some examples, the calculated adjustment correlates to a target improvement, for example, by reducing strain on target muscle groups, torque on joints, etc..

[0048] In some examples, the step of obtaining motion data comprises obtaining target motion data of the user; calculating the target biomechanical model based on the target motion data and generating, according to the target biomechanical model, a target footwear component design comprising one or more target footwear parameters; calculating a target adjustment that corresponds to a reduction in the difference of the target footwear component design and the footwear component design, wherein the first criterion represents the target adjustment. The first criterion may, in such embodiments, be configured such that the generating of the footwear design component is performed such that a modelled motion of the user wearing, using, or having their motion influenced by the modelled footwear design component is adjusted towards the target motion. In other words, the first criterion, which may in these embodiments be referred to as a motion adjustment criterion, is configured so as to cause the footwear component to influence the biomechanical forces exerted in the user body so that they are more similar to the target motion. Accordingly, the generated footwear component design may achieve user motion closer to the target motion. In this way, the first criterion may be used to determine, optimise or guide the target adjustment.

[0049] It may be advantageous, in some applications, to generate a footwear component design that approximates or reproduces the performance of an existing unknown article of footwear. For instance, it may be desired to emulate the characteristics and biomechanical impact on a wearer of an existing unknown footwear article, such as a favoured running shoe, or other footwear component worn by a user, that is no longer in production, or is not available from a particular manufacturer or distributor, is altered intentionally or through wear and tear, is a prototype etc. The method may be performed so that a design is generated with similar properties or with characteristics that reproduce similar biomechanical benefits in a wearer as the unknown shoe. An unknown footwear article, and target motion data representing motion of a wearer using that article, may be obtained, for example, from video data of an action of that wearer, as is described in this disclosure in relation to motion data generally. In some examples, the step of obtaining motion data comprises obtaining motion data from a user (or plurality of users) using a known footwear with a known footwear component design, the footwear component design, and additionally obtaining target motion data from a user (or plurality of users) using an unknown footwear with an unknown footwear component design, the target footwear component design; calculating the target biomechanical model based on the target motion data and generating, according to the target biomechanical model, a target footwear component design comprising one or more target footwear parameters; calculating a target adjustment that corresponds to a reduction in the difference of the target footwear component design and the footwear component design, or the differences in one or more footwear parameters.

[0050] The target motion data can refer to motion data of at least one user (or multiple users) that refers to, or is taken from / measured during, a desired motion of the user (or users) that the user (or users) aims to achieve when carrying out the motion. For example, a user may wish to improve their motion while running, therefore, by obtaining target motion data of the user exhibiting the target motion data by, for example, wearing a pre-existing footwear component, or combination of footwear components, that is able to facilitate the target motion of the user, a target biomechanical model can be calculated from the target motion data. Other examples might include obtaining target motion data prior to a surgical intervention, or obtaining target motion data from specific surfaces, or under specific conditions, or any other situation in which it is desirable to capture for the purposes of replicating the motion data or footwear component design at a future moment. In this way, target motion data may comprise an indication or measure of the motion of a user performing motion or during movement or whilst executing specific actions or activities.

[0051] By calculating the target biomechanical model based on the target motion data, the properties of the target footwear component can be inferred or deduced in the model such that the generated target footwear component design comprising one or more footwear parameters that are intended to emulate forces subjected to, or generated by, the user, in the target biomechanical model. Therefore, the generated footwear components are configured to adjust the target motion and / or to adjust the biomechanical forces in the body towards the those resulting from the target footwear parameters of, for example, a pre-existing footwear component which facilitates the targeted motion.

[0052] Preferably, the target biomechanical model is representative of the user and the target footwear component. When calculating a target biomechanical model of the user, it is useful to consider both the footwear component the user is wearing as well as the body of the user. This is because the shoe is the contact point between the foot of the user and the ground, such that it directly influences the forces and pressure exerted on or by the body muscles and joints. In this way, the target biomechanical model better represents the interaction of the forces and joints of the user’s body with the target footwear component. Preferably, the step of calculating a target biomechanical model may comprise calculating at least one of a target increase to the biomechanical model; and a target decrease to the biomechanical model. The calculated increase and / or decrease to the biomechanical model may comprise directly or indirectly calculating an increase and / or decrease to any one or more of magnitude of a spring constant, a damping constant, energy conversion (e.g., kinetic to potential energy), a force, direction of a force exerted, during motion, to, within or through one or more user body parts that are modelled. In this way, a targeted improvement in the motion or action or activity of the user, can be used to generate a design for a footwear component.

[0053] In some embodiments, the user may be one of a plurality of users, and wherein the step of calculating a biomechanical model may comprise calculating an average biomechanical model for the plurality of users. For example, the plurality of users may be of a specific demographic or population, such that, by calculating an average biomechanical model for the plurality of users, a design for a footwear component may be generated, wherein the footwear component be adapted to improve motion for a user of the specific demographic or population.

[0054] In particular, the step of calculating an average biomechanical model may comprise calculating at least one of an average biomechanical load distribution. In some examples, the step of calculating an average biomechanical model comprises calculating an average stiffness for the plurality of users.

[0055] For example, calculating the average biomechanical model may comprise firstly calculating values for the magnitude and direction of forces exerted upon a plurality of parts of the body for each user of the plurality of users, based upon the motion data for each user and using a computational mechanical model of the body of the user. Calculating the average biomechanical model may comprise calculating an average magnitude and an average direction offerees exerted upon a plurality of parts of the body for the plurality of users using the individually calculated magnitude and direction of ferees for each user.

[0056] For example, calculating the average stiffness may comprise calculating a spring constant for each user of the plurality of users, and calculating an average spring constant for the plurality of users. The same process may be done to calculate the damping constant for a plurality of users.

[0057] In some examples, the step of generating the one or more footwear parameters may comprise using a machine learning model.

[0058] In some examples, the machine learning model comprises one or more of: a Deep Neural Network, Convolutional Neural Network and Spiking Neural Network. Another type of machine learning model that can be used is an unsupervised regression model. This can comprise a deep neural network with a minimum of 3 layers, including 1 input layer, 1 or multiple hidden layers and one output layer. The input layer can comprise of raw or processed sensor data, information about the pre-existing footwear design worn by the user, user anthropometries or information describing the locomotion of the user, including cadence, contact time, flight time, stability, stride time, stride length, step length, balance, flight time / contact time ratio amongst other. The output comprises the generated footwear design, comprising one or more footwear parameter(s).

[0059] In some examples, the machine learning model inputs comprise at least one biomechanical parameter. In some examples, the biomechanical parameter comprises at least one of: speed, cadence, step length, foot strike, contact time, flight time, balance, stability, impulse, leg lift, strike angle, toe-off, vertical loading rate, mechanical power, biomechanical loading values for at least the portion of the body, and / or any further portions or structures of the body.

[0060] The machine learning model may be trained or otherwise configured in accordance with the first criterion. For example, the machine learning model may be configured to generate, for a given input set of data comprising or derived from motion data, a footwear component design defining one or more physical characteristics which, when the footwear component is used by the user, influence the motion of the user in accordance with the first criterion. The first criterion may correspond to some target modification or improvement in the motion of the user, and the machine learning model may be trained to generate a footwear component design specifically adapted to improve or otherwise modify the user motion towards fulfilling that criterion, in particular towards footwear component- modified motion that more closely meets or satisfies the criterion in comparison with the monitored, unmodified motion. In some examples, the method further comprises training the machine learning model to predict at least one of increasing or maximising an efficiency index, decreasing or minimising an injury risk index, increasing or maximising a stability index, decreasing or minimising biomechanical loading values for at least one portion of the body, decreasing or minimising biomechanical loading values for at least one portion of the body, increasing or maximising a spring constant, increasing or maximizing a dampening constant, increasing or maximising mechanical power, decreasing or minimising mechanical power for a user. In some examples, the method further comprises using the machine learning model to predict at least one of increasing or maximising an efficiency index, decreasing or minimising an injury risk index, increasing or maximising a stability index, decreasing or minimising biomechanical loading values for at least one portion of the body, decreasing or minimising biomechanical loading values for at least one portion of the body, increasing or maximising a spring constant, increasing or maximizing a dampening constant, increasing or maximising mechanical power, decreasing or minimising mechanical power for a user.

[0061] In some embodiments, the step of obtaining motion data may comprise obtaining video data of the user indicating the body motion of the user performing an action and deriving the body motion of the user using the video data. Preferably, the step of deriving the motion data using the video data may comprise using a machine learning model.

[0062] This model can comprise of sequential model, like convolution neural network with a minimum of 3 layers, including 1 input layer, 1 or multiple hidden layers and one output layer. The layers follow each other in a stacked manner and between each layer batch normalization shall be used, a method used to make training of artificial neural networks faster and more stable through normalization of the layers' inputs by re-centring and re-scaling. Moreover, to avoid overfitting, drop-out shall be applied to the layers, a technique where randomly selected neurons are ignored during training. The input layer can comprise of raw or processed video data of the user, information about the pre-existing footwear design worn by the user, user anthropometries or information describing the locomotion of the user, including cadence, contact time, flight time, stability, stride time, stride length, step length, balance, flight time / contact time ratio amongst other. The output comprises the generated footwear design, comprising one or more footwear parameter.

[0063] Specifically, when using a machine learning model to obtain motion data of the user, the input layer may be a convolutional 2D layer, followed by a max pooling 2D layer (hidden), that iterates between conv2D and maxPooling2D (Down sampling the input along its spatial dimensions (height and width) by taking the maximum value over an input window (of size defined by the pool size) for each channel of the input) until the output layer which is a Dense layer. Layers early in the network architecture (i.e., closer to the actual input image) learn fewer convolutional filters while layers deeper in the network (i.e., closer to the output predictions) will learn more filters. Conv2D layers in between will learn more filters than the early Conv2D layers but fewer filters than the layers closer to the output, which is why the maxPooling2D layers will down sample the learned features, squeezing the important information together. The last layer is used to predict metric values.

[0064] In other embodiments, the machine learning model comprises at least one ensemble learning machine learning model, such as an ensemble neural network. The ensemble learning machine learning model may comprise any one or more of: a Convolutional Neural Network, a Deep Neural Network, a Recurrent Neural Network, and a Spiking Neural Network. This ensemble can typically integrate temporal data such as, video, raw and processed sensor data. All models may have their unique input and hidden layers while they share the same output layer. A first model may be a convolutional neural network to handle temporal data streams, such as video data, typically by processing sequences of frames through a series of TimeDistributed Conv2D and maxPooling2D layers to extract spatial features, typically further comprising, preferably followed by, a Long Short-Term Memory (LSTM) layer to capture temporal dependencies. A second model may be an additional convolutional neural network, which typically processes raw sensor data, typically utilizing ConvI D and MaxPoolingI D layers to extract features from the time series data, typically further comprising an LSTM layer again capturing temporal patterns. A third model may be a deep neural network associated with or dedicated to processed data, which typically passes this data through one input and one or more hidden layers. The outputs from the set of models may then be concatenated to form a combined feature vector, which may be fed into one or more dense layers that produce the final prediction (last dense layer is the output layer). Typically, the motion data comprises an indication of the velocity and / or orientation of one or more parts of the body of the user during the performing of an action. By measuring or monitoring the speed or velocity of one or more user body parts, for example, and using the data collected in a computational mechanical model of the user body, for instance, the biomechanical load distribution and / or stiffness may be monitored by relating such recorded motion data to resultant forces distributed in the body using the model.

[0065] It may be advantageous, particularly for actions related to running or walking, or any form of motion, to monitor the motion of the feet of the user in particular. Therefore, in preferred embodiments, one or more parts of the body are monitored, including one or both of the feet of a user, and the motion data for each of the monitored feet comprises an indication of the velocity or orientation of one or both of the user’s feet and / or other parts of the body of the user. In particular, the motion data for each of the monitored feet comprises an indication of the velocity or orientation of the foot during the stance phase of the gait cycle. Although data may likewise be collected during other phases of the gait cycle, it may be most advantageous to monitor the motion of the feet of the user during the stance phase, when a given foot is in (typically indirect) contact with the ground.

[0066] The velocity or acceleration may be measured prior to and during the strike phase and / or the stance phase, which can be used to calculate the impulse, energy, power, or force exerted upon the feet of the user during motion or during the gait cycle, and thereby calculate the forces that are generated by and transmitted to the foot and footwear and other parts of the body of the user and calculate the biomechanical load distribution arising during those parts of the cycle. A biomechanical load distribution will provide information about the forces, torques, moments, stresses and strains experienced by parts of the body, and also the those generated by the muscles to generate or counter the load distribution. In this way, the generated footwear component comprises physical characteristics of the footwear configured to adjust the load distribution and thereby the forces that are generated by and transmitted to the body of the user in order to meet the improvement criteria, thereby improving the movement, motion, or at least one phase of the gait cycle of the motion of the user.

[0067] In other examples, the velocity may be measured during phases of various movements or actions, for example, velocity measurements may be taken during the phases of a jumping such as at the counter movement phase, the propulsive phase, the flight phase and / or the deceleration phase. Other measurements such as the distribution of pressure or forces on the feet and legs may also be obtained. In this way, the measurements can be used to calculate a force and / or impulse force exerted upon the feet during the cycle, thereby calculating the force generated by the muscles and transmitted to other parts of the body of the user and calculate the biomechanical load distribution arising during those parts of the cycle.

[0068] In addition to, or alternatively to, motion data, data indicating the force or pressure exerted between a foot of the user of the running or walking surface or ground, may be used in calculating the monitored biomechanical load distribution and / or the stiffness. The motion data therefore comprises an indication of the pressure exerted upon one or more regions of the foot as a result of a contact force exerted upon that foot by the ground during motion. The motion data may comprise an indication of the motion of one or more regions of the body or foot during motion.

[0069] In some examples, the footwear component is adapted to improve locomotion. In this way, the footwear design, comprising the one or more parameters, is modified and adapted so as to enhance a user’s ability to perform locomotion more efficiently. In other words, the parameters are calculated to provide a footwear component with features that facilitate an improved locomotion of the user. In a specific example, a sole with more cushioning (or dampening) can reduce the stress on the joints and muscles. This may help reduce the risk of injury of the user.

[0070] In some situations, the user may intend to perform motion in conditions that will affect their technique. Examples of this may be running uphill, downhill, or across terrain that is uneven. Further examples of performance-affecting conditions are softer surfaces such as grass or sand, harder surfaces such as concrete or asphalt, performing the action in a windy environment, or in extreme temperatures. In some embodiments therefore, for these cases, that the biomechanical model may be adjusted to reflect the influence of these external conditions on the runner or other type of user, such as a football player on a soft or hard pitch. Therefore, in some embodiments, the method may further comprise obtaining environment data including an indication of the terrain and / or environmental conditions in which the user is intended to perform an action, wherein the step of calculating the biomechanical model is further based on the environment data.

[0071] In some situations, the user may intend to perform motion as part of a an activity. The activity may be a sporting activity such as tennis, football, squash or basketball etc. The sporting activities may additionally be defined by the intended intensity of performance of the user such an indication that the sporting activity is highly competitive or a fast-paced race, or by contrast, is a low intensity daily jog or a daily walk. In some embodiments, the method may further comprise obtaining one or more activities, the one or more activities including an indication of the activity in which the user is intended to perform an action, wherein the step of calculating the biomechanical model is further based on the one or more activities. The type and / or intensity of the activity has an implication on the distribution of forces throughout the user’s body, therefore, to modify the motion of the user during the activity, the activity may be accounted for in the biomechanical model, such that the effects of motion of the body during the activity are accounted for in the design of the footwear component. Preferably, the activities comprises at least one of sporting activities, outdoor activities, rehabilitation activities, health activities and work activities.

[0072] In some embodiments, the one or more footwear parameters may be generated so as to optimise a muscle group used when the user performs the action as part of the activity, for example by optimising the activity or recruitment of the muscle group. In some embodiments, the method may further comprise obtaining user sizing data comprising at least one of: a foot size of the user; a shoe size of the user; a foot length of the user; a foot width of the user; a foot height of the user; and a foot arch height of the user; and wherein the step of calculating the biomechanical model is further based on the one or more user sizing data. Such data is useful for calculating the biomechanical model with improved accuracy. Foot sizing can, for example, influence the length of a moment arm, or other relevant distances used for calculating functions of, or related to force or pressure or other mechanical aspects. Furthermore, the foot sizing information may be used to generate the footwear component dimensions to ensure the appropriate fit.

[0073] In some embodiments, the method may further comprise obtaining anthropometric data comprising at least one of: a body mass index, BMI, of the user; a weight of the user, a foot size of the user and a height of the user; and wherein the step of calculating the biomechanical model is further based on the anthropometric data. Additionally, or alternatively, the method may further comprise obtaining other anthropometric data such as age or gender or the history of specific pain or injuries, pathologies, diseases (e.g. Parkinson’s or Alzheimer’s), disabilities, surgical interventions, areas prone to blisters or pain, or performance metrics such as strength (e.g. 1 rep max of e.g. squat, grip strength etc.) flexibility (e.g. sit and reach test), speed (e.g. average running speed over 5km, max speed over 100m, etc.), VO2Max, maximum heart rate, resting heart rate, or level of experience, expertise or level of competition (e.g. recreational, beginner, expert, elite, international), or preference or habitual information such as preferred running distance, preferred surfaces, intended exercise intensity, hobbies, exercise frequency, exercise seasons, fit preferences (e.g. snug, loose, tight, etc.) any of which can be used to adapt the footwear component to improve the motion of the user or optimise the footwear component design for the intended purpose. For example, if the user provides preferences such as frequent long distance, outdoor use in wet climates, the software may be programmed to recognise this type of input and limit design freedoms such as material choices to ensure a footwear component design maybe further optimised to provide higher durability and water resistance. In a further example if the user provides input regarding their running experience level, since some research has suggested that higher levels of support and / or cushioning may additionally help reduce injury risk for more novice runners, for example, Malisoux et al. 2016, Injury risk in runners using standard or motion control shoes: a randomised controlled trial with participant and assessor blinding. (https: / / www.researchgate.net / publication / 289688183_lnjury_risk_in_run ners_using_standard_or_motion_control_shoes_A_randomised_controlled_trial_ with_participant_and_assessor_blinding) the software may be programmed to recognise this type of input and limit design freedoms such as mechanical or material choices to ensure a footwear component design maybe further optimised to, for example, have higher torsional stiffness, or later / medial hardness, to provide a minimum threshold of support to more novice runners.

[0074] In some embodiments, the method may further comprise obtaining injury susceptibility data for the user, the injury susceptibility data including an indication of one or more injury-susceptible parts of the body of the user; and wherein the step of calculating the biomechanical model is further based on the injury susceptibility data. Preferably, the one or more footwear parameters are generated such that the footwear design is configured to minimise the forces that are exerted upon the one or more injury-susceptible parts as a result of the motion of the user. Preferably, the indicated parts of the body correspond to joints or muscles that are susceptible to injury. Additionally, or alternatively, the method may comprise obtaining injury data for the user, the injury data including an indication of one or more parts of the body of the user in which a previous injury has occurred. The footwear parameters may be generated in the same way as that of the injury susceptibility data.

[0075] For example, if the user has indicated they are prone to injury at a part or parts of the body, the one or more parameters will be generated so that while the user is wearing the footwear component, the forces exerted on joints or muscles that are susceptible to injury, will be minimised.

[0076] In some examples, the method further comprises assigning an injury priority weighting to the obtained injury susceptibility data, wherein the one or more footwear parameters are generated based on the injury priority weighting. In this way, the obtained data is assigned a relative priority or importance value based on a level of significance or severity. In some examples, this may comprise assigning a numerical value to each of the above-mentioned obtained data. In this way, the generated footwear design accurately reflects the importance of each injury susceptibility data in the footwear component, so as to maximise the benefit of the footwear component to the user.

[0077] In some examples, the obtained motion data may comprise motion data for one or more monitored parts of the body including one or both of the feet of the user, and wherein the motion data for each of the monitored feet comprises an indication of the velocity and orientation of the foot. Preferably, the motion data for each of the monitored feet comprises an indication of the velocity and orientation of the foot during a phase of the gait cycle. The phase of the gait cycle may include the stance phase (including for example, heel strike, foot flat, mid stance, heel-off, toe-off) and swing phase (including for example, acceleration, midstride and deceleration).

[0078] The improvement criterion may be configured in accordance with physiological, biomechanical or user defined objective data. Objective data may be received from a user input device and the biomechanical model can be calculated in accordance with the objective data. As such, wherein the first or improvement criterion is configured in accordance with objective data, such that the biomechanical model is calculated in accordance with the objective data; and wherein the objective data corresponds to a primary objective for the user that includes any of: performance improvement, efficiency improvement, strength improvement, power improvement, stability improvement, force reduction or increase, injury risk reduction, improving mobility and improving health and fitness. In some examples, the objective data may be physiological objective data.

[0079] For example, it may be desirable for some individuals to emphasise training of a particular muscle group or set of muscles, such as the glutes quadriceps, and / or calves. For walking or running performance, examples may be a lower energy cost at higher speed or reduced fatigue for longer distance capability, or reduce risk of local fatigue of specific muscle groups, such as reducing the risk of muscle soreness, cramp or lactic acid production in the glutes, quadriceps, calves, etc. For injury risk reduction, a specific region of the body can optionally be selected, such as foot and ankle or hip and back and type of injury to be avoided or of particular interest or concern, such as soft or hard tissue, connective tissue, joints, muscles, tendons, etc. alternatively specific muscle skeletal structures can be selected such as the left leg Achilles tendon, right knee anterior cruciate ligament, iliac-sacral joint, vastus medialis etc. otherwise an overall minimum risk can be the objective. Therefore, it is advantageous for the first criterion to be configured in accordance with any of the above-described objective data obtained from a user.

[0080] For example, if the first or the improvement criterion is to maximise muscular strength related performance training effects, the footwear component will be generated and adapted such that the load experienced during motion while wearing the footwear will be maximised in joints or muscles that are expected to improve performance in the activity or to exercise or stress specific anatomical structures to stimulate adaptations such as muscle growth. If the first criterion is an improvement criterion is for an individual to lose weight, the footwear component will be generated and adapted such that the load will be maximised in specific muscle groups likely to improve calorific bum, or likely to stimulate beneficial hormonal responses, such as the glutes and quads. If the improvement criterion is to improve endurance, the footwear component will be generated and adapted to result in the biomechanical load distribution will be distributed more evenly to reduce onset of (local) fatigue in any one muscle group or joint and reduce overall energy cost.

[0081] For health and fitness, example objectives can be losing or gaining weight, improve muscle tone, improve muscle strength, increase muscle size, hormone stimulation, improve joint mobility, improve stability, improve cardiovascular health or increase endurance capability for any relevant activity in which the feet are used such as running, walking, lifting weights, resistance training, home fitness, rehabilitation, healthy working conditions, etc. In some examples, the method includes assigning an improvement criterion priority weighting to the objective data, wherein the one or more footwear parameters are generated based on the improvement criterion priority weighting. In some examples, this may comprise assigning a numerical value to each of the above-mentioned obtained data. In this way, the generated footwear design accurately reflects the importance of each improvement criterion in the footwear component, so as to maximise the benefit of the footwear component to the user.

[0082] In accordance with a second aspect of the invention there is provided a system configured to generating a design for a footwear component, the system comprising: a motion data module configured to obtain motion data derived from monitoring motion of a user; a biomechanical modelling module configured to calculate, according to the motion data, a biomechanical model of at least a portion of the body of the user, wherein the biomechanical model is representative of a biomechanical load distribution through at least the portion of the body of the user and wherein the portion of the body of the user comprises one or more anatomical structures of the body other than a foot of the user; and a design module configured to generate, according to the biomechanical model and a first criterion, a footwear component design comprising one or more footwear parameters, each footwear parameter defining a physical characteristic of the footwear component.

[0083] In some examples, the design module is further configured to generate the footwear component so as to modify motion of the user according to the first criterion.

[0084] The system may further comprise a manufacturing module configured to manufacture the footwear component. In some examples, the manufacturing module is at least one of: a three-dimensional printer, a laser cutter; an additive manufacturing machine; an injection moulding machine; a cutting device, a stitching device, a weaving device, a milling machine; stitching device; a weaving device; and a lathe. The motion data module, the biomechanical modelling module and the design module may each be or comprise a computer processor. Alternatively, one computer processor may comprise any one or more of these modules. In some examples, the one or more computer processors are coupled via a wired or wireless connection. A “smart phone” or other mobile device, may comprise at least one of the motion data module, the biomechanical modelling module and the design module. The motion data module may be configured to receive motion data via a wired or wireless connection, the biomechanical modelling module is configured to use algorithms for computing the biomechanical model and the design module is configured to generate the footwear component in the abovedescribed way. Further, the system may comprise a memory and power supply.

[0085] The system may be configured to implement any of the above-mentioned methods.

[0086] In particular, the system may be capable of obtaining motion data from a sensor arrangement. Therefore, the motion data module may comprise at least one of a force sensor configured to be arranged on a foot of the user; a pressure sensor configured to be arranged on a foot of the user; a pressure sensor configured to be arranged on a floor / under the foot of the user; a force sensor configured to be arranged under the foot of the user; a camera; a photodetector sensor; an inertial measurement unit, I MU, sensor; an accelerometer sensor; a gyroscope sensor; an electromyography, EMG, sensor, a local positioning sensor, and a global positioning sensor, GPS, sensor.

[0087] The system may comprise the sensor arrangement or may comprise a receiver unit adapted to receive motion data directly or indirectly therefrom. In some embodiments, the sensor arrangement comprises at least one pressure sensor. Preferably, it is configured to monitor the pressure exerted upon one or more regions of a foot of the user as a result of a contact force exerted upon that foot by the ground during motion or locomotion; or an indication of the pressure exerted by foot during locomotion. For performed actions that involve motion, such as walking, jogging, or running, weight lifting, resistance training, home fitness, rehabilitation by the user, it may be advantageous for the purpose of calculating the biomechanical model, to measure or monitor the contact force exerted upon the foot by the ground throughout, or at one or more times or moments during, the foot of the user being in contact with the ground. Contact with the ground in this context typically reverts to indirect, rather than direct contact, since the user will typically be wearing some form of footwear such as shoes or trainers while the motion data is acquired, and so during locomotion, will typically make indirect contact with the ground, or the surface upon which the user is performing locomotion, through the shoe or trainer, and in particular the sole of the footwear. The user may be wearing a “neutral shoe” in which all footwear parameters are minimised to limit the influence of footwear mechanics upon the natural biomechanics, for example, the heel to toe drop is zero, there is a minimised stack height and minimal support and stiffness parameter. In other examples, the user will not use a neutral shoe, in which case the user will input the footwear type, use a bar code scan to identify the footwear type, or use imaging or statistical correlations from previous data to estimate the footwear type.

[0088] In some preferred embodiments, the at least one pressure sensor is positioned or attached, or is configured to be positioned or attached, in the sole of a piece of user footwear, or on the ground under the foot of a user. For example, the sensor arrangement may include an inner sole comprising one or more pressure sensors located at one or more positions corresponding to one or more respective locations on the foot of the user. In such embodiments, the sensor arrangement may measure the pressure, or the force exerted between the inner sole of the footwear and the foot of the user, at however many locations within the footwear at which a sensor is positioned. This force or pressure data may be used, for example, to calculate the force exerted upon various parts of the body, transmitted through the foot of the user, and arising from an impact or period of contact between the foot of the user and the ground.

[0089] In some embodiments, the sensor arrangement further comprises or alternatively comprises one or more inertial measurement units configured to monitor the linear acceleration and the rotational rate of a user’s body part for which it is attached, such as the foot of the user, an ankle, an arm etc and wherein the monitored motion data comprises data representative of the monitored linear acceleration and rotational rate.

[0090] In preferred embodiments, the sensor arrangement is configured, using an inertial measurement unit, to measure or monitor the linear and / or angular velocity and / or acceleration of a part of the body of the user, such as the foot of a user, and may be configured to do so in one, two, or three spatial axes for each of linear and angular measurements. In some embodiments, an inertial measurement unit (I MU) is included in the sensor arrangement and is attachable to the foot or footwear of a user. For example, an IMU may be configured to be in electronic communication with the other parts of the system and may be provided as an integral part of an inner sole or sole insert for user footwear or may comprise a clip for a fixing to user footwear, or may be adapted for or have a shape suitable for being secured or positioned within a recess within user footwear, or under the foot or insole of the user.

[0091] Thus, in some preferred embodiments, an IMU may be provided for one or both feet of a user performing an action so as to enable the velocity of the feet of the user to be measured, for example during the strike phase and / or the stance phase of a gait cycle, which can be used to calculate the impulse, energy, power, or force exerted upon the feet of the user during the gait cycle, and thereby calculate the forces that are transmitted to other parts of the body of the user and calculate the biomechanical model arising during those parts of the cycle.

[0092] In some embodiments wherein the sensor arrangement comprises a plurality of inertial measurement units, wherein each of the inertial measurement units is attachable to a part of the body or clothing of the user. Each IMU may be configured to monitor the linear acceleration and the rotational rate of the part to which it is attached, and the monitored motion data may comprise data representative of the monitored linear acceleration and rotational rate from each of the plurality of inertial measurement units. In this way, some embodiments may include sensors to monitor the motion of any part of the person of the user to which a sensor may be attached. For example, in addition to, or as an alternative to, the sensors attached to the feet of a user, motion sensors may also be provided in the sensor arrangement which are attachable to or configured to monitor the motion of the arms, hands, head, or torso of a user, for example. The system may be configured to identify the part of the body which it is attached, in use based upon a detected pattern of motion which may be associated with a predetermined body part. Measurements from other parts of the body may be used in collating, using data from a plurality of motion sensors distributed around the body for instance, a collection of data representing the overall movement of the body and / or various parts thereof. This may then be used in conjunction with a computational model of the body of the user in order to calculate the biomechanical model.

[0093] Thus, the sensor arrangement may comprise different combinations of sensor types in different embodiments. Some embodiments may include a pressure sensor configured to monitor pressure exerted upon the foot of the user, and these embodiments may comprise an optional IMU that is attachable to the foot. Some embodiments may comprise a foot pressure sensor with a plurality of I Mils that may be positioned at different parts of the body of the user. Some embodiments may comprise one or a plurality of IM Us, with no foot pressure sensor. Sensor arrangements comprising each of these sensor type combinations may be capable of calculating the biomechanical model. Additionally or alternatively, the system may use motion data obtained via one or more image sensors or cameras.

[0094] In particular, the system may be capable of obtaining motion data from a recording device. In some embodiments, the system comprises at least one camera or image sensor to record the user in motion or locomotion.

[0095] The algorithms that are typically used may be implemented in an artificial intelligence advice module that is optionally some form of embedded computing device or any other computing device. The algorithms may be run on servers or can be on-chip locally on a wearable device, or can be locally processed on a mobile device, on a smart watch, or on a conventional computer system / laptop. In some embodiments, the system further comprises a user interface, an information delivery device and / or a display screen configured to provide the footwear design to the user. Any of the previously mentioned devices may further comprise a device for providing any of visual, audible, haptic, CAD, physical footwear items, or other forms of information or signal to a user that may provide the footwear design.

[0096] In some embodiments, the user interface is configured to obtain an input from the user relating to at least one of environment data, exercise activities, daily activities, a body mass index of the user, a height of the user, a weight of a user, a gender of the user, an age of the user, an injury history of the user, a physical fitness of the user, a health of the user, injury susceptibility data for the user, one or more primary objectives (e.g. biomechanical, performance, physiological, health, functional) and physiological objective data.

[0097] In accordance with a third aspect of the invention there is provided a system configured to generate a design for a footwear component according to the method of the first aspect. The system may comprise any one or more of the features, components, and modules described in relation to the second aspect.

[0098] In accordance with a fourth aspect of the invention there is provided a non- transitory computer-readable storage medium configured to store computer executable code that when executed by a computer configures the computer to: obtain motion data derived from monitoring motion of a user; calculate, according to the motion data, a biomechanical model of at least a portion of the body of the user, wherein the biomechanical model is representative of a biomechanical load distribution through at least the portion of the body of the user and wherein the portion of the body of the user comprises one or more anatomical structures of the body other than a foot of the user; and generate, according to the biomechanical model and an improvement criterion, a design for a footwear component, the footwear design comprising one or more footwear parameters, each footwear parameter defining a physical characteristic of the footwear component. Throughout the specification, the terms “shoe”, “article” and “article of footwear” may be used interchangeably to mean a footwear component.

[0099] BRIEF DESCRIPTION OF DRAWINGS

[0100] Figure 1 shows a part of an example system for obtaining motion data;

[0101] Figure 2 shows a part of an example system according to the invention including a shoe insole with pressure sensors;

[0102] Figure 3 is a box diagram depicting an example system according to the invention;

[0103] Figure 4 is a diagram of a link segment model for the human leg which may be used in an example method according to the invention;

[0104] Figure 5 is a diagram of a free body model for a single element which may be used in an example according to the invention;

[0105] Figure 6 is a diagram of an anatomical model of an ankle / foot of a user which may be employed in an example according to the invention;

[0106] Figure 7A is a diagram of a user carrying out a squatting action and illustrates the corresponding displacement-time graphs, force-time graphs and power-time graphs;

[0107] Figure 7B illustrates a model of the human body which may be used in an example method according to the invention;

[0108] Figure 8 is a diagram of a spring-mass model for a leg of a user which may be employed in an example according to the invention;

[0109] Figure 9A is a diagram of a spring-dampener-mass model for a leg of a user which may be employed in an example according to the invention;

[0110] Figure 9B illustrates a scatter graph of average flight time plotted against average contact time grouped by colour according to cadence range; Figure 9C illustrates a scatter graph of flight time and contact time plotted against running speed;

[0111] Figure 10 illustrates a flowchart illustrating an exemplary monitoring and storage process according to the present invention;

[0112] Figure 11 illustrates a flowchart illustrating an exemplary biomechanical modelling process according to the present invention;

[0113] Figure 12 illustrates a flowchart illustrating an exemplary biomechanical targeting process according to the present invention;

[0114] Figure 13 illustrates a flowchart illustrating an exemplary design generation process according to the present invention;

[0115] Figure 14 illustrates a flowchart illustrating a feedback process according to the present invention;

[0116] Figure 15 illustrates a flowchart illustrating an exemplary method according to the present invention;

[0117] Figures 16A, 16B, 16C and 16D illustrate an exemplary presentation of the generated footwear design to a user according to the present invention;

[0118] Figure 17. illustrates an exemplary presentation of the biomechanical loading model.

[0119] DETAILED DESCRIPTION

[0120] In a first example, a system is described in which a method according to the present invention may be implemented.

[0121] To obtain motion data, the system uses data from a sensor. Optionally, the sensor is a foot sensor. In other examples, the sensor may be an image sensor, camera, or electromyogram (EMG sensor). The sensor may be used in conjunction with a GPS location sensor, gyroscope, accelerometers, temperature monitors, magnetic sensors and I Mils. Further sensors such as a heart rate monitor can also be added to the system to improve injury risk assessment and / or performance assessment and / or health assessment.

[0122] A schematic of an exemplary foot sensor is shown in Figure 2. The foot sensor may be worn on the foot of a user as part of a shoe or may be attached to a shoe and comprises one or more of a force or pressure sensor. The optional additional sensors may also be included in with the foot sensor as part of a foot sensory system or may be located elsewhere on the body of the user. The motion data for a user can be obtained from said sensors during one or more running interval(s) on a treadmill carried out by a user, as shown in Figure 1.

[0123] A computer processor, memory and power supply are contained within a small module that is attached to, or placed inside the user’s shoe, foot, or body, or alternatively in or on the ground under the foot of the user. The module can communicate wirelessly with a remote computer that is configured to provide a generated footwear design to the user. In some embodiments, the footwear design can instead be generated at the module. The remote computer can be a “smart phone”, tablet, multimedia entertainment system or any other suitable computer. Alternatively, input and output devices can communicate directly with the shoe module by wired or wireless link. While the user is moving or in motion, for example running, data can be stored in the shoe module or optionally transmitted to a remote computer and after the run the data can be read out to a remote computer for data analysis and presentation of generated footwear design.

[0124] Additionally, or alternatively, motion data is obtained from video data of the user indicating the body motion of the user during locomotion or other activity. The system therefore can include a recording device such as a video camera, included in as part of a “smart phone”, or a webcam for example, and is arranged so as to record the user while in motion, for example, while running on a treadmill. The recording device may be included in the remote computer or is configured to communicate wirelessly with the remote computer, such that the video data can be transmitted to the remote computer or stored locally and later transferred to the remote computer. Preferably, the video data captures the whole of the part or parts of the body intended to be modelled by the biomechanical model.

[0125] In some examples, motion data may be obtained by manually labelling movement data in video, such that deterministic algorithms may be used to determine join positions. In some examples, markers can be used to obtain motion data. In some examples image processing can be used to obtain motion data.

[0126] In some examples, machine learning can be used to derive and measure motion data of the user from the video data. The processor of the remote computer can be configured to use a machine learning model that tracks the motion of the part or parts of the body in the video data. As above, a machine learning model, such as the described neural network, obtains motion data comprising an indication of the velocity and / or the orientation of one or more parts of the body.

[0127] In some examples, using the recording device, video footage comprising a plurality of frames of the part or parts of the body is inputted into the machine learning model which can carry out image segmentation. The machine learning algorithm segments the parts of the body in each frame by locating the edges of the said body parts so as to track the changes in the position or location of the body part across the frames, thereby deriving the motion of the user from the footage. Preferably, a neural network (NN), deep neural network (DNN), spiking neural network (SNN), or convoluted neural network (CNN) can be used to track the main joints of the body in each frame of the plurality of frames, and based on the relative positions of the joints, a (part(s) of a) human or humanoid skeleton and its movements can be reconstructed for the user.

[0128] The obtained motion data is used to calculate a biomechanical model for the body of the user which can comprise a biomechanical load distribution and / or a stiffness for a part of the body (e.g., the foot) and / or a leg stiffness. Both calculated component of the biomechanical model will be discussed below.

[0129] By the remote computer, the method includes calculating a numerical representation of the mechanical forces and / or torques / moments and / or stresses / strains at different position in the body. Such values are included in the “biomechanical load distribution” and are calculated by using a mechanical model for the body and deriving the load, in terms of ferees and moments on individual joints, from external sensor measurements and / or derived video data measurements. The system can continue to measure and recalculate the forces and moments at regular intervals throughout an action or movement. One measure of the relative biomechanical load distribution at different locations is obtained by determining the average over the action or movement for all the calculated values of force or moment for each location in the body. Alternative measures are possible such as the maximum, minimum, median, mean, mode, range or standard deviation over the duration of the action or movement. The principles of inverse dynamics are well known (see for example https: / / en.wikipedia.org / wiki / lnverse_dynamics) where the body limbs are approximated by a link-segment model and forces and moments are computed from measurements of the motion of limbs and external forces such as ground reaction forces.

[0130] In this first embodiment measurements from spatially distributed pressure sensors, embedded in an insole, are used to calculate the forces experienced by the ankle joint and which muscle groups are most actively engaged to exert the moment / torque about the joint during the stance phase of running. In this embodiment, the user may partake in a running session on a treadmill wherein the user runs on a treadmill for a given period of time while wearing spatially distributed pressure sensors, as shown in Figure 1. Because there is no direct measurement of limb movement in this embodiment, the limb positions have to be deduced from the foot pressure measurements and the following example shows how this is achieved.

[0131] The forces experienced by various segments or anatomical structures may also be calculated through statistical correlation parameters derived from measurements from a sensor or plethora of sensors mounted on the person, and population data from lab-based measures and inverse dynamics or other relevant reference data. Alternatively, machine learning, optionally in the form of a neural network, may be employed.

[0132] Additionally, joint angles can be estimated from statistical correlations that relate the position of the measured pressure, or the timing of the measured pressure, to reference data, for example, pressure applied at the forefoot when the foot first contacts the ground during running can be correlated to average foot or ankle angle (Mann et al. 2013, Reliability and validity of pressure and temporal parameters recorded using a pressure-sensitive insole during running) https: / / www.researchgate.net / publication / 256927941_Reliability_and_validity_of _pressure_and_temporal_parameters_recorded_using_a_pressure- sensitive_insole_during_running ).

[0133] Figure 4 shows the forces and moments operating on a single segment. In the bottom-up inverse dynamics procedure, the forces and moment for the distal joint, together with the mass, dimensions and acceleration of the segment are used to determine the forces and moment for the proximal joint, using the equations of motion. Equal and opposite reactive forces and moment for this proximal joint are then used as the forces and moment for the distal end of the next segment that is closer to the body. In the stance phase of running, the foot is contacting the ground and experiences a ground reaction force which is the first external force in the link segment chain. An example of a more anatomical representation which may be needed to determine forces and moments in some circumstances, is shown in Figure 5. In this example a more anatomical representation of the foot is used to determine the forces and moments for the distal end of the foot, and this is shown in Figure 5.

[0134] During the stance phase in stable state running, a person is supporting the body on one foot with spatially distributed pressure sensors embedded in an insole under the supporting foot. The ground reaction force Ry3is measured via the spatially distributed pressure sensors embedded in the insole. The centre of pressure (COP) is calculated from the known positions of the pressure sensors in the insole and the model assumes all pressure to act through this calculated point. The total force acting through the COP can be derived from the foot pressure sensor measurements but the mass and position of COM of the foot segment have to be inferred from the person’s available anthropometric data and population statistics. The angle of the foot is inferred from the pressure and position of the COP relative to the rest of the foot during the stance phase by using the correlation of angle (measured by a motion analysis system) with pressure and COP propagation for a representative population of runners. Similarly, the velocity and acceleration of the ankle can be estimated. This follows a similar statistical approach used for example by Mann et al (Gait & posture 39(1), August 2013, “Reliability and validity of pressure and temporal parameters recorded using a pressure-sensitive insole during running” https: / / www.researchgate.net / publication / 256927941). For greater accuracy it is also possible to take foot angle measurements from video or alternatively from an inertial measurement unit (I MU) mounted below the ankle joint (on the shoe if present) to determine angular rotation without needing to make statistical estimates.

[0135] The system may be configured to obtain anthropometric data of the user for use when calculating the biomechanical load distribution. The user may input this data via the remote computer or a device in communication with the remote computer, such as via a smart phone. The anthropometric data may include at least one of a body mass index (BMI), a weight, and a height, a gender, an age etc of the user, which can be used in the calculations of the biomechanical load distribution. The anthropometric data may also be obtained from an image or multiple images of the user using for example image processing or a machine learning algorithm such as a neural network (e.g. a convolutional neural network) as described in this specification or any other suitable algorithm.

[0136] In a specific example, at a particular point in time for a person, the COP is 0.03 m from the ankle joint. The force acting through the COP calculated from the pressure measurement (Ry3) is 686.7 N. In case the size of the foot is not known or input by the user, it can be estimated from the size of the sensing insole and position of the sensors, thus, based on a scaling factor derived from average population statistics and the person’s anthropometric data, the centre of mass is estimated to act 0.05 m from the ankle joint. The person’s mass is 70 kg, and based on average population anthropometric data, the mass of the foot is 1 kg. In the position in time captured in Figure 5, the foot is flat on the ground, ay= 0 and in steady state running, the acceleration ax= 0. Therefore, applying the equations of motion

[0137] Ry3 = 686.7 N.

[0138] 1. = m.ax,

[0139] Rx3P+ RX3 = m.ax= 0

[0140] 2. Fy= m.ay,

[0141] Ryzp + Ryz - m.g = m.ay

[0142] Ry3P+ 686.7N - 1 x 9.8 = 0

[0143] Ry3P=-676.9N

[0144] 3. About the COM, _ MP= .a,

[0145] Mp- Ryz x (0.05-0.03) - Ryzp x 0.05 = 0

[0146] Mp = 686.7 0.02 + (-676.9 0.05) = -33.85N.m

[0147] In this example the net muscle moment Mp is negative which indicates that the plantar flexor muscle groups are active in generating the moment / torque necessary to maintain the ankle angle. This means forces are experienced or exerted by the anatomical structures collectively commonly referred to as the “calf’ which includes muscle groups such as the gastrocnemius, soleus, and tendons such as the Achilles and planta fascia.

[0148] The force and moment calculated for the proximal (ankle) end of the foot segment 3 is then used to calculate the equal and opposite reactive components used as input to the next segment in the biomechanical chain, namely segment 2 knee / lower-leg. The mass, length and COM of the leg are estimated from the person’s available anthropometric data and population statistics. The angle of the knee can be estimated from the sensor measurements and any available anthropometric data for the person using correlations to kinematic data obtained by a motion capture system using a large population of runners. Thus, with the instantaneous estimates of heel position, velocity and acceleration, leg length and knee angle, the equations of motion can be solved to obtain the force and moment at the knee joint. In this case, the resulting muscle moment indicates how actively the quadriceps muscle groups are engaged. Optionally, additional measurements from sensors such as, video, goniometers, EMG, or one or more inertial measurement unit (I MU) mounted above, below, or on relevant articulating joints can be used to determine foot or limb position and angle and thus improve the correlation estimate of knee angle.

[0149] This process of using data from foot mounted sensors, anthropometric data and optionally correlations established from kinematic studies on a large population of runners is used to estimate forces and moments for all the linked segments. The accuracy of these estimates decreases the further the limb segment is further from the foot, but accuracy can be improved by adding further sensors at relevant locations on the body.

[0150] Described above is a method for calculating the forces and moments when the user is in locomotion (for example when the user is running or walking, etc.) however a biomechanical load distribution may be calculated when a user is carrying out a different activity such as squatting, weightlifting, resistance training, jumping etc. For example, Figure 7A shows force-time curve metrics of a user carrying out a squatting action. From the timing of the applied pressure, which may be obtained from at least one of the above-described sensors, it is possible to determine the counter movement and propulsion movement in a jump or squat, from which it would be possible to correlate this timing to ankle, and / or knee angles during a squat or vertical jump. Using this information, a biomechanical load distribution of the user is calculated.

[0151] In some examples, the biomechanical loading of the user may be calculated from statistical correlations between one or more metrics derived from measured movement data, and reference data from a database of user data or from scientific literature such that the metric may serve as an indirect indicator of loading, force, torque, moment, stress, or strain. For example, for running, from measured movement data and reference data from a database of user data or from scientific literature, there is a statistical correlation between cadence and knee loading, with lower cadence associated with higher knee loading. There is also a statistical correlation between forefoot landing and increased Achillies tendon and calf muscle loading and between rearfoot landing and higher knee, quadricep, hip and lower back loading. There is also a statistical relationship between stride length and loading at low cadence, with longer stride lengths at low cadence correlated to higher loading.

[0152] Further the biomechanical loading of the user may be calculated from statistical correlations between multiple metrics derived from measured movement data and reference data from a database of user data or from scientific literature such that the combination of metrics may serve as indirect indicators of biomechanical loading distribution through the body. One way to calculate such a correlation is to utilise a mathematical function such as sum, multiply, divide, whilst optionally applying a weighting factor or off-set to each of the input metrics to calculate a resulting, potentially unitless, loading values for , the associated body segment or location in the form of a single integer that can be correlated to load, force, torque, moment, stress or strain.

[0153] Further machine learning, for example a neural network, may be used to establish statistical relationships or correlations between sensor data, and / or motion metrics and reference body segment loading values measured and calculated in controlled conditions such as in a laboratory. Such methods have been published in scientific literature so would be familiar to an expert in the field, for example measured movement data and reference data from a database of user data or from scientific literature, Stetter et al. 2019 (https: / / www.ncbi.nlm.nih.gov / pmc / articles / PMC6993119 / ).

[0154] Additionally or alternatively, by the remote computer processor, the method can comprise calculating numerical representations of a spring constant associated with at least a part of the body. In particular, the system can calculate a spring constant associate with a leg of the user. The spring constant is the “stiffness” of the part of the body and is calculated using a mechanical model. The part of the body by which the spring constant is calculated may be the same as the part of the body by which the biomechanical load is calculated or may be a different part of the body. Take for example a runner’s leg, as the foot contacts the ground during for example a stance phase, the leg “spring” compresses, storing elastic energy until mid-stance and returning mechanical energy from mid-stance through ground contact. In this way, the musculoskeletal structures of the leg, will store and return energy similarly to a spring such that the leg can be modelled simply as a massless spring loaded with the runner’s body mass as a point mass. In practice, the materials of the footwear and the ground surface will also contribute to the exchanging of energy when a user is in motion. For example, factors such as surface stiffness, inclines or declines will affect how the leg behaves and therefore will affect the spring-like properties of the leg. Therefore, in some examples, the influences of the materials of the footwear and the ground surface will be accounted for when modelling the leg as a spring.

[0155] Figure 8 shows an exemplary spring-mass model of the leg. In this embodiment the user may carry out a running session on a treadmill as shown in Figure 1 , and measurements from spatially distributed pressure sensors, embedded in an insole, are used to calculate the forces experienced during the stance phase of running. Anthropometric data may also be obtained for the user.

[0156] The motion of the user is modelled with a spring-mass system, as described by Moores et al. (2019) https: / / www.frontiersin.org / articles / 10.3389 / fspor.2019.00053 / full. The stiffness is calculated by dividing the maximal ground reaction force during contact by the peak displacement of the leg spring. The sine-wave-method from Dalleau et al. (2004) https: / / pubmed.ncbi.nlm.nih.gOv / 15088239 / #: ~:text=Abstract,measured%20by% 20a%20contact%20mat., used to model vertical jumps, is used to calculated the peak vertical force. The centre of mass displacement at mid-stance (ACoM) can be inferred from the person’s available anthropometric data and population and the length of the leg (L) can be inferred from the height of the participant. Using the following equations, the leg stiffness can be calculated:

[0157] Wherein, M or m represent weight [kg], g represents gravity [rr / s 2], Ta or tf represents flight time [s], Tc or tc represents contact time [s], v represents running speed [rr / s], L or LO represents leg length [m] = 0.53 * height, K or K_leg= leg stiffness [N n], co represents natural frequency [-], DF represents duty factor [-], ACoM represents centre of mass vertical displacement [m], AL represents leg spring deformation [m], Fmax or Fpeak represents maximal vertical ground force [N], z_spring represents centre of mass position during spring, z_projectile represents centre of mass position during projectile.

[0158] To improve on the model for calculating a leg stiffness to better replicate the biological structure of the human body or leg, or of the common materials used or found in footwear and surfaces, a viscoelastic mechanical model can be used. Therefore, another method to calculate the stiffness is to add complexity to the model with a dashpot, creating mass-spring-damper model, and increase the accuracy of estimating the stiffness. Viscoelastic mechanical models model the behaviour of viscoelastic materials using a linear combination of springs and dashpots to represent the elastic and viscous components respectively. The method of calculating the spring constant for the part of the body may therefore comprise using, for example, a Maxwell model, a Kelvin-Voigt model, or the Standard Linear Solid model.

[0159] Figure 9A shows an exemplary model of a leg of a user represented by a singlebody MSD. The leg is modelled as a single spring-dashpot representation, with a dashpot (i.e., a viscous damper) and an elastic spring connected in parallel and loaded with a mass. The mass is attached to the ground by both the spring and the damper. An external force is applied on top of the mass, which represents weight of the user. Therefore, the external force may be calculated using the person’s available anthropometric data, by for example, multiplying their mass, m, by the gravitational constant, g. By using equations of mechanical vibrations, the spring constant for the leg may be calculated: mz" + cz' + kz = Fext= —mg

[0160] Where z is the centre of mass position and z(t) is the displacement of the mass relative to its equilibrium position, k is the spring constant, c is the dampening constant.

[0161] For spring - damper models, the resonance, rate or time of energy loss or energy return may in some cases be relevant. In these cases, the Quality Factor or Q factor can be relevant to calculate. With the Q factor it is possible to optimise the relationship between spring stiffness and damper to match that of a specific user and / or footwear component. Q factor may be calculated according to one of the exemplary equations as follows:

[0162] Wherein, Q is the Quality factor or Q factor, E is the maximum energy stored during one cycle, AE is the energy lost during one cycle.

[0163] Wherein, Q is the Quality factor or Q factor, co is the angular frequency, Emax is the maximum energy stored, AP is the power loss, Wherein, Q is the Quality factor or Q factor, fris the resonant frequency, Af is the resonance width

[0164] Wherein, Q is the Quality factor or Q factor, m is the mass, k is the spring constant, D is the damping coefficient.

[0165] In some examples, the stiffness of the user may be calculated from statistical correlations between a single metric derived from measured movement data, and reference data from a database of user data or from scientific literature such that the metric may serve as an indirect indicator of stiffness, or alternatively, efficiency.

[0166] An example of such a metric is the flight time to contact time ratio. Figure 9C demonstrates flight time and contact time in relation to running speed. As shown, at lower running speeds, the average contact time is much higher than average flight time. However, as running speed increases, typically running efficiency also increases and the two average times begin to converge. The ratio of these times under certain conditions can be correlated to leg stiffness or efficiency during running, with higher ratios correlated to higher leg stiffness and / or higher running efficiency.

[0167] Figure 9B illustrates a scatter graph of average flight time plotted against average contact time grouped by colour according to cadence range illustrating further statistical relationships between flight time, contact time, and cadence.

[0168] Further the stiffness of the user may be calculated from statistical correlations between multiple metrics derived from measured movement data and reference data from a database of user data or from scientific literature such that the combination of metrics may serve as indirect indicators of stiffness, and / or efficiency. One way to calculate such a correlation is to utilise a mathematical function such as sum, multiply, divide, whilst optionally applying a weighting factor or off-set to each of the input metrics to calculate a resulting, potentially unitless, stiffness index and / or efficiency index, in the form of a single integer that can be correlated to stiffness and / or efficiency and thus used as a stiffness indicator and / or efficiency indicator. An example of this approach would be to use the inputs of flight time to contact time ratio and contact time as used in the following exemplary calculation:

[0169] Stiffness index or Efficiency index = FT / CT + 0.9934 x (1 - CT / 1000)

[0170] Where: FT = Flight time, CT = Contact time, and FT / CT = flight time to contact time ratio.

[0171] The stiffness index, or efficiency index, may also / further be calculated utilising logic to optimise the stiffness index or efficiency index as used in the following exemplary calculation:

[0172] Stiffness index or Efficiency index = FT / CT + k x (1 - CT / 1000)

[0173] IF Cadence > 190, then: k = 0.9974. IF 180 < Cadence < 189, then: k = 0.9952 IF 170 < Cadence < 179, then: k = 0.9934. IF 160 < Cadence < 169, then: k = 0.9913. IF 150 < Cadence < 159, then: k = 0.9895. IF 140 < Cadence < 149, then: k = 0.9879

[0174] Or alternatively: Stiffness index or Efficiency index = (CT*-1+296) / CT

[0175] Stiffness index or Efficiency index = (CT*m+c-k) / CT

[0176] Where CT = contact time, m = gradient of contact time to flight time ratio best fit from user data, c = intercept of contact time to flight time ratio best fit from user data, k = cadence normalisation constant

[0177] IF Cadence > 190, then: m = -1 , c = 296.9, k = 0. IF 180 < Cadence < 189, then: m = -1 , c = 330, k = 33.1. IF 170 < Cadence < 179, then: m = -1 , c = 345, k = 48.1. IF 160 < Cadence < 169, then: m = -1 , c = 365, k = 68.1. IF 150 < Cadence < 159, then: m = -1 , c = 390, k = 93.1. IF 140 < Cadence < 149, then: m = -1 , c = 410, k = 113.1 In other examples, the leg may be separated into one or more components, with each component modelled as a spring-dashpot representation, which collectively form a spring-dashpot system. The equations of mechanical vibrations may be adapted accordingly.

[0178] In addition to foot pressure sensors to measure ground contact forces, I MU sensors can be connected to other parts of the body in order to get a more direct estimate of limb positioning, rather than having to use statistical correlation using measurements from a population of runners. In Kim et al “Estimation of Individual Muscular Forces of the Lower Limb during Walking Using a Wearable Sensor System” Hindawi Journal of Sensors Volume 2017, Article ID 6747921 https: / / doi.org / 10.1155 / 2017 / 6747921 , IMUs are attached to the body and are used to estimate limb positions and accelerations to enable muscular forces to be estimated.

[0179] In some examples, only IMU sensors are used to determine body kinematics and no foot pressure sensor is employed. The number and placement of IMU sensors determines how accurately the biomechanical model can be determined and statistical correlation modelling with a population of runners is required to estimate the ground reaction forces. This approach has also been applied to ski jumping for example (Logar and Munih, 2015, Sensors 2015, 15, 11258-11276; https: / / doi.org / 10.3390 / s150511258).

[0180] To further improve accuracy, especially regarding mechanical or metabolic efficiency, further physiological parameters can be obtained if suitable additional sensors are worn. For example, blood oxygen level can be estimated with an SpO2 monitor using light transmission through capillaries in the skin and breathing rate can be measured using an IMU strapped to the chest, or strain gauge embedded in the chest strap Other physiological parameters can be obtained by using, for example, a heart rate monitor or an EMG sensor.

[0181] The biomechanical load distribution indicates how much load is experienced or generated by the joints and how much by the muscles and tendons. Joint load is generally related to segment forces and muscle / tendon loads are generally related to segment moments. The distribution can therefore indicate how hard muscles are working and which structures are exposed to extra loading and the related effects such as potential risk of injury.

[0182] In the example case of running, whilst the user will usually be able to alter aspects of their running style to change cadence, stride length, which part of foot makes initial contact with the ground, knee flexion, pre-tension of specific muscles such as abdominal muscles, leaning forward or backward, degree of pelvic rotation, reduce bounce in the run or adjusting the relative time they spend on each foot or how hard they push on each foot, they will find it more difficult or impossible to change physiological and / or biomechanical parameters such as, pronation, impulse, power, contact time, flight time or stability. Similarly, whilst the user can alter or modify their form when for example weight training or squatting for example by changing stance width, foot positioning, load or range of motion, they will again find it difficult to change physiological and / or biomechanical parameters. The same issues occur when a user is carrying out a variety of actions or activities. Therefore, the generated footwear component is adapted to modify motion the motion of the user without requiring them to actively attempt to alter their style. In a specific example, the generated footwear component is adapted to improve motion the running style of the user without requiring them to actively attempt to alter their running style.

[0183] Anthropometric data may be obtained via any suitable device. As above, the method may obtain anthropometric data such as height, limb dimensions, weight and gender rather than the system relying on population averages. The user can additionally or alternatively enter their body mass index (BMI) or their BMI may be calculated from other obtained anthropometric data such as height and weight. Some estimate of anthropometric data may also be estimated or calculated from the data collected, for example, height, limb dimensions, gender, etc. may be extracted from video footage, weight may be extracted from force or pressure measurements, etc. This data is used to scale the metrics to make them more relevant to the user. For example, the weight of physical components may be scaled to be a factor of the user’s weight so as to prevent the footwear component from being too heavy for the user, step length maybe scaled to a proportion of the user’s height, forces may be scaled to a proportion of the users body weight, etc. The obtained data may comprise sizing information about the user pertaining to their foot size, including adding parameters such as shoe size, foot length, foot width, foot height, foot arch height. Preferably this information is automatically determined from a photodetector or an image (for simplicity) or multiple images (if higher accuracy is required) of the user’s foot, insole or shoe, or optionally both feet, insoles or shoes, using image processing to establish the foot sizing parameter or parameters. Optionally this is achieved by automatically determining the boarders / outline of the foot, shoe or insole from a digital image with sufficient contrast between the foot, shoe or insole and background of the image. Alternatively, a machine learning approach may be used to identify the borders / outline of the foot, or the key foot sizing parameter(s), which may be in the form of a convoluted neural network. The processing may be performed in 1 D, 2D or 3D depending on the amount of input data and required level of accuracy for the specific application. The foot sizing parameters will then be used as input to the footwear design to determine the optimal shape, size, or topology for the user or application.

[0184] In some embodiments, the session may comprise multiple actions, activities, motions or locomotion types, from which specific actions can be characterised to identify and isolate relevant actions for the improvement criteria and footwear. Optionally the generated footwear design will attempt to optimise the improvement criteria for multiple actions, either by averaging, or through weighted criteria to mathematically prioritise specific actions and / or improvement criteria for one or all actions.

[0185] Whilst there are well established example methods to classify actions within a series of movement data, (for example Feature-Free Activity Classification of Inertial Sensor Data With Machine Vision Techniques: Method, Development, and Evaluation, Veiga et al. 2017 https: / / www.ncbi.nlm.nih.gov / pmc / articles / PMC5562934 / or “Classification of basic daily movements using a triaxial accelerometer Mathie et al. 2004 https: / / www.researchgate.net / publication / 8213211_Classification_of_basic_daily _movements_using_a_triaxial_accelerometer or Mobile Sensor Data Classification for Human Activity Recognition using MapReduce on Cloud, Paniagua et al. 2012, https: / / www.researchgate.net / publication / 257719519_Mobile_Sensor_Data_Clas sification_for_Human_Activity_Recognition_Using_Map_Reduce_on_Cloud) one further example simple method of classifying actions while running, walking or idling is to use metrics such as (CT), swing time (ST) and flight time (FT). First, the frequency response (using the Discrete Fourier Transform) of the FT / CT ratio (of both feet) is determined and used to find the number of steps that where taken. Typical running cadences or step frequencies generally range between 110 to 250 steps per minute, whereas typical walking cadences or step frequencies generally range between 30 and 140 steps per minute. Next, contact time (CT) and swing time (ST) are calculated from the movement data. Every step is categorized in one of the following categories using a set of if-statements.

[0186] 0. idle

[0187] 1. walking

[0188] 2. running

[0189] If there was a long period of inactivity (no steps detected) before the current step, the step is categorized as idle.

[0190] A step is considered a running step if all requirements below are met:

[0191] CT < 0.5 & (CT + ST) < 1.0

[0192] If the step did not meet the requirements for a running step, it is checked whether it meets all the requirements for a walking step:

[0193] CT < 2.0 & FT < 0.0 If the requirements for a walking step and for a running step are both not met, then the step is categorized as idle. The output of this algorithm is the step type, a list containing the step category (0-2) of each of the steps.

[0194] As shown in Figure 10, the method can further include obtaining data from the user on any previous injuries and optionally when they occurred, and / or includes obtaining an indication from a user of parts of the body that are prone to injury. The user may indicate their previous injury by inputting the relevant information to the system audibly or using a user interface in communication with the computer. In an embodiment in which video data is obtained, the user may indicate in the video data by motioning, such as by pointing to, the part of the body, joints or muscles that are prone to injury. Therefore, to reduce risk of injury, the generated article of footwear will be generated so as to maintain similar performance while accounting for these previous injuries. The footwear component can be generated so as to minimise the force exerted on joints or muscles that are susceptible to injury.

[0195] The method can further include obtaining environmental data including an indication of the terrain and environmental conditions in which the user is intended to perform an action. Terrain and environmental conditions will affect how the user will perform and this can be taken into account. For example, if the user intends to focus on improving their road running abilities, the biomechanical model will be calculated taking this factor into account, and a footwear design will be generated to that has a spring stiffness better suited for firmer surfaces, compared to if the footwear would be optimised towards softer surfaces like grass, or sand for example.

[0196] For example, in wet conditions the ground may become slippery, as is the case with asphalt, concrete, or grass, and some surfaces may become softer such as grass or earth. In these example cases, softer ground can be included in the model within the, or as additional spring and dampening constants. Slippery surfaces can be modelled by including a co-efficient of friction in the calculations such as: M = |Ff| / |FN|

[0197] Where p = coefficient of friction, Ff= frictional force and FN= normal force. In Figure 7B below an example of how to utilise the co-efficient of friction in a relevant example is provided:

[0198] Where N = normal force, FpuSh = vector force applied by the foot of user on ground and Fg= Gravitational force applied to user’s centre of mass In the case that the foot slides or slips across the floor the frictional force can be included in the biomechanical model to calculate the resulting forces, torques, moments, stresses and strains, accelerations and velocities. The forces applied during ground contact may be measured or calculated from movement data, and, in the case that the foot slips across the surface, movement of the foot across the ground can also be measured or calculated from movement data. In some cases, for example, when only force or pressure data is available, it may be preferable for the user to report if the foot has slipped during an action or step. Optionally the system may be configured to assume the foot has, or has not, slipped during a motion Optionally the system may be set-up to be configured to be able to select a surface with known co-efficient of friction, which may be adapted according to the conditions, for example if it is wet or dry, hot or cold, etc. optionally the system may be configured to allow user to enter a user defined co-efficient of friction or use a predetermined default. .

[0199] In an example case if the co-efficient of friction is insufficient over all or part of the footwear, and the footwear is slipping more than is required (which may for example reduce deceleration, cutting or change of direction performance or other performance or injury risks) for all contact areas or partial contact areas, the generated footwear design is likely to include, for example, in the relevant contact area or areas, one or more of, outsole materials with higher co-efficient of friction, midsole or outsole materials that are softer and likely to conform better to the underlying surface, effectively increasing the surface contact area, increased outsole area, additional lugs, deeper lugs, wider or narrower lugs, additional spikes, longer spikes, additional studs, longer studs, wider or narrower studs, etc. depending on the type of surface, the softness of the surface, etc.. In the case that the co-efficient of friction is too high over all or part of the footwear, and the footwear is not slipping enough or locking on to the underlying surface too easily (which may for example increase torsional stress on the body, increase risk of falling or tripping, or other performance or injury risks) for all contact areas or partial contact areas, the generated footwear design is likely to include, for example, in the relevant contact area or areas, one or more of, outsole materials with lower co-efficient of friction, midsole or outsole materials that are harder and likely to conform less to the underlying surface, effectively decreasing the surface contact area, decreased outsole area, fewer lugs, shallower lugs, wider or narrower lugs, fewer spikes, shorter spikes, fewer studs, shorter studs, wider or narrower studs, etc. depending on the type of surface, the softness of the surface, etc.

[0200] The method can further comprise obtaining one or more activities such as sport activities, the one or more sport activities including an indication of the sport activity in which the user is intended to perform an action, for example, the action of running, jogging, walking, jumping, cutting / changing direction, squatting, lunging, or sports like football, athletics, basketball, or other relevant situations such as elderly at risk of falling, motor neurone or brain disease patients, safety in the work place for physical workers, security guards, first responders etc. A user intending to wear the footwear component when walking may require a shoe with a lower stiffness and greater flexibility than a user intending to wear the footwear component for a competitive sprint which may require footwear with a higher stiffness. Therefore, calculating the biomechanical model can be further based on the one or more activities so as to optimise a muscle group used when the user is performing the activity or to minimise a muscle group that is susceptible to injury during the performing of the activity. The biomechanical model will be calculated taking this factor into account. Multiple activities or actions may share the same improvement criteria (e.g., more efficiency during running and walking), and a single action may have multiple improvement criteria (e.g. higher efficiency and lower injury risk when running). In these cases the design elements / footwear parameters will be optimised to meet the improvement criteria(s) for all actions, and in the case that a design change will compromise one improvement criteria for the benefit of another, the design will be balanced to limit the compromise / benefit equally, or will prioritise the improvement criteria or action considered to be most important, either predetermined or as selected by the user, or will weigh the design changes according to a weighted order or priority for the improvement criteria(s) or action(s). e.g., increasing the stack height of a shoe may help increase stride length and subsequently running or walking speed or efficiency, whilst also increasing the cushioning of the shoe, potentially reducing loading or impacts and injury risks for the knee in both running and walking. In this case the design can be optimised to meet both an improvement criteria of running economy and injury risk for both actions / activities of running and walking, In another example, stiffening a carbon plate within a shoe, may improve the energy return and overall efficiency of an individual during running, but may increase loading / torques / forces experienced by the Achilles tendon during running. In this case the design may be optimised to compromise the positive effective of stiffening on economy for running with the negative effect on injury risk resulting in a somewhat less stiff resulting shoe that balances the benefit of the increased stiffness on efficiency with the benefit of reduced stiffness on injury to achieve a moderated result for both, or alternatively according to a weighting from predetermined defaults or according to user defined priorities, criterion or objectives.

[0201] An improvement criterion can be deduced by the system or can be inputted by the user using an input device. The improvement criterion can relate to objective data such as performance improvement, injury risk reduction, improved mobility and improved health and fitness, stronger muscles, lose weight, lower impact, reduce back pain, reduced local fatigue, reduced overall fatigue, increased comfort etc. For running performance, examples include higher speed or longer distance capability. For injury risk reduction, increase strength, increase comfort, reduced pain, etc. a specific region of the body can optionally be selected similarly to that described injury susceptibility data, such as foot and ankle or hip and back and for injury and / or pain additionally the type of injury to be avoided can also be indicated such as soft or hard tissue, joints, tendons, muscles, or ligaments etc. An improvement criterion relating to improved health and fitness can comprise improving muscle strength, muscle tone, hormone production, calorie bum, weight loss, joint mobility, endurance capability and / or cardiovascular health in the user.

[0202] In the above-described examples wherein anthropometric data, environmental data, injury data, sport activity data, objective criterion data etc are obtained, a priority weighting may be assigned to one or more of these factors. In this way, the obtained data is assigned a relative priority or importance value based on a level of significance. In some examples, this may comprise assigning a numerical value to each of the above-mentioned obtained pieces of data. In a specific example, a knee injury may be assigned a high priority weighting, while an Achilles tendon injury may be assigned a low priority weighting. These weightings can be accounted for when generating the one or more footwear parameters, as shown in Figure 10. In some examples, the weightings may be used in calculating the biomechanical model or a target biomechanical model.

[0203] Alternatively, a priority can be determined, pre-set or provided by the user to, for example, only generate designs that fully meet the criteria of efficiency, allowing limited adaptations to be made to the design that would benefit the improvement criteria of injury, (in the carbon plate example above this would result in a generated design that would “accept” the increased injury risk in order to meet the efficiency improvement criteria by stiffening the carbon plate to optimise running efficiency fully). Alternatively a weighting can be provided to the improvement criteria priority to, for example, account for at least 80% of the full potential of design elements that contribute to an improvement criteria such as efficiency, to allow for some wider degree of design freedom to better meet the injury criteria (in the carbon plate example this would mean the carbon plate stiffness could be reduced to the point of providing 80% of the full potential efficiency benefit if required to somewhat reduce the compromise to the injury improvement criteria). 0.8

[0204] Wherein, K_target represents the target in leg stiffness [I i] that should at least be met, K_potential represents potential leg stiffness [N / m] and K_current represents current leg stiffness [N / m], Similarly, a weighting can be provided to a damper constant, or for loading, forces, torques, moments, etc.

[0205] A comparable method can also be applied to scale damping, loading, force, torque, moment, stress, strain, Q factor, etc., and / or to prioritize loading or injury risk in specific locations in, or on, the body, body segments, joints, muscles, tendons, bones or other tissue

[0206] In the case that multiple factors must be prioritized a weighting or scaling factor can be applied to each. In one exemplary method, the priority of the factors will be translated into a hierarchy, with the highest priority factors addressed first, with compromises to the generated design made in order within the limits calculated by the weighting factor, ensuring that the top of the hierarchy is addressed to at least the target, and every subsequent factor potential target calculated on the basis of the limitations incurred from the higher ranked factors.

[0207] In an example case, a user requires footwear for running and has reported injuries in both the knee and the Achilles tendon and has input objectives of injury risk reduction and increased efficiency. The user has indicated a severity level of 1 for the knee injury, and a severity level of 3 for the Achil lies tendon injury. The user has indicated a preference for reduced injury risk over improved efficiency with a weighting of 60% to 40% respectively.

[0208] From the user’s motion data it is apparent that the user has a mid to forefoot landing, with a relatively high leg spring stiffness, and a biomechanical loading profile that shows a relatively high loading in the foot, Achilles tendon and lower leg, with a medium loading in the knee and upper leg, and a relatively low loading in the lower back. It is likely that, to reduce injury risk, it may be favorable to reduce (or limit as far as possible) loading, and to optimize efficiency it may be favorable to closely match the spring stiffness of the footwear to the leg spring stiffness of the user. Without any other information from the user or other sensor data, a weighting for the priority of the injury risk and the improved efficiency will be calculated accordingly. An exemplary calculation is shown below: 0.60 0.40

[0209] Wherein, in this case, F Target represents the target loading [N] that should at least be met, F Potential represents the optimal achievable loading [N], F Current represents the current loading [N], Q target represents the target Q factor, Q potential represents the potential achievable Q factor and Q Current represents the current Q factor.

[0210] Without any other information from the user or other sensor data, the relative priority of the separate reported injuries will then be calculated according to the severity, an exemplary calculation is shown below:

[0211] Knee injury severity = 1

[0212] Achilles injury severity = 3

[0213] Total severity score = 1 + 3 = 4

[0214] Knee injury priority weighting = 1 / 4 = 25%

[0215] Achilles injury priority weighting = 3 / 4 = 75%

[0216] Since the Achilles tendon injury has the highest priority, the calculated target biomechanical model will first aim to reduce Achilles tendon loading as much as possible, but at least to 75% of F Target.

[0217] From the example user, looking for a running shoe to lower loading in the Achilles tendon might, for example, result in a generated footwear component design that has a high heel to toe drop, that is likely to adjust the landing position of the user’s foot to a position that is further back from the current mid to forefoot landing, more towards the mid or rear foot, thus reducing the range of motion of the heel and stretch or deformation of the Achilles tendon, and moving the center of pressure of the foot further back proportionally reducing the moment and ultimately the forces experienced by the Achi Hies tendon, however with less loading and less stretch / deformation of the Achilles tendon over the time of the step it is likely that the forces experienced by the knee will have a relatively higher vertical loading rate and peak force. Therefore, to account for the lower priority injury in the knee, the resulting generated design would likely be modified to a slightly lower heel to toe drop that would reduce the loading in the Achilles tendon to 75% of the full potential reduction. As the 2ndpriority injury, and the next factor in the hierarchy, the design must now attempt to reduce loading at the knee within the existing design constraint introduced by the higher priority Achilles tendon injury, in this example case limiting the range of possible heel to toe drop adjustment. By influencing the spring stiffness of the footwear, and / or by increasing the damping, it may be possible to reduce the vertical loading rate and peak force experienced by the user, and therefore the knee, thus it is likely the generated design will additionally comprise a higher stack height with a mid-sole material that is relatively soft with a moderate spring stiffness and high damping with a Q factor tending towards critical damping. However, in this example case, a high damping and moderate spring stiffness, is likely to be sub-optimal for improving efficiency since this would results in a high level of energy loss, and a lower Q factor than might be possible with a lower damping and higher spring stiffness that would likely be preferable for an optimal rate of energy return for the example user. Therefore, to account for the lower priority objective to increase efficiency, the resulting generated design would likely be modified to a slightly stiffer spring stiffness and slightly reduced damping that would ensure at least 60% of the potential loading reduction is met whilst optimizing the energy return as far as possible without exceeding the target loading.

[0218] Alternatively, if a more balanced result is desirable, a regression may be used to calculate the optimum configuration for all variable footwear component design parameters to meet the newly calculated target values derived from the priority weightings as closely as possible, for example, by finding the least mean square difference.

[0219] In some examples, the method may include calculating a target increase or decrease in biomechanical models, as shown in Figure 13. The target increase or decrease is based on the biomechanical model, including at least one of the biomechanical load distribution and / or the stiffness. In a specific example, the target model may be calculated so as to simulate a reduction in one or more of the deformation, elongation, stretch and elastic energy stored in the Achilles tendon in order to account for a specific injury experienced by the user. By identifying target changes in the joint load or stability of the user, a footwear component may be generated based on the target biomechanical model.

[0220] In some embodiments, the user is one of a plurality of users. Motion data and other relevant data can be obtained in any of the ways described above for each user in the plurality of users. Preferably, the plurality of users comprises a sample group representative of a chosen population or demographic. For example, the population can comprise users of one or more of similar age, gender, race, height weight, similar injuries, common movement patterns, etc. In some examples, the users may be classified into populations by using a binning system, for example, a user with a weight of 78 kg may be classified into a population ranging from 70- 80 kg. Similarly, in some examples, users with hip pain and upper leg pain may be placed into the same categorised demographic.

[0221] An average biomechanical model can be calculated for the given population or demographic. In a specific example, an average biomechanical model can be calculated using data from multiple female users with a weight of 70-80 kg and a lower leg injury that habitually land on their forefoot with a cadence of between 140 and 160 steps per minute when running. Specifically, an average biomechanical load can be calculated for the given population and demographic. The biomechanical load will be calculated in the same way as described above for each user, such that estimated forces and moments are calculated for the segment(s) and average estimated forces and moments are calculated for the population using this data. Similarly, an average stiffness can be calculated by calculating a spring constant for the specified part of the body of each user and finding an average spring constant for the population.

[0222] A footwear component, for example, a single article of footwear can be generated for the plurality of users in a demographic based on the average biomechanical model. The generated article is therefore suitable for other users who fall within this demographic, allowing for a simplified system to be employed that obtains demographic data from a user such as age, gender, race, height weight, movement patterns and / or injuries of the user and generates the footwear component by comparing the demographic data with the population classification data, that is the data about each demographic, so as to identify a classification the user may be assigned to and provides the article according to this classification. This advantageously reduces the computation power required to generate a footwear component.

[0223] Having calculated the biomechanical model, the method includes generating a footwear design comprising one or more footwear parameters according to the calculated biomechanical model for one or more users and the improvement criterion, as shown in Figure 13. The one or more footwear parameters define a physical characteristic. The physical characteristic can be a mechanical or physical characteristic of the footwear component such as an insole, a midsole, an outsole, studs or spikes or lugs, plate (plastic / carbon fibre / glass fibre / nylon etc.), an upper, laces or fasteners, a truss, a sock-liner, a heel counter, a heel collar, component topology, a stack height, a heel to toe drop, an anti-pronation post, a heel lift, a metatarsal lift, a metatarsal spreader, an arch support, dualdensity foam, or a tongue with specific mechanical properties.

[0224] In a specific example, the parameter may be a stiffness / support parameter defining a physical characteristic that is a midsole with a low density configured to compress under pressure, thereby reducing the load on the body and improving the running style of the specific user. There are many other footwear parameters that may be generated, for exemplary purposes, a heel to toe parameter and a heel stack height parameter are described in detail below.

[0225] A heel to toe drop parameter represents the difference in height between heel and forefoot in a footwear component. In running shoes, the heel to toe parameter will have a value that is typically in a range of 0-14 mm. Typical categories for the heel to toe drop parameter are: zero drop (0 mm), low drop (1-4 mm), mid drop (5-8 mm) and high drop (8+ mm). In other examples, the categories may be low drop (0-8 mm) and high drop (8-13 mm). Since the heel to toe drop parameter affects the forces acting on different parts of the body, the selected heel to toe parameter of the footwear component has the potential to change aspects of the users running style such as their cadence, foot strike, strike angle, maximum ankle moment, maximum knee moment, ankle planta / dorsiflexion moments and net joint ankle flexion. In this way, the performance of the user during motion can be improved by generating an appropriate heel to toe drop for the user, according to the biomechanical model and the improvement criterion for the specific user.

[0226] If the user has any injuries, it is preferable that uninjured parts of the body absorb the impact when the user is in motion or locomotion. For example, a lower heel to toe drop will load the ankles, calves and Achilles of the user more than a high heel to toe drop and allows for more ankle flexion during landing. Further, a lower heel to toe drop promotes a foot strike more towards the midfoot and forefoot, which can aid with pre-existing injuries such as anterior knee pain, gluteal overuse syndrome, upper leg or lower back injuries. Low heel to toe drops places greater stress on the foot, ankle, lower leg. Alternatively or conversely a higher heel-toe drop can also be used to help reduce loading on the Achilles and calf muscles, thus potentially relieving pain, or reducing injury risk of these structures.

[0227] The calf muscles, Achilles and associated connective tissue behaves elastically and can be modelled as a spring. A higher drop elevates the heel and can reduce ankle flexion during landing which limits how far these structures will be stretched, and thus reduces the potential for elastic energy to be stored. Conversely the midsole of the heel will be compressed more thus increasing the potential for elastic energy to be stored within the footwear. By adjusting heel-toe drop and adjusting the materials used in the construction of the mid-sole, it is thereby possible to directly influence the amount of elastic energy that can be stored during landing, and subsequently returned to the user during propulsion. Further it is possible to influence where elastic energy is likely to be stored, which may be relevant, for example, for an individual experiencing pain in their relatively short, stiff, Achilles tendon, it may be favourable to reduce the stretch and elastic energy stored in the Achilles tendon, in preference to store elastic energy within the materials of the midsole of the footwear.

[0228] In this way, by generating a heel to toe drop parameter according to the biomechanical model, which can account for a pre-existing injury of the user that has been deduced from the biomechanical load distribution or has been obtained by user input, user’s motion or locomotion can be improved. Further, generating a heel to toe drop parameter according to an improvement criterion that in some examples is directed to reduced injury can improve the locomotion or motion of the user.

[0229] In a further example during a squat motion, heel to toe drop can influence the range of motion of the knee, ankle and hip, and the activity of different muscles or muscle groups such as the calves, quadriceps, glutes, hamstrings or back and core stability muscles. A higher drop will allow the user to support their weight more towards the heel without requiring a large range of motion around the ankle, reducing the stretch required from the Achilles tendon, and reducing the moment around the ankle, thus relieving the force or load in the lower leg muscles and tendons In turn, this generally allows the user to utilise a wider range of motion around the knee and hips, enabling the user to use bigger muscle groups such as the quadriceps and glutes, more effectively for longer periods of the squat. In this way heel to toe drop can be optimised to increase or reduce loading on or generated by specific muscle groups, tendons, other tissues or any relevant footwear component materials, or to influence the efficiency of the motion, or to influence where energy may be stored or released during the motion from specific muscles, tendons or other tissues, or materials from any relevant footwear component, or influence the stiffness ordamping, of any part of the biomechanical model. In this way, by generating a heel to toe drop parameter according to the biomechanical model, which can account for pre-existing injuries or injury risks of the user that has been deduced from the biomechanical load distribution or has been obtained by user input, user’s motion can be improved. Further, generating a heel to toe drop parameter according to an improvement criterion that in some examples is directed to increase strength of a specific muscle group or muscles groups such as the quadriceps or glutes, can improve the motion of the user.

[0230] A weight parameter relates to the weight of the footwear component and can influence the inertia of the foot and leg and the amount of energy required to move the foot. In the case of running a heavier shoe will require more energy to lift and move during each step and will also increase the inertia of the leg, that can slow the acceleration and deceleration of the foot, potentially reducing the rate at which steps are made, which in the case of running is likely to result in a lower running cadence.

[0231] A stiffness and / or softness parameter relates to a deformation of the physical characteristic. The stiffness and / or softness parameter can be categorised the physical characteristic into categories of soft, medium soft, medium, medium firm and firm. Softer shoes, with a midsole that compresses under pressure, typically have lower shore values and can increase the dampening effect or dampening constant (or damping coefficient), which can reduce the rate of deceleration and extend the duration of the collision between foot and ground (the foot strike) potentially reducing the peak forces and load on the body while running. During the propulsion phase, the same footwear design providing the same soft damping surface can also have a reduction effect on acceleration, potentially requiring relatively more muscle energy to create propulsion sufficient to maintain running speed. Whilst also relevant for many activities, in the case of running, some research studies (for example Soft-tissue vibration and damping response to footwear changes across a wide range of anthropometries in running, Behling et al. 2021 , https: / / journals.plos.org / plosone / article?id=10.1371 / joumal. pone.0256296) have shown that softer footwear with lower shore values, are generally correlated with higher damping coefficients in many runners. However, since many materials utilised in footwear design can be more elastic or viscoelastic in nature, stiffness or softness and damping effect of the footwear should also be taken in consideration with the elasticity or spring constant (or spring coefficient) of the footwear. Optimally the loading and unloading time, or the conversion from potential energy, stored as the material is compressed, to kinetic energy, released as the material reforms, which may be modelled as the oscillation of the system, should match that of the users body or structures of the body engaged during the motion, to ensure that there is some synergy or resonance between the footwear and the user to ensure a greater degree of energy return during the propulsion phase. As is typical in viscoelastic systems, damping and spring coefficients are dependent on the timing and magnitude of loading, as well as the tissue or material properties, which in the case of, for example, human users, can be complex and highly varied between individuals, footwear types and conditions of use. In many cases increasing the damping coefficient will negatively affect how much energy return is possible during the propulsion phase of a movement or activity, and thus is likely to require optimisation to maximise the potential for reducing and spreading loads and forces, through tuning damping effects, and maximising the potential for energy return during the propulsion phase through tuning spring stiffness. The complexity and compound effects of materials and design means, in some cases, it may be relevant to further categorise or separate the stiffness or softness parameter to represent one or more of, the shore value, the spring constant (or spring coefficient), and the damping constant (or damping coefficient), to ensure more representative or accurate results. If the stiffness is calculated to work in phase with the user’s propulsion phase, for the user’s particular way of moving, then the stiffness parameter may be generated as part of the footwear component design to achieve an objective or improvement criterion related to, for example, improving running performance, or increasing running efficiency. If the user has reported injuries, or an objective or improvement criterion that additionally relates to, for example, reducing injury risk, then the stiffness or softness parameter is likely to be calculated to optimise the balance between spreading or reducing forces, loading or energy, and reducing energy loss and maximising the potential energy return by optimising the damping and spring coefficients of the footwear to provide a balance in the efficiency profile and injury risk profile of the footwear. This may be further optimised to reflect a weighted priority, for example, if injury risk is given a higher priority than efficiency, it is likely the resulting footwear design will compromise spring stiffness in favour of improved damping.

[0232] A stack height parameter represents a height of material between the foot of the user and the ground when the user is wearing the footwear component. Often, the stack height parameter will affect how “intensely” the user will feel the ground through the article. Typically, the stack height will range anywhere from 3-50 mm. Typical categories for stack height are: barefoot (3-8 mm), minimalist (3-13 mm), regular (9-29 mm) and maximalist (30-50 mm). In other examples, the stack height may be categorised as minimalist (0-23 mm), regular (23-28 mm) and maximalist (28-40 mm) in some cases extreme maximalist shoes are also available in which the stack height is over 40mm. The stack height may be determined by the sum of the thickness of each individual component on the lower portion of the footwear component, for example it may be the sum of the thickness of the insole, the midsole and the outer sole. In a specific example, a barefoot stack height may comprise a particularly thin midsole of a few mm such that the overall stack height of the shoe is in a 3-8 mm range.

[0233] For a minimalist stack height, there is less midsole material to spread forces over a larger area or over a longer time, therefore, the foot, leg and body of the user will likely experience forces more directly, with higher vertical loading rates, quicker decelerations and thus potentially higher impacts and forces when the user’s foot contacts the ground and therefore, if the user has an injury that might be vulnerable to such effects (e.g. a tibial stress fracture), a minimalist stack height will not be generated. For users without such an injury, a minimalist stack height is likely to reduce the amount of material in the midsole and thus can reduce the weight of the shoe, and for well-trained users sufficiently experienced to have sufficient leg stiffness and / or strength, it may be possible to utilise the forces to store elastic potential energy in tissues of the foot, leg and body that may be returned during the propulsion phase. These factors are likely to improve the user’s running economy and performance, and so a minimalist stack height may be generated as part of the footwear component to achieve this objective or improvement criterion. Alternatively, higher stack heights are likely to increase the amount of material in the midsole and thus can increase the weight of the shoe, which will in turn take more energy to lift, but if the material is a damping material this will mean more energy is lost, that may be beneficial in reducing the impact experienced by the user. These factors are likely to spread loading and potentially help reduce pain or injuries and so may be generated as part of the footwear component to achieve an objective or improvement criterion related to, for example, reduced knee pain, or reduced injury risk. However, if the additional stack height material is a more elastic material, it will deform and store energy as potential energy in the midsole similar to a spring, which may be returned when the material reforms. If the timing of the energy return is synchronised with the propulsion phase of the user, this may help benefit the efficiency or running economy of the user. In addition, a high stack height may help add length to the user’s leg enabling them to make bigger steps which may allow the user to use relatively fewer steps to travel the same distance. If the elasticity and additional height is calculated to contribute more to increasing the user’s running efficiency than any reduction the additional energy cost of lifting the heavier weight of the shoe may have for the user’s particular way of moving, then a maximalised stack height may be generated as part of the footwear component to achieve an objective or improvement criterion related to, for example, improving running performance, or increasing running efficiency.

[0234] A stability or support parameter relates to footwear component design aspects that influence the motion of the foot during contact with the ground. This may relate to the flexibility or torsional stiffness of the shoe, or topographical features like arch-support, metatarsal spreaders, or lifts, or additional supporting structures such as heel counters, or motion control or guidance systems or technologies such as dual density foams, trusses, anti-pronation posts, or other forms of reinforcement within the footwear. Stability or support can be categorised further into, for example, the type of support and the amount of support provided. Type of support may include, lateral, neutral, or medial support, and may further be considered with regards to the location of the support, for example, one or more of, rearfoot, heel, midfoot, arch, forefoot, or any combination that may include guidance, motion control, or support throughout the entire footwear. The amount of support may be further categorised to qualify the degree of support provided, for example, extreme, very high, high, mid or medium, light, very light, extremely light or no support. Support and stability can be highly relevant for a variety of activities, exercises and sports. For example in running, particularly relevant can be the rate and range of pronation (or supination), and inter-step variability parameters, especially for runners that use their heel during the stance phase. Increased support will generally help control the rate and / or range of pronation, research (Malisoux et. al. 2016) has shown that, for recreational runners, particularly those that pronate, injury risk can be reduced by introducing support technology designs such as dual density foams. However, too much support has been shown in some research studies to increase risks of injury e.g. ( The effect of three different levels of footwear stability on pain outcomes in women runners: a randomised control trial, Ryan et al. 2010, https: / / bjsm.bmj.eom / content / 45 / 9 / 715.info), and thus the degree of support should be optimised to the user’s way of moving. Increasing the support in the shoe will influence the rate and range of motion of the foot and ankle during a movement. This will influence the pressure, forces, moments etc. that can be calculated by a biomechanical model. For a beginner runner user that has excessive lateral or medial motion of the knee or foot or hyper-pronation of the foot during a step a dual density foam construction, with firmer foam built into the medial side of the footwear component, may help to reduce the rate and range of this motion to a more optimal rate and range for the user’s particular way of moving. These factors are likely to manage loading and potentially help reduce pain or injuries and so may be generated as part of the footwear component to achieve an objective or improvement criterion related to, for example, reduced knee pain, or reduced injury risk. In the example of running, in some cases reducing lateral or medial motion may also help reduce energy lost in these undesirable motions, ensuring energy is primarily used to move forwards potentially also resulting in a higher efficiency or improved running economy. Any of the above-mentioned parameters may be included in the generated footwear design.

[0235] The one or more footwear parameters can be generated using a machine learning model.

[0236] One type of machine learning model that can be used is an unsupervised regression model. This can comprise a deep neural network (DNN) with a minimum of 3 layers, including 1 input layer, 1 or multiple hidden layers and one output layer. Input data may be split in 2 sets, 80% for training and 20% for testing (validation). During training, a 10-fold cross validation technique may be used to estimate how well the model will perform on unseen data. The technique randomly splits training data in 10 subgroups, the DNN may then train on the 9 subgroups, and attempts to predict performance scores for the 1 subgroup that was left out. An evaluation score may be calculated between the true and predicted efficiency index of the left out subgroup. The technique iterates until all 10 subgroups have been used as the left out subgroup (10 iterations), preferably / optionally the whole procedure may be repeated multiple times, for example, 3 times, meaning 3 times random subgroups of 10s. In this case, the final evaluation score is the average of all 30 (3 repeats x 10 splits) iterations.

[0237] Preferably / optionally a k-fold cross validation procedure may be implemented to estimate the skill of the model on unseen (test) data. The general procedure is as follows [3]:

[0238] 1 . Shuffle the dataset randomly

[0239] 2. Split the dataset into k groups

[0240] 3. For each unique group: a) Take the group as a hold out or test data set b) Take the remaining groups as a training data set c) Fit a model on the training set and evaluate it on the test set d) Retain the evaluation score and discard the model

[0241] 4. Summarize the skill of the model using the sample of model evaluation scores

[0242] When using 10 splits (10-fold cross validation) the average accuracy of the model was shown to be increased by ~1 %, and based on literature research the optimal k is between 5 and 10.

[0243] The input layer can comprise of raw or processed sensor data. When training the DNN, data or information of a user’s anthropometries when wearing a known or pre-existing footwear design may be used. In other words, a user may carry out a session in a pre-existing footwear design, and measurements and information will be obtained from the sensors which can then be fed into the input layer of the DNN. Table 1 below shows exemplary raw or processed input data of seven users comprising their user anthropometries or information describing the motion of each user, including cadence, contact time, flight time, stability, stride time, stride length, step length, balance, flight time / contact time ratio amongst others. Table 1

[0244] Since features or parameters are likely to be of different scales, data is preferably scaled to correct for this variability. Normalization is one method that may be used to bound values between two numbers, typically, between [0,1] or [-1 ,1]. Standardization transforms the data to have zero mean and a variance of 1 , so data is made to be unitless [1], Table 2 shows the input anthropometric data from

[0245] Table 1 once it has been normalised to two decimal places.

[0246]

[0247] Table 2

[0248] Tables 1 and 2 show exemplary input data obtained for seven users carrying out a session comprising running on a treadmill for a given period of time. In practice, the DNN is trained on a much larger dataset comprising input data from a large number of users each carrying out a session. The number of steps (taken by a user in motion) during a session can very, for example, a session may comprise 50-100 steps taken per user. As such, in one specific example, the DNN may be trained using a database comprising 85 million steps obtained from the sessions carried out by the users.

[0249] Various neural network analysis methods may be utilised, including classification or regression. In the example of running, one option to enable regression analysis is to calculate continuous unitless indexes in the range of 0.0 to 1 .0 representative of running efficiency and running related injury risk. Table 3 comprises unitless indexes calculated for each of the exemplary seven participants representing a flight time over contact time ratio (FTCT), a running efficiency and an injury risk.

[0250] Table 3

[0251] One example method to calculate an efficiency index comprises the parameters of flight time and contact time since these have been shown in research to be indicative of running efficiency independently or as (parts of) functions representative of, for example, leg stiffness during running. In an exemplary calculation the equation for efficiency index may be calculated as follows:

[0252] Flight time

[0253] Efficiency Index = 0.75 x - 1- 0.25 % (1 — Contact time)

[0254] Contact time

[0255] In this exemplary case, efficiency index ranges between 0 (least) and 1 (best).

[0256] In some examples, an efficiency index may further, or alternatively be a function of energy, power or impulse, in which a lower energy use, power, or impulse for the same motion, movement or action, performed at the same rate or speed, is correlated to a higher efficiency. For example, in the case of a foot striking and leaving the floor, in a manner that may be modelled as a collision, mechanical energy may be calculated based on measures of force, time, mass, and / or velocity as shown below:

[0257] Impulse = F x t

[0258] Impulse = A momentum

[0259] F x t = m x A v

[0260] F x t A v = - m (F x t)2

[0261] Energy = -

[0262] 2m

[0263] In some examples, the biomechanical load distribution further comprises an injury risk of the user that may be calculated from statistical correlations between a single metric derived from measured movement data and reference data from a database of user data or from scientific literature such that the metric may serve as an indirect indicator of injury risk. One example method to calculate an injury index comprises the parameters of inter-step variability or stability and cadence since these have been shown in research to be indicative of running related injury risk independently or as (parts of) functions representative of, for example, muscle recruitment of loading during running. Further the injury risk may be calculated from statistical correlations between multiple metrics derived from measured movement data and reference data from a database of user data or from scientific literature such that the combination of metrics may serve as indirect indicators of injury risk.

[0264] One way to calculate such a correlation is to utilise a mathematical function such as sum, multiply, divide, whilst optionally applying a weighting factor or off-set to each of the input metrics to calculate a resulting, potentially unitless, injury index in the form of a single integer that can be correlated to reference data and thus used as an injury risk indicator. An example of this approach would be to use the inputs of inter-step variability and cadence (since higher cadence at the same speed under certain conditions is correlated to lower risk of injury during running) as used in the following exemplary calculation:

[0265] In an exemplary calculation the equation for injury index may be calculated as follows:

[0266] Injury Index = 0.75 (1 — VariabilityX^) + 0.25 % Cadence

[0267] Where Variability is the Inter-step variability. Injury risk indexes range between 0 (higher risk) and 1 (lower risk). Flight time over contact time ratio or a lower contact time are used as performance-efficiency indicators, while step length and footstrike Y need to be part of a broader metrics picture (correlation to the other metrics) to estimate performance and efficiency. Similarly, the metrics used as injury indicators are 4, lower stability X and Y, higher cadence and closest to 0 balance, respectively.

[0268] The output of the DNN may be the generated one or more footwear parameters. Table 4 shows exemplary efficiency scores given per footwear component design per user as an output of the DNN.

[0269] Table 4 In Table 4, each exemplary footwear component has been assigned a numerical value between 0 and 4, as shown in the first row. It should be understood by the skilled person that these values are place holders representing any one or more of the mentioned footwear parameters.

[0270] To evaluate DNN regression analysis results, the mean absolute error (MAE) may be used to measure the difference between the predicted and true performance values of the known pre-existing footwear component (25 for SMOTE option, 10 for SDV option).

[0271] In examples where the method comprises comparing the generated footwear design to each pre-existing footwear design from a database of pre-existing footwear designs, and a pre-existing footwear design may be identified. For example, a heel to toe drop parameter may be generated to define a physical characteristic having a heel to toe drop of 7 mm, this heel to toe drop value will then be compared to heel to toe values of the pre-existing designs. In this specific example, the closest pre-existing heel to toe value to 7 mm may be a pre-existing design with a value of 8 mm, such that this pre-existing design will be identified. This method can be repeated for each parameter, and the optimal pre-existing design with footwear parameters most closely matched to those of the generated footwear design will be identified. In other words, a footwear design is identified that most closely matches those of the generated footwear design.

[0272] In order to determine the most closely matched pre-existing footwear design, a pre-design design may be identified that optimised the improvement criteria(s) for all actions, and in the case that a pre-existing design will compromise one improvement criteria for the benefit of another, the design identification will be balanced to limit the compromise / benefit equally, or will prioritise the improvement criteria or action considered to be most important, either predetermined or as selected by the user, or will weigh the design changes according to a weighted order or priority for the improvement criteria(s) or action(s).

[0273] Alternatively, the generated parameter may be categorised into one of the categories outlined above, for example, 7 mm may be assigned to a medium drop category (5-8 mm), and the user will be provided with a generalised footwear design with a medium heel to toe drop. In this way, there may be a number of preexisting designs that are in this category, giving greater choice and footwear options to the user. In some examples, the pre-existing designs are ranked in order of how closely they match the generated design. The highest-ranking preexisting designs may then be identified. Further / optionally specific footwear design components may be prioritised or placed in a hierarchy to further prioritise the closest matching pre-existing designs.

[0274] In an example case, a female user that has reported Achilles tendon pain is measured with pressure sensitive insoles and foot mounted I Mils during running. In this example case the users measurements show the user is a forefoot lander, with a leg spring stiffness of around 2.35kN typical of a less trained beginner runner in minimal neutral footwear, which, according to relevant statistical data, is estimated to reduce by 5% to 10% to around 2.2kN in softer more built up shoes, and a biomechanical loading profile that calculates a relatively high proportion of the forces and torques are located in the lower leg, with high activity in the calf muscles, and a Achilles tendon and foot experiencing a lot of load during each step. A shoe design is calculated that is expected to reduce the loading on the lower leg and through the Achilles tendon, with a elasticity or spring / damper profile that is expected to resonate with the calculated leg spring stiffness of the runner. The resulting generated design comprises a high heel to toe drop of 11 mm, that has been calculated is likely to help reduce the biomechanical loading in the lower leg and Achilles tendon, with a maximalist stack height of 30mm comprising a midsole of medium-soft EVA foam with Shore A value of 19.5 and spring stiffness of 2.2kN, that has been calculated is likely to help spread the rate of loading over a longer period and store and return energy in phase with the users leg-spring stiffness to contribute to the propulsion phase of the runner, a light support system with medium-hard polyurethane reinforcement with elastic moduli of 40 MPa on the medial side of the midsole, that has been calculated is likely to help provide more guidance or stability during each step to help reduce medial / lateral range of motion thus helping to reduce loading or force on the medial side of the ankle and knee to helping to reduce torsion and potentially injury risk in the Achilles tendon and additionally help reduce propulsion impulse, expected to help reduce fatigue, thus potentially improve performance.

[0275] At the location where the user has been measured only pre-existing footwear designs are available so a calculation is required to rate which pre-existing footwear designs most closely match the calculated footwear component design.

[0276] Preferably, as mentioned above, a useful first step is to translate the design to a specification. In the given example In the given example the resulting footwear component design can be translated to the following specifications:

[0277] Heel-toe drop: 11 mm (High) Stack height: 30mm (Maximalist)

[0278] Softness (Shore A): 19.5 (Medium-soft)

[0279] Spring constant: 2.2kN (Low-medium responsiveness)

[0280] Support level (0-5): 2 (Medial 40 MPa PU) (Light-mid support) Support type (1-3): 1 (Medial 40 MPa PU) (Light medial support)

[0281] Comparably the pre-existing footwear component designs can be translated into comparable specifications. This step can be performed by a user measuring or assessing the designs themselves, or by sourcing product specifications from the manufacturer or distributor, or may be gleamed from the internet from online databases, product reviews, third party service providers, or any other relevant applicable method. In the given example 3 pre-existing footwear designs are available at the user’s location with the following specifications:

[0282]

[0283] Table 5

[0284] Table 5 shows exemplary parameters with exemplary values and their corresponding categorisation.

[0285] The specifications of the pre-existing footwear can then be compared with the generated footwear component design specifications. If an exact match exists this option will be suggested to the user, if multiple exact matches exist then all matches may be suggested to the user, allowing the user to choose based on their personal preference, or alternatively the suggestion may be prioritised or ranked according to other metrics such as cost, stock level, brand, material types, suitability for specific surfaces or environments (e.g. trail, water proof, etc.) or other relevant metrics that may be defined by a user, or by predetermined defaults, or according to additional available inputs. If no exact match exists then a comparison will be made to assess which existing footwear is the closest match. For this a calculation or scoring system may optionally be implemented that may for example calculate the closeness of a match according to a statistical method such as a subtraction, least square regression, or other statistical method, to compare the absolute values (e.g. 10mm, 2.1 kN, etc.), or numerical representations of classification categories, for example, heel to toe drop may be categorised into zero, low, mid, and high drop, with a numerical representation between 0 and 3 in which zero = 0, low = 1 , mid = 2, high = 3, for support type this may be categorised into medial, neutral and lateral with a numerical representation between 1 and 3 in which medial = 1 , neutral = 2, lateral = 3, for support level this may be categorised into, no support, light support, medium support, high support, with a numerical representation between 0 and 5 in which no support = 0, light support = 1 , light-mid support = 2, mid support = 3, mid-high support = 4, high support = 5, etc. Any numerical values may also optionally be normalised, standardised, or calculated relative values to make comparison more representative, accurate, effective or simpler to compute. The closeness of the match maybe performed per specification so the closest matching footwear can be identified within a single specification category, for example, heel to toe drop values are compared between each shoe, and then stack height values compared between each footwear etc. to establish closeness values or scores according to each specification category, or alternatively, a single closeness value or score can be calculated per footwear by summing, subtracting, multiplying or dividing, the closeness values or scores per specification category. Optionally specific categories can be prioritised, for example, by applying an additional weighting or scaling factor.

[0286] In the given example it is possible to apply a simple scoring system in which the available existing footwear design specifications are ranked per specification category according to the modal difference to the generated footwear component design specifications with the ranks summed and the lowest scoring footwear design, or designs in case of matching scores, can be suggested to the user as illustrated in Table 6:

[0287] Table 6

[0288] This approach would result in providing the user with a recommendation for Shoe A as the closest match. Optionally it may also be useful to provide further suggestions regarding the closeness of the match for the remaining shoes, or a sub-set of those that are within a certain range of closeness, or the rank order, scores, relative or normalised percentage matches to allow the user to make more informed choices or decisions based on the feedback from the available selection of pre-existing footwear designs.

[0289] In a further embodiment, the closeness rating, may also be used to generate footwear component design modification for the pre-existing designs that would increase the closeness of the match. In the above example, Shoe B, may be further modified by inserting an insole with an additional 3mm heel to toe drop, bringing the modified heel to toe drop to 11 mm, exactly matching the generated footwear design, and with the additional insole also providing additional medial support, it may be possible to improve the match even further for the support categories. Thus with the addition of the extra insole, Shoe B could provide a better match than Shoe A without any modification. Further, applying the same method to Shoe A could potentially help further improve the match to create an even closer match with an additional insole or heel lift of only 1 mm. In this embodiment the existing footwear design suggestions would be provided to the user together with additional potential footwear design modifications that may further improve the closeness of the match.

[0290] Alternatively a generated footwear component design may be compared to preexisting designs utilising the dimensions and shapes of the footwear component designs or parts of the designs. Using statistical methods such as least square regression to established which dimensions, shapes or lines are closest matching to the generated design. This approach would be particularly suitable in the case that the output is a digital 2D or 3D CAD file, drawing, schematic, or another graphical medium. In this case it may be favourable to normalise designs to eliminate sizing differences e.g. to facilitate comparing a UK size 6 footwear design with a UK size 10 design.

[0291] Figure 14C shows illustrates ways in which the method can comprise providing the footwear design to the user. In some examples, a written or audible description of the one or more of physical characteristics of the footwear component to the user. In some examples, a 2D or 3D visualisation of the generated footwear design is provided to the user is displayed to the user. Accordingly, the system can further comprise a display screen coupled to the remote computer via a wire or wirelessly.

[0292] Figure 15 illustrates an exemplary method according to the present invention. Specifically, Figure 15 depicts the interconnection between the processes described in Figures 10-14, which collectively comprise the method of the present invention. Shown are the outputs from one process being fed into the next process as inputs.

[0293] Figures 16A, 16B, 16C and 16D illustrate an exemplary display screen displaying the generated footwear design to the user. In each figure, the generated article of footwear is schematically shown in an exploded view. In this way, the user can visualise the various components which form the footwear component as generated by the method and system. In the specific example shown in Figure 16A, a biomechanical model has been calculated for a user with a midfoot strike and who is an inexperienced runner. There is no indication of injury. Based on the biomechanical model, the system has generated a design for a footwear component with a low heel to toe drop, a high stack height and a soft shoe. The low heel to toe drop is represented in the displayed illustration using a line indicating that the insole be at a low, acute angle of inclination. The high stack height is represented by the thickness of the insole in the displayed illustration. The stiffness of the footwear is represented by the compression of the springs in the displayed illustration. The user can use an input device to select the various components of the shoe so as to configure the display screen to display more information or options for that component.

[0294] Having generated the footwear component, the method can further comprise manufacturing components of the article of footwear, as shown in Figure 14B. File types for manufacturing the design may include at least one of CAD, STL, VRML, ipt, sldprt, dwg etc. Materials for manufacturing include EVA, PU, gels, silicone, foam, air pockets, rubber, carbon, carbon fibre, fiberglass, synthetics (nylon), plastics, bamboo, cotton, algae, piezo-electric / resistive materials, dual-density mid-sole, anti-pronation post, etc.

[0295] The footwear industry is traditionally and typically divided into seasons such as spring, summer, autumn, winter, low, peak, etc. and / or regions / territories, such as Germany, US, Benelux, Europe, ASIA, EMEA, etc. for which specific design considerations are often considered to tune the performance of the footwear towards the predominate conditions, surfaces, market conditions, economic conditions, regulations, etc. , and it may be favourable to sell particular shoe models prior to, during or after these seasons. Therefore, in a preferred embodiment, season and location, will be inputs provided by the user, or automatically determined from the time and / or location of use of the system / method / software / hardware. The season and location input will be used to optimise the footwear design, for example, to limit the design freedom to selected number or types of materials that may be, for example, more sustainable materials, less costly, or have particular favourable properties such as being waterproof for wet seasons, locations or conditions, heat resistance for high temperature seasons, locations or conditions (e.g. running on asphalt in the summer in the middle east), highly durable for seasons, locations or conditions that may be particularly extreme, etc.

[0296] Any one or more of the motion data module, the biomechanical modelling module, and the design module may be implemented on any of the following computer components. A processor, which may exemplarily comprise a central processing unit (CPU), a microcontroller, and digital signal processor (DSP), configured to execute specific algorithms for motion data analysis for example. The processor may include one or more arithmetic logic units (ALUs), registers, and control units. A co-processor may be included for handling specific tasks such as floating-point calculations, signal processing, or cryptographic functions. The co-processor may include specialized ALUs and control units. Memory, which may be volatile memory (e.g., RAM) and non-volatile memory (e.g., ROM, flash memory) for storing calculation algorithms, intermediate data, and biomechanical model data, biomechanical load distribution data, footwear component design data, or any other information forming part of the methods described herein. Memory may be communicatively coupled to the processor via a memory bus and may include memory controllers. A field-programmable gate array (FPGA) may be provided, programmed to perform the specific calculations of the module. The FPGA may include configurable logic blocks, interconnects, and input / output blocks. An application-specific integrated circuit (ASIC) may be provided for performing particular calculation tasks related to motion data analysis, biomechanical modelling, or component design. The ASIC may include dedicated processing units and memory for specific algorithms. A digital signal processor (DSP) may be provided for real-time signal processing tasks. The DSP may include specialized ALUs, memory units, and control units. A graphics processing unit (GPU) may be provided, for training of machine learning models. The GPU may include multiple processing cores and memory interfaces. Additionally the motion data module may comprise any one or more sensor components configured to obtain the motion data, with the aforementioned computer components being configured to control those sensor components and / or process the monitored motion data and / or provide that data to the biomechanical modelling module.

[0297] The design module may be configured to output the generated component design in a manner that makes the design available, usable, or accessible in physical or virtual form. The design module may comprise any one or more of the following hardware components: a display, which may be a screen (e.g., LCD, OLED, E- ink) for visually presenting the generated component design. The display may include a display driver circuit for controlling the screen; a speaker for providing auditory feedback or alerts, which may include a digital-to-analogue converter (DAC), an amplifier, and / or a speaker driver circuit. A haptic feedback module is a motor or actuator that provides tactile feedback through vibrations or movements. The module may include a control circuit for managing haptic feedback. A wireless communication module, such as Bluetooth, Wi-Fi, or NFC modules, may be configured for transmitting the generated component design to external devices. Each module may include a transceiver, antenna, and supporting circuitry for communication protocols. Awired communication interface may include interfaces such as USB, serial ports, or other wired interfaces for direct data transfer. The interface may include connectors, transceivers, and protocol controllers. A printer interface may be provided that connects to a printer for producing physical copies of the generated component design. The interface may include a print controller and relevant drivers.

[0298] The manufacturing module described in relation to some implementations may comprise, in addition to any of the computer components described above, on which the module may be implemented, any one or more of the following hardware components: a 3D printer, which may include a print head, build platform, control electronics, and software for managing the printing process; a computer numerical control (CNC) machine configured to shape materials into a desired form according to a design. The CNC machine may include motors, control software, tool heads, and a control unit. A laser cutter may be provided to cut or engrave materials based on the generated component design. The laser cutter may include a laser source, mirrors, lenses, and control systems for precise cutting. An injection moulding machine may be provided for producing plastic parts by injecting molten plastic material into moulds. The machine may include an injection unit, mould assembly, cooling system, and control electronics. An assembly robot may be provided for assembling components into a footwear component according to a generated design. The assembly robot may include robotic arms, grippers, sensors, and a control system for precise operations. A quality control system may be provided, comprising a sensor arrangement or camera arrangement for inspecting the manufactured object to ensure it meets specified criteria. The system may include image processing software and defect detection algorithms.

[0299] The modules and components of the system may be configured to be in data communication with each other communication pathways that may comprise any one or more of the following: bus systems, such as I2C, SPI, UART, and CAN bus for data transfer between sensors, processors, memory, and other components. These buses may include data lines, clock lines, and control lines. Wireless communication may include Bluetooth, Wi-Fi, and NFC modules for wireless data transfer between the system and external devices. These modules may include transceivers, antennas, and protocol stacks for managing communication. Wired communication may include USB, Ethernet, and serial port interfaces for wired data transfer. These interfaces may include connectors, transceivers, and protocol controllers. Peripheral interface controllers may be dedicated controllers for managing communication between the main processor and peripheral devices such as displays, speakers, haptic feedback modules, and printers. Memory interfaces may include memory controllers and interfaces such as DDR, LPDDR, and flash memory controllers for data communication between the processor and memory units. Inter-integrated circuit (I2C) may be a multi-master, multi-slave, packet-switched, single-ended, serial computer bus used to connect low-speed peripherals to the processor and other higher-speed components. Serial peripheral interface (SPI) may be a synchronous serial communication interface used for short-distance communication, primarily in embedded systems, for connecting sensors and other peripherals. Universal asynchronous receivertransmitter (UART) may be a hardware communication protocol that uses asynchronous serial communication with configurable baud rates, used for communication between the processor and other modules.

Claims

CLAIMS1. A computer-implemented method of generating a design for a footwear component, the method comprising: obtaining motion data derived from monitoring motion of a user; calculating, according to the motion data, a biomechanical model of at least a portion of the body of the user, wherein the biomechanical model is representative of a biomechanical load distribution through at least the portion of the body of the user and wherein the portion of the body of the user comprises one or more anatomical structures of the body other than a foot of the user; and generating, according to the biomechanical model and a first criterion, a footwear component design comprising one or more footwear parameters, each footwear parameter defining a physical characteristic of the footwear component.

2. The method according to claim 1 wherein the footwear component is adapted to modify motion of the user according to the first criterion.

3. The method according to any preceding claim wherein the footwear component is adapted to modify at least one of: a spring coefficient according to the first criterion; a damping coefficient of the user according to the first criterion; a Q factor of the user according to the first criterion; a biomechanical loading of the user according to the first criterion; an efficiency of the user according to the first criterion; and an injury risk of the user according to the first criterion.

4. The method according to claim 1 wherein the biomechanical load distribution comprises data representative of the distribution within the portion of the body of the user offerees exerted upon or by the body as a result of the motion of the user.

5. The method according to any of the preceding claims wherein the biomechanical model is calculated in accordance with the footwear component design.

6. The method according to any preceding claim wherein the biomechanical model is representative of a stiffness associated with at least the portion of the body of the user.

7. The method according to claim 7 wherein the stiffness of the user comprises data representative of at least one of: a spring constant for at least a portion of the body of the user as a result of the motion of the user; and a damping constant for at least a portion of the body of the user as a result of the motion of the user.

8. The method according to any preceding claim wherein calculating the biomechanical model comprises calculating values for the magnitude and direction of ferees exerted upon a plurality of parts of the body of the user, based upon the motion data and using a computational mechanical model of the body.

9. The method according to any of the preceding claims wherein the method further comprises providing the footwear design to the user.

10. The method according to any of the preceding claims wherein the footwear component comprises at least one of: an article of footwear, an insole, a midsole, an upper, a plate, an outsole, or an outsole spike configuration, an outsole stud configuration, an outsole lug configuration and an outsole grip configuration.11 . The method according to any of the preceding claims wherein the method further comprises manufacturing at least a portion of the footwear component.

12. The method according to any of the preceding claims wherein the method further comprises comparing the generated footwear design to each pre-existing footwear design from a database of pre-existing footwear designs to identify at least one pre-existing footwear design, wherein the at least one identified preexisting footwear design comprises one or more footwear parameters most closely matched to those of the generated footwear design.

13. The method according to any of the preceding claims wherein the method further comprises comparing the generated footwear design to each pre-existingfootwear design in a database of pre-existing footwear designs so as to obtain an ordering of the pre-existing footwear design, the ordering being based on a matched level of the one or more footwear parameters of each of the pre-existing footwear designs to those of the generated footwear design, and identifying at least one pre-existing footwear designs based on the ordering.

14. The method according to claim 12 or 13 wherein one or more differences between the generated footwear design and the most closely matched preexisting footwear design are provided to the user as proposed modifications to the pre-existing footwear design.

15. The method according to any of claims 12-14 wherein the method further comprises providing the at least one identified pre-existing footwear design to the user.

16. The method according to any of the preceding claims wherein the first criterion is an improved biomechanical function of the user.

17. The method according to any of the preceding claims wherein a parameter of the one or more footwear parameters comprises one of: a support / stiffness parameter, a heel to toe drop parameter, a heel stack height parameter, a spring constant parameter, a dampening parameter, a shoe size parameter, a shoe shape parameter, an insole parameter, a weight parameter, a grip parameter, a friction parameter, a stud / spike / cleat / lug parameter, an upper parameter, a traction parameter, a cushioning parameter, a tightness parameter, an upper flexibility parameter, a width parameter, a length parameter, a height parameter.

18. The method according to any of the preceding claims wherein the step of calculating a biomechanical model comprises calculating a target biomechanical model.

19. The method according to claim 18 wherein the step of obtaining motion data comprises obtaining target motion data of the user;calculating the target biomechanical model based on the target motion data and generating, according to the target biomechanical model, a target footwear component design comprising one or more target footwear parameters; and calculating a target adjustment that corresponds to a reduction in the difference of the target footwear component design and the footwear component design, wherein the first criterion represents the target adjustment.

20. The method according to claim 18 or 19 wherein the target biomechanical model is representative of the user and a target footwear component.

21. The method according to any of claims 18-20 wherein the step of calculating a target biomechanical model comprises calculating at least one of a target increase to the biomechanical model; and a target decrease to the biomechanical model.

22. The method according to any of the preceding claims wherein the user is one of a plurality of users, and wherein the step of calculating a biomechanical model comprises calculating an average biomechanical model for the plurality of users.

23. A method according to claim 22 wherein the step of calculating an average biomechanical model comprises at least one of: calculating an average biomechanical load distribution; and calculating an average stiffness for the plurality of users.

24. The method according to any of the preceding claims wherein the step of generating the one or more footwear parameters comprises using a machine learning model.

25. The method according to claim 24 wherein the machine learning model comprises one or more of: a Deep Neural Network, a Convolutional Neural Network, a Recurrent Neural Network, and a Spiking Neural Network.

26. The method according to claim 24 or 25 wherein the machine learning model inputs comprise at least one biomechanical parameter.

27. The method according to claim 26 wherein the biomechanical parameter comprises at least one of: speed, cadence, step length, foot strike, contact time, flight time, balance, stability, impulse, leg lift, strike angle, toe-off, vertical loading rate, mechanical power, and biomechanical loading values for at least the portion of the body.

28. The method according to any of claims 24-27, wherein the machine learning model comprises of one or more of an ensemble learning machine learning model comprising one or more of: a Convolutional Neural Network, a Deep Neural Network, a Recurrent Neural Network, and a Spiking Neural Network, wherein the outputs from the ensemble of models are concatenated to form a combined feature vector, which is input into one or more of the dense layers.

29. The method according to any of claims 24-28, further comprising training the machine learning model to predict at least one of maximising an efficiency index, minimising an injury risk index, maximising a stability index, minimising biomechanical loading values for at least one portion of the body, minimising biomechanical loading values for at least one portion of the body, maximising a spring constant, maximizing a dampening constant, maximising mechanical power, minimising mechanical power for a user.

30. The method according to any of claims 24-29 further comprising using the machine learning model to predict at least one of maximising an efficiency index, minimising an injury risk index, maximising a stability index, minimising biomechanical loading values for at least one portion of the body, minimising biomechanical loading values for at least one portion of the body, maximising a spring constant, maximizing a dampening constant, maximising mechanical power, minimising mechanical power for a user.31 . The method according to any of the preceding claims wherein the step of obtaining motion data comprises obtaining video data of the user indicating the body motion of the user performing an action and deriving the body motion of the user using the video data.

32. The method according to claim 31 wherein the step of deriving the motion data using the video data comprises using a machine learning model.

33. The method according to any of the preceding claims wherein the motion data comprises at least one of: an indication of the velocity and orientation of one or more parts of the body of the user during the performing of an action; and an indication of the pressure exerted upon one or more regions of the foot as a result of a contact force exerted upon that foot by the ground during motion.

34. The method according to any of the preceding claims wherein the method further comprises at least one of: obtaining environment data including an indication of the terrain and environmental conditions in which the user is intended to perform an action, wherein the step of calculating the biomechanical model is further based on the environment data; obtaining one or more activities, the one or more activities including an indication of the activity in which the user is intended to perform an action, wherein the step of calculating the biomechanical model is further based on the one or more activities; obtaining user sizing data comprising at least one of: a foot size of the user; a shoe size of the user; a foot length of the user; a foot width of the user; a foot height of the user; and a foot arch height of the user; and wherein the step of calculating the biomechanical model is further based on the one or more user sizing data; and obtaining anthropometric data comprising at least one of: a body mass index, BMI, of the user; a weight of the user, a foot size of the user, an age of the user, a gender of the user, anatomical dimensions of the user, a flexibility of the user, performance metrics of the user, a maximum heart rate of the user, a resting heart rate of the user, habitual information of the user and a height of the user; and wherein the step of calculating the biomechanical model is further based on the anthropometric data.

35. The method according to claim 34 wherein one or more footwear parameters are generated so as to optimise a muscle group used when the user performs the action as part of the activity.

36. The method according to any of the preceding claims wherein the method further comprises obtaining injury susceptibility data for the user, the injury susceptibility data including an indication of one or more injury-susceptible parts of the body of the user; and wherein the step of calculating the biomechanical model is further based on the injury susceptibility data.

37. The method according to claim 36 wherein the one or more footwear parameters are generated such that the footwear design is configured to minimise the forces that are exerted upon the one or more injury-susceptible parts as a result of the motion of the user.

38. The method according to claim 36 or 37 wherein the indicated parts of the body correspond to joints or muscles that are susceptible to injury.

39. The method according to any of claims 36-38 wherein the method further comprises assigning an injury priority weighting to the obtained injury susceptibility data, wherein the one or more footwear parameters are generated based on the injury priority weighting.

40. The method according to any of the preceding claims wherein the first criterion is configured in accordance with objective data, such that the biomechanical model is calculated in accordance with the objective data; and wherein the objective data corresponds to a primary objective for the user that includes any of: performance improvement, efficiency improvement, speed improvement, pain reduction, injury risk reduction, improving mobility improving stability and improving health and fitness.41 . The method according to claim 40 wherein the method further comprises assigning an improvement criterion priority weighting to the objective data, wherein the one or more footwear parameters are generated based on the improvement criterion priority weighting.

42. A system configured to generate a design for a footwear component, the system comprising: a motion data module configured to obtain motion data derived from monitoring motion of a user; a biomechanical modelling module configured to calculate, according to the motion data, a biomechanical model of at least a portion of the body of the user, wherein the biomechanical model is representative of a biomechanical load distribution through at least the portion of the body of the user, and wherein the portion of the body of the user comprises one or more anatomical structures of the body other than a foot of the user; and a design module configured to generate, according to the biomechanical model and a first criterion, a footwear component design comprising one or more footwear parameters, each footwear parameter defining a physical characteristic of the footwear component.

43. The system according to claim 42 wherein the motion data module comprises at least one of a force sensor; a pressure sensor; a camera; a photodetector sensor; an inertial measurement unit, IMU, sensor; an accelerometer sensor; a gyroscope sensor; an electromyography, EMG sensor a global positioning sensor, and a GPS sensor.

44. The system according to any of claims 42 and 43 further comprising a manufacturing module configured to manufacture the footwear component.

45. The system according to claim 44 wherein the manufacturing module comprises at least one of: a three-dimensional printer, a laser cutter; an additive manufacturing machine; an injection moulding machine; a milling machine; a stitching device; a weaving device; and a lathe.

46. A system configured to generate a design for a footwear component according to the method of any of claims 1-41 .

47. A non-transitory computer-readable storage medium configured to store computer executable code that when executed by a computer configures the computer to:obtain motion data derived from monitoring motion of a user; calculate, according to the motion data, a biomechanical model of at least a portion of the body of the user, wherein the biomechanical model is representative of a biomechanical load distribution through at least the portion of the body of the user and wherein the portion of the body of the user comprises one or more anatomical structures of the body other than a foot of the user; and generate, according to the biomechanical model and a first criterion, a footwear component design comprising one or more footwear parameters, each footwear parameter defining a physical characteristic of the footwear component.

Images