Deformable scaffold with mechanical tunability for bone tissue engineering

Abstract

A scaffold device for allowing distraction movement between segments of a bone, the scaffold device comprising: a first platform, a second platform, and struts extending from the first platform to the second platform. The scaffold device is configured to be deformed into position between segments of a bone, in a deformed state, from an undeformed state. When in the deformed state, the struts buckle and / or flex to bias the first platform and the second platform away from one another toward the undeformed state.

Description

[0001] DEFORMABLE SCAFFOLD WITH MECHANICAL TUNABILITY FOR BONE TISSUE ENGINEERING CROSS-REFERENCE TO RELATED APPLICATION

[0002] The present application claims the priority of United States Patent Application No. 63 / 664,251 , filed on June 26, 2024, the content of which is incorporated herein by reference.

[0003] TECHNICAL FIELD

[0004] The application relates implants and devices used in bone tissue engineering, for example in distraction osteogenesis and in other uses, such as devices known as scaffolds.

[0005] BACKGROUND

[0006] Management of bone defects is a highly prevalent challenge in orthopedic surgery. Bone defects may be the result of comminuted fractures, tumour resections, or aggressive bone debridement due to infections. As the size of the defect increases, advanced strategies are required for effective management and recovery of bone, while reducing the risk of failure. Filling a defect with autografts and / or allografts as a substitute for lost bone is a common approach. However, autografts involve harvesting bone from the patient’s own body, which can lead to donor site morbidity and limited availability. Allografts, on the other hand, come from a different individual and may carry the risk of immune rejection or disease transmission. Using tissue engineering scaffolds on the other hand removes the limitations of previous approaches, while showing similar or better outcomes.

[0007] One major challenge in treating bone defects is the prolonged healing period. The natural process of bone regeneration is complex and slow, especially in the case of large bone defects, in procedures like distraction osteogenesis. Tissue engineering scaffolds have emerged as a promising solution for enhancing bone regeneration. They can be made from a variety of materials, including biodegradable polymers, ceramics, and composites, and are often combined with growth factors and stem cells to further promote osteogenesis. The use of tissue engineering scaffolds can potentially reduce healing time, improve the quality of the regenerated bone, and allow for the repair of larger defects that would otherwise be challenging to treat.

[0008] The application of tissue engineering scaffolds is not without limitations. One of the primary challenges is the adjustability of their mechanical and structural properties. The scaffold must withstand physiological loads, yet be flexible enough to accommodate the natural movements of the bone. Therefore, stiffness plays a crucial role in getting the suitable bone healing outcome. Additionally, the scaffold's porosity is crucial for facilitating cell infiltration, nutrient diffusion, and waste removal, but achieving an optimal balance between porosity and mechanical strength is a challenge and highly dependent to the fundamental structure of the scaffold. Therefore, it is important to be able to adjust the stiffness and porosity of scaffolds to adapt to different treatments. Moreover, the degradation rate of the scaffold material must be carefully controlled to match the rate of new bone formation. If the scaffold degrades too quickly, it may fail to provide adequate support; if it degrades too slowly, it may impede new tissue formation or cause inflammation. Another important limitation is high cost and complexity of manufacturing custom-designed scaffolds.

[0009] In specific treatments such as distraction osteogenesis, which is the gradual elongation of a small bone defect to a large defect to stimulate bone growth, a lengthy consolidation phase where new bone slowly matures is disruptive specially among pediatric patients. This extended healing time can lead to complications such as infections, mechanical failure of fixation devices, and significant discomfort for the patient. Therefore, there is a critical need for devices and / or methods that can accelerate bone healing to improve patient outcomes and reduce the overall recovery period. A tissue engineered scaffold that can fit inside a small bone defect, as in distraction osteogenesis, and then be able to elongate with the gap elongation without any failure can be the solution.

[0010] SUMMARY

[0011] In a first aspect, there is provided a scaffold device for allowing distraction movement between segments of a bone, the scaffold device comprising: a first platform, a second platform, and struts extending from the first platform to the second platform, wherein the scaffold device is configured to be deformed into position between segments of a bone, in a deformed state, from an undeformed state, and wherein, when in the deformed state, the struts buckle and / or flex to bias the first platform and the second platform away from one another toward the undeformed state.

[0012] Further in accordance with the first aspect, for instance, the device has a monolithic body of biocompatible material.

[0013] Still further in accordance with the first aspect, for instance, the monoblock is made of a biodegradable material. Still further in accordance with the first aspect, for instance, the biodegradable material is polycaprolactone.

[0014] Still further in accordance with the first aspect, for instance, the struts are arranged into at least a first annular distribution of interconnected struts.

[0015] Still further in accordance with the first aspect, for instance, at least a second annular distribution of interconnected struts is present.

[0016] Still further in accordance with the first aspect, for instance, the first platform and / or the second platform is an annular disk.

[0017] Still further in accordance with the first aspect, for instance, the first platform and / or the second platform is a polygon.

[0018] Still further in accordance with the first aspect, for instance, the struts are arranged into a zig-zag pattern, in which adjacent ones of the struts merge at or adjacent to the first platform or the second platform.

[0019] Still further in accordance with the first aspect, for instance, the adjacent ones of the struts merging emulate clamped joints.

[0020] Still further in accordance with the first aspect, for instance, the adjacent ones of the struts merging emulate hinged joints.

[0021] Still further in accordance with the first aspect, for instance, the struts in the zig-zag pattern include longer struts and shorter struts, wherein, in the deformed state of the scaffold device, a majority of the longer struts is in tension and a majority of the shorter struts is in compression.

[0022] Still further in accordance with the first aspect, for instance, in the undeformed state of the scaffold device, a majority of the longer struts is in compression and a majority of the shorter struts is in tension.

[0023] Still further in accordance with the first aspect, for instance, in the zig-zag pattern, the struts alternate in sequence between longer strut and shorter strut.

[0024] Still further in accordance with the first aspect, for instance, the struts include longer struts and shorter struts, wherein, in the deformed state of the scaffold device, a majority of the longer struts is in tension and a majority of the shorter struts is in compression. Still further in accordance with the first aspect, for instance, the struts all have an equal length.

[0025] Still further in accordance with the first aspect, for instance, the struts of the first annular distribution all have an equal length.

[0026] Still further in accordance with the first aspect, for instance, at least some of the struts of the second annular distribution have a length differing from the equal length of the struts of the first annular distribution.

[0027] In accordance with a second aspect, there is provided an assembly for use in distraction osteogenesis comprising: at least two of the scaffold device as described above, with the contacting platforms optionally glued or connected mechanically, wherein the at least two scaffold devices are deformed into position between segments of a bone in distraction osteogenesis into the deformed state, and wherein, when compressed to the deformed state between the segments of the bone, adjacent ones of the at least two scaffold devices contact one another with the first platform of one of the scaffold devices being against the second platform of the other of the scaffold devices.

[0028] Further in accordance with the second aspect, for instance, the platforms in contact with the segments of the bone have at least one attachment part projecting from the circumference of the platform to the bone outer surface.

[0029] Still further in accordance with the second aspect, the attachment part has a hole.

[0030] DESCRIPTION OF THE DRAWINGS

[0031] Reference is now made to the accompanying figures in which:

[0032] Fig. 1 is a schematic view of a scaffold device in accordance with a variant of the present disclosure;

[0033] Fig. 2 is a perspective view of an exemplary embodiment of the device of Fig. 1 ;

[0034] Fig. 3 is a perspective view of another exemplary embodiment of the device of

[0035] Fig. 1 ;

[0036] Fig. 4 is a perspective view of a pair of the device of Fig. 2 as used in end-to-end assembly;

[0037] Fig. 5 is a perspective view of a pair of the device of Fig. 3 as used in end-to-end assembly; Fig. 6 shows pairs of the device of Fig. 1 (A) being in a deformed state and (B) having expanded from the deformed state of (A) toward an undeformed state;

[0038] Fig. 7 is a diagram showing responses of a device of Fig. 1 using a finite element model;

[0039] Fig. 8 is a diagram showing evaluated elasticity and strain as a function of device configurations;

[0040] Fig. 9 is a Bayesian machine learning flowchart to create predictive models of structural and mechanical properties of the device of Fig. 1;

[0041] Fig. 10 is a series of equations applicable to an optimization of design of the device of Fig. 1 ;

[0042] Fig. 11 is a Four-objective optimization flowchart of the device to maximize elastic modulus or strain energy and surface area to volume ratio, and minimize rotation, while keeping porosity close to 80%;

[0043] Fig. 12(a) is a FE model of the device with loading and boundary conditions; while (b) is a CFD model of the fluid phase, in which fluid inlet is evenly distributed across the outer surface of the device, zero pressure at the outlet is applied at the entire surface of the inner surface of the device, and the top and bottom surfaces have non-slip boundary conditions;

[0044] Fig. 13 is a GP classification diagram illustrating the effects of varying input parameters on the foldability of the device 10 unit cell; (Top) Unit cells with h = 5 mm corresponding to the 4UC design group; (Bottom) Unit cells with h = 10 mm corresponding to the 2UC design group; Foldable and not foldable regions are indicated with tone changes, respectively; The design space for classification is predefined to correspond to 4UC and 2UC models, which is provided in Fig. 11 ;

[0045] Fig. 14 is a GPR prediction diagram illustrating the effect of varying input parameters on structural and mechanical properties of the device of Fig. 1. (a) Porosity, (b) Surface area to volume ratio, (c) Maximum rotation, (d) Normalized elastic modulus, (e) Strain energy. In each column, top row corresponds to the outer layer of the device of Fig. 1 with h = 5 used for the 4UC design group. Bottom row corresponds to the outer layer of the unit cage with h = 10 used for the 2UC design group. Bars represent the range of output values derived from the predefined design space corresponding to 4UC and 2UC models, which is provided in Fig. 11 ; Fig. 15 is a diagram showing a multi-objective space of P, S / V, ip, and U derived from testing 10000 feasible inputs as described in Fig. 11 for: (top) Unit cells, i.e., the devices of Fig. 1 , with h = 5 mm corresponding to the 4UC design group, (bottom) Unit cells with h = 10 mm corresponding to the 2UC design group; The Pareto optimal points are highlighted with red and the closest point to the Utopia point is labeled with a black star in a dotted circle; The gradient represents ip across the objective space;

[0046] Fig. 16 presents CAD models of the optimized designs and the fabricated samples via SLS 3D printing;

[0047] Fig. 17 presents different sizes of the same design printed with SLS 3D printing;

[0048] Fig. 18 is a diagram showing FEA outcomes of model A (4U) and model A (2U); (a) Front and top view of undeformed unit cage with imperfections included in the geometry; (b1 ) Axial displacement contour under 12.5% strain; (b2) local strain contour under 12.5% strain; (c1 ) Axial displacement contour under 25% strain; (c2) local strain contour under 25% strain;

[0049] Fig. 19 is a diagram showing FEA outcomes of model B (with 2 and 4 UC configurations); (a) Front and top view of undeformed unit cage with imperfections included in the geometry; (b1 ) Axial displacement contour under 12.5% strain. (b2) local strain contour under 12.5% strain; (c1 ) Axial displacement contour under 25% strain; (c2) local strain contour under 25% strain;

[0050] Fig. 20 is a diagram showing FEA outcomes for single unit cage model, stressrotation response of each model;

[0051] Fig. 21 is a diagram showing CFD outcomes of model A (with 2 and 4 UC configurations); (a) Geometry and boundary conditions of the ROI for fluid phase in undeformed and deformed configuration; (b) Velocity pathlines for each model; (c) Pressure contour for each model; (d) wall shear stress due to surrounding fluid flow;

[0052] Fig. 22 is a diagram showing CFD outcomes of model B (4U) and model B (2U); (a) Geometry and boundary conditions of the ROI for fluid phase in undeformed and deformed configuration; (b) Velocity pathlines for each model; (c) Pressure contour for each model; (d) wall shear stress due to surrounding fluid flow;

[0053] Fig. 23 is a diagram showing the permeability of each model in strain free and 25% strain condition; Fig. 24 is a diagram demonstrating the key structural properties of each model in strain free and 25% strain condition and their relation with permeability; (a) Change in porosity, (b) Change in surface are to volume ratio, (c) Combine effect of change in P and S / V to permeability of each model in undeformed and defamed configuration;

[0054] Fig. 25 is a diagram demonstrating experimental results: (a) Stress-strain curves of models under axial compression test; Points where each sample buckled and their corresponding images as well as points where the test was terminated as well as the corresponding images are shown; (b) linear elastic modulus of each model before buckling; (c) strain energy of each model at 25% strain; and

[0055] Fig. 26 is a table of output values and their corresponding inputs for the compromised point in each layer for the four selected designs.

[0056] DETAILED DESCRIPTION

[0057] Referring to the drawings are more particularly to Fig. 1 , a scaffold device for filling a gap between segments of a bone is generally shown at 10. The expression “device” is used herein, by other expressions could be used to identify the device 10, such as a scaffold, an implant, a prosthesis. The device 10 is implanted in the body between bone segments, has the shape and configuration of a scaffold, and is temporarily a bone prothesis. Other expressions could be used to name the device 10.

[0058] The device 10 may have various uses. In an embodiment, the device 10 is used in distraction osteotomy, as positioned between bone segments pursuant to a resection. This is one use among others in tissue engineering uses. Other uses may include cartilage and tendon tissue engineering, where high compression or tensile strains must be tolerated. The device 10 is configured to be compressed into position between segments of a bone. When compressed into position between the segments of the bone, the device 10 exerts a biasing force to fill the gap between bone segments. The device 10 may have a first platform 11 , a second platform 12. Struts 13 extend from the first platform 11 to the second platform 12. The struts deform, such as by buckling or bend as a result of a compressive force between the first platform and the second platform. The device 10 is a passive structure that may enable osteogenesis. In a possible use, as bone segments are gradually distracted using a distraction device (internal or external), the scaffold device 10 that is compressed into a gap between the bone segments may progressively deploy or be biased to an uncompressed state by action of the struts 13. The scaffold device 10 may not exert a compressive force between bone segments or may exert a negligible compressive force.

[0059] As illustrated in Fig. 2, the first platform 11 and the second platform 12 are shown as being plates, with a central opening, respectively shown as 11A and 12A. The platforms 11 and 12 may also be referred to as plates, base plates, etc. An outer periphery of the platforms 11 and 12 is shown as being circular, but could have other shapes, including oval, a polygon with n sides, etc. For example, in a variant, the outer peripheries may have patient-specific geometries, i.e., they may be shaped based on imaging and may thus have a periphery that matches the periphery of the bone segments at the resection. The platforms 11 and / or 12 may optionally be tied to the bone segment against which they are laid. For example, it is considered to used one or more fastener, an adhesive, surface feature, for the platforms 11 and / or 12 to be affixed to the bone surface.

[0060] The platforms 11 and 12 may be substantially planar as observed. Indeed, if the bone segments are the result of a resection, the surface of the cut at the resection may be substantially planar, whereby a planar bottom surface for the platform 11 and a planar top surface for the platform 12 are well suited for abutment of the resected bone surfaces against these surfaces. Other geometries are considered, including some with non-planar geometries depending on the surfaces against which they will be positioned. In a variant, a vector representative of a compressive force Z of the device 10 is normal to a plane lying in X-Y of the planar bottom surface for the platform 11 and / or of the planar top surface for the platform 12. The platforms 11 and 12 are made of a biocompatible material, that may be porous. The platforms 11 and 12 may be said to be osteoinducing and / or osteoconducting. In a variant, deformation may actively modulate permeability and wall shear stress by altering porosity and surface complexity of the platforms 11 and 12, the large deformation of the struts changing the permeability and wall shear stress.

[0061] Still referring to Fig. 1 , the struts 13 are shown being in a Kresling pattern, as one possible type of pattern that may be used. As observed, an exemplary strut 13A extends from the platform 11 to the platform 12. The strut 13A may not be parallel to direction Z and may have its opposed ends rigidly connected to the platforms 11 and 12. For example, the strut 13A extends at an angle 01 relative to a plane of the platform 11 . Another exemplary strut 13B extends from a point of connection between the strut 13A and the platform 12, and has its opposed end connected to the platform 11. Optionally, the strut 13B may not be parallel to direction Z (though it may be) and may have its opposed ends rigidly connected to the platforms 11 and 12 (and to the strut 13A). As observed, the angle 02 between the struts 13A and 13B may be acute, though this is optional. The value ranges for angles 01 and 02 is defined by the number n of struts that are distributed on the perimeter of the platforms 11 and 12 and angle cp. The angle <p may be described as being between the ends of a strut 13, and a center of the one of the platforms 11 or 12. The number n can be any more than 3. The angle <p can be between -2rr / n to + 70. Fig. 8 shows the parameters of elasticity and strain based on angle cp, number n and a d / h ratio, based on Fig. 1.

[0062] Another exemplary strut 13C extends from the platform 11 to the platform 12. Strut 13C extends from a point of connection between the strut 13B and the platform 11 , and has its opposed end connected to the platform 12. It may therefore be said that the adjacent struts merge and are interconnected. The interconnection may behave as a clamped joint or a hinge joint, depending on the size of the strut cross-section and material composition. For example if the strut diameter is close to the thickness of a paper (100 micron) the clamped (fixed) joint will have almost no resistance to bending moments exhibiting the same behavior as a hinge joint. A similar behavior is expected if we use a comparatively softer material when fabricating the joint. Printing different materials is achievable with a method called multi-material 3D printing. The strut 13C may not be parallel to direction Z and may have its opposed ends rigidly connected to the platforms 11 and 12. As observed, the angle between the struts 13B and 13C may be acute, though this is optional. In a variant, the angle 02 between all interconnected struts is acute, and is less than 45 degrees. The pattern of struts 13A, 13B, 13C may be repeated until a strut 13n (not shown) is connected to the point of connection between the strut 13A and the platform 11. Accordingly, in the Kresling pattern, the adjacent struts 13 are interconnected at a junction with one of the platforms 11 or 12, and form a triangle with a virtual segment at their opposite ends, the virtual segment potentially lying in the other platform 11 or 12. In a variant, there are only two strut length, i.e., a first strut length being that of the struts 13A, 13C, etc, and a second strut length being that of the struts 13B, 13D, etc. However, it is possible for the struts 13 to be present in more than two lengths. Moreover, while adjacent struts 13 meet at one of the platforms 11 or 12, it is also possible for one of the struts 13 to connect to one of the platforms 11 or 12, with the other strut connecting to the first strut near the intersection with the platform 11 or 12, but not at the platform 11 or

[0063] 12. If there are two lengths of struts, a majority (or all) the longer struts may be in tension when the device 10 is in a deformed state between bone segments, while a majority (or all) the shorter struts may be in compression when the device 10 is in a deformed state between bone segments. Thus, the device 10 may have multiple concentric layers (i.e., annular arrangements) with distinct geometries, selected to achieve matched rotational and foldable displacement when the device 10 is in the deformed state, with a selected rotation-constrained strategy.

[0064] In a variant, there is a single annular arrangement of struts 13 as in Fig. 1. However, as observed from Figs. 2 and 3, more than one annular arrangement of struts 13 may be present, such as by having one annular arrangement inside the other. The struts 13 or strut sets in a first annular arrangement may be the same as the struts 13 or strut sets in a second annular arrangement. However, it is possible to have different struts or strut sets in the annular arrangements, i.e., each annular arrangement may have its own strut or strut sets. The annular arrangements may also be known as unit cells. In an example, the unit cells may be spaced 400 pm apart (± 50 pm in a variant) with struts defined by three parameters: cross-sectional diameter (dstrut), orientation angle (4>), and repetition number n, wherein n is the number of short-long strut pairs repeating on the perimeter of each of the annular arrangements.

[0065] The angles between struts 13 and platforms 11 and 12 may be tuned according to the desired stiffness of the device 10. Comparatively, Figs. 2 and 3 show devices 10 having similar platforms 11 and 12, and a similar number of annular arrangements, but with different angles. A radius R, or a maximum distance from a center of the platforms 11 or 12 may be between 5 and 15 mm. A height h between platforms 11 and 12 may be between 0.3 and 10 mm.

[0066] In a variant, when the device 10 is placed between bone segments, such as for promoting osteogenesis, an assembly of two or more of the devices 10 are positioned end to end, as demonstrated in Figs. 4 and 5. Indeed, in Figs. 4 and 5, it can be observed that a pair of the devices 10 are positioned end to end, such as by the platform 11 of an upper device 10’ being laid onto the platform 12 of the lower device 10”. A third, a fourth device 10 could also be added. While the platforms 11 and 12 of adjacent devices 10 are against one another, it is possible to have different arrangements, such as platforms 11 against one another, platforms 12 against one another, or the lower device 10” having the platform 11 on top, and the upper device 10’ having the platform 12 at the bottom. In a variant, the adjacent devices 10’ and 10” are laid against one another, and may or may not be secured to one another via the coplanar platforms 11 and 12. In a variant, the devices 10’ and 10” are configured to provide opposite rotations, i.e., counterrotations, to allow rotation of the coplanar platforms 11 and 12, while the other platforms 11 and 12 that are in contact with the bone segments are fixed. In an alternative arrangement, the platforms 11 and / or 12 are not fastened to one another. Stated differently, the devices 10’ and 10” can slide with respect to another, namely they can translate relative to one another and / or rotate relative to one another. Indeed, as the platforms 11 and 12 of a device 10 may rotate relative to one another when the device 10 is compressed and regains its shape, i.e., when h varies, the sliding joint between adjacent devices 10 prevents a rotation between one of the devices 10 and the bone surface it contacts. In a variant, the devices 10 may not move relative to one another, by having the devices 10 mirror one another. For example, as one of the devices 10 turns in a given direction when decompressing, the other of the devices 10 may be arranged to rotate in the opposite direction when decompressing, such that the platforms 11 and / or 12 that are against one another without relative rotation when the devices 10 expand. In a variant, the devices 10 used in a pair or more may have opposite chirality to eliminate net rotational displacement at scaffold boundaries during axial compression, i.e., chiralitybased cancellation would be present.

[0067] Therefore, the device 10 may be referred to as an origami-like 3D-patterned metamaterial bone scaffold with a deformable mechanical structure. In a variant, the Kresling pattern may be described as a sequence of tessellated triangles, in which adjacent triangles share a strut 13, and in which each triangle is defined by a pair of struts 13 and a segment on either one of the platforms 11 and 12 being a straight line between the ends of the pair of struts 13. For example, 13A, 13B and the platform 11 form a first unit triangle, 13B, 13C and the platform 12 form a second unit triangle, and so forth and so on. The struts 13 form polygons that are tessellated in an annular arrangement, such as in an annular pattern, that may optionally be repeated to achieve a desired density.

[0068] The platforms 11 and 12 are shown as being annular, with openings 11A and 12A. This configuration contributes to bone regeneration, as a passage is defined between the surfaces of the bone segments. Moreover, as the struts 13 are in an annular arrangement, a central passage is maintained, again favorizing bone regeneration. In other variants, it is considered to have the platforms 11 and / or 12 made of a permeable material, or to have the platforms 11 and / or 12 with interstitial spaces. The devices 10 can be used on their own in various tissue engineering uses, such as when the device 10 is used as a skin scaffold. When used for distraction osteogenesis, the devices 10 are used in combination with a bone distraction device, such as an external fixator, telescopic intramedullary nails, external fixators and internal telescopic plate fixator. Hence, the assembly of devices 10 can accelerate bone regeneration, such as in distraction osteogenesis. For example, as shown in Fig. 6, a pair of the devices 10 are shown in a deformed or compressed state at (A). The devices 10 are designed to fit in small osteotomy gaps and to deploy during the elongation process and support growing callus tissue. In the deformed or compressed state, because of their tessellated arrangement, the struts 13 buckle and / or bend. In such a condition, the struts 13 provide a distracting force, to remain in contact with the bone segments as the gap between the bone segments increases, leading to (B). Stated differently, the device 10 has its strut 13 buckle and / or flex and / or bend to bias the first platform and the second platform away from one another toward the undeformed state, with repulsive forces between the platforms 11 and 12. However, the force applied on the bone segments may be negligible or absent.

[0069] The struts 13 can buckle and / or flex regardless of the joint type or the pattern that they are in. The buckle modes (form of the deformation in the struts 13) and the kinematics of folding are dictated by the boundary conditions at the junction of the struts 13 with the platforms 11 and 12, by the pattern of the struts 13, such as the Kresling pattern shown and described herein as an embodiment among others. In an optimized configuration, the junction of the struts 13 with the platforms 11 and / or 12 should behave similar to a fix joint which has no kinematic motion, and stores energy in the form of strain. Joints of the struts 13 with the platforms 11 and / or 12 can also be designed to behave similar to a pin joint which releases energy through kinematic rotation letting the structure have two strain free states. This may be known as bistability, and less likely to have strains that reach the yield limit at the junction of strut 13 with platform 11 and 12.

[0070] Various materials may be used for the device 10. In an embodiment, polycaprolactone (PCL), is one possible biocompatible and biodegradable polymer that may be used for the device 10. The mechanical properties of PCL, characterized by a controlled rate of biodegradation, align well with the requirements for bone regeneration. The bone tissue can gradually replace the PCL material. In a variant, PCL may be combined with other biomaterials, to enhance the properties of the device 10, such as bioactivity and the capacity to imitate the composition of natural bone. For example, hydroxyapatite and p-tricalcium phosphate may be used for osteoconductivity properties. In some uses, the device 10 is only required to provide load bearing capacity, and no compressibility, whereby in such cases ceramics and metals such as titanium may be used.

[0071] PCL can be stretched from 20 to 100% depending on the processing techniques and composition, making it a suitable choice for the device 10. The device 10 has been developed through computer-aided design and optimized by identifying the appropriate constraints to drive the optimization process (number, size, material of the components used in the design). As PCL can be 3D-printed, the manufacturing method for the device 10 may be selective laser sintering (SLS) as a possibility.

[0072] Design of highly deformable tissue-engineered scaffolds is challenging due to limitations in material choice, method of additive manufacturing, and uncertainty in the macro / micro mechanical properties. The device 10 is a scaffold capable of relatively large deformations through buckling in its struts 13. A parametric design of the device 10 was created to explore the effects of geometrical parameters on its mechanical properties. Finite element method was leveraged to generate computational structural (global) stress-strain and local maximum principal strain data corresponding to the entire solution space. Using Bayesian machine learning, the sensitivity of the output to the input parameters was analyzed and regression tasks was done. Scaffold was classified as Linear, non-linear, and bi-stable, which can be identified depending on the input design values. From the solution space, five designs were selected and 3D printed monolithically by selective laser sintering (SLS) method and with a commercially available powder with similar elongation limit as polycaprolactone (PCL). All the selected designs could withstand 80% recoverable strain in quasi static and cyclic loading, while exhibiting different mechanical behavior. The present disclosure demonstrates the viability of the design and manufacturing of origami inspired super deformable scaffolds with highly-tunable mechanical properties, which can be used for various tissue engineering applications. The device(s) 10 may be designed using method with expansive properties using Bayesian machine learning or artificial neural network (ANN) to predict and optimize porosity, surface area-to- volume ratio, stiffness, energy absorption and rotation, such as by using surrogate modeling or Bayesian optimization. Figs. 7 and 8 illustrate the responses of the device 10 based on different configurations, indicating that the device 10 can be tuned based on the needs. In a variant, the device 10 may therefore be used as distraction osteogenesis scaffold that is implanted between the two bone segments to accelerate bone consolidation and healing. As mentioned above, uses of the device 10 may not be limited to limb distraction osteogenesis. The device 10 is well suited in its design to promote bone regeneration / healing. For example, the device 10 may enhance angiogenesis, may align collagen fibers. The device 10 has the potential to improve and accelerate the osteosynthesis. All of these features may result in a reduction of treatment duration.

[0073] The pattern shown and described in Figs. 1-5, that may for example include a triangular strut arrangement such as a Kresling pattern, may offer a wide range of possibilities for creating structures that can undergo large and complex yet predictable shape transformations. The device 10 may be regarded as a versatile pattern for various applications.

[0074] The pattern of the device 10 is such that the device 10 may exhibit sensitivity to changes in initial geometric design. This may allow a more precise control over the behavior of the structure. Different types of spring behavior such as linear, non-linear, quasi-zero, and bi-stable can be achieved by simply altering the geometry of the triangles of struts 13, making the device 10 adaptable to specific needs or conditions. Moreover, the stiffness of the device 10 featuring the Kresling pattern can be controlled and greatly changed by just changing the geometric arrangement of the struts 13.

[0075] Moreover, the illustrated embodiment, in which side struts and diagonal struts in the pattern enable the transfer of compression and tension providing stability to the structure. In Fig. 1 , the side struts are the shorter struts such as 13B and 13D, while the diagonal struts are the longer struts, such as 13A and 13C. Any translational shear loads applied to the platforms 11 and / or 12 may be resisted.

[0076] Contrary to other scaffolds, the device 10 as detailed above is suitable for distraction osteogenesis, in spite of the relatively small initial osteotomy size (1-2 mm). By way of the design explained above, inspired by Kresling origami patterns, the device 10 may be capable of withstanding relatively large deformations, which is useful for the typical sizes of an osteotomy gap during the callus distraction (e.g., from 1 to 20 mm gap size). The device 10 can accelerate bone regeneration by increasing angiogenesis, a process by which mesenchymal stem cells are exposed to more oxygen required to promote intramembranous ossification, aligning collagen fibers, and controlling the mechanical stimuli inside the callus. The device 10 has the potential for a variety of tissue engineering applications as its mechanical behavior is tunable and as the device 10 can be manufactured in different scales and configurations.

[0077] As an exemplary analysis of the effectiveness of the device 10, an automated virtual lab integrated with a BML framework was used to predict the behavior of the device 10 (Fig. 9). Five input variables define a unit cell of the device 10: dstrut, <P, n, cell diameter (dceii), and height (h). Design space bounds were determined through trial and error based on manufacturing and geometric constraints. A Sobol sequence generated 1 ,500 samples using SAlib, each described as a 5-component input vector Xi (Fig. 9 and Fig. 10a).

[0078] Inputs were normalized, and dstrut and dceii were scaled by h. For the present application, h = 10 mm was used in the workflow. The resulting design space covered 0.0014 < dstrut / h < 0.07 and 1.4 < dceii / h < 4 for h = 5 or 10 mm. CAD models were generated via Python scripting in ABAQUS V6.24 (Velizy-Villacoublay, France), and FE simulations of unit cells of the device 10 under 80% compressive strain (0.8h) were performed using the Arc-length method (Fig. 9).

[0079] Porosity (P) and free surface-to-volume ratio (S / V) were calculated for each sample and considered the key structural parameters influencing the biology of bone healing. Mechanical outputs included the maximum rotation at the top disc (ip), indicative of local shear strain on the surrounding tissue; the elastic modulus (E), representing the linear stiffness at low strain; and strain energy density (U), reflecting the nonlinear mechanical response. Each sample was therefore represented as an output vector Yi of five components (Fig. 10b), in which Ytdenotes the normalized outputs for the ithsample.

[0080] These were paired with corresponding input vectors Xi to form the full dataset. Samples that failed to reach 0.8h strain were excluded using a Bayesian binary classifier and labeled as unfoldable, and the remaining 1 ,085 foldable samples were retained. Of these, 800 were used to train Gaussian Process Regression (GPR) models using as an example GPflow, while the rest were reserved for testing. Sampling continued until all models achieved R2> 0.98. The trained predictive models of the outputs are expressed as in Fig. 10c, in which X is an arbitrary normalized input vector within the design space of the trained data to visualize the input-output relations in the 4-dimensional design space, dceii / h was fixed at selected values (4, 3.2, 2.8 for 4UC; 1.4, 1.6, 2 for 2UC), corresponding to constant diameters for the outer, intermediate, and inner annular arrangement of each device 10.

[0081] An exemplary workflow for the optimization of the device 10 is depicted in Fig. 11. A total of 10,000 samples were generated for each design space adapted for 4UC and 2UC using a Sobol sequence, ensuring efficient and unbiased coverage of the multidimensional design space. Each sample was checked for foldability using the trained Bayesian classifier; only foldable samples were retained for further analysis.

[0082] For each retained sample, the five key outputs P, S / V, ip, E, and U were predicted using the trained surrogate models. These five outputs formed a 5D multi-objective optimization space.

[0083] Two optimization goals were defined:

[0084] • Goal A: Target porosity = 80%, maximize SA / , minimize ip, and maximize E;

[0085] • Goal B: Target porosity = 80%, maximize S / V, minimize ip, and maximize U.

[0086] To formalize the optimization, cost functions were constructed for each goal, as in Fig. 10d, where ptarget is the normalized value corresponding to 80% porosity, and each fi(X) represents one of the five surrogate model outputs. Equal weights were assigned to each component of the cost functions, assuming structural and mechanical properties contribute equally to the device 10. The multi-objective Pareto optimal solution is defined as a set of non-dominated solutions — those that cannot be improved in one objective without degrading another. The boundary formed by these solutions is known as the Pareto front, from which all points are considered optimized. In the present application a compromised solution was selected based on its minimum Euclidean distance from the Utopia point, a hypothetical ideal point in the objective space where all outputs reach their optimal values simultaneously.

[0087] This entire process was repeated for the three concentric layers of the unit cage in both 4UC and 2UC scaffold configurations. The resulting compromised points were used as input to the inverse design step. Fig. 11 provides a schematic overview of this four-objective optimization process to maximize E or U and S / V, minimize ip, and maintain P near 80%.

[0088] The optimized design points from the Pareto front were integrated across the three concentric layers (i.e., the annular arrangements of struts 13) of the device 10 to construct full scaffold geometries. Each layer behaves like a parallel spring, which constrains the system such that all layers must undergo the same rotational displacement. To satisfy this constraint, the layer with the lowest i was selected as the reference point (RP). If the compromised points from the other layers had rotation values matching the RP, they were selected directly. Otherwise, alternative points were selected from the corresponding Pareto front to minimize deviation from the RP’s rotation. This approach ensured kinematic compatibility across layers while maintaining optimal mechanical and structural performance. The effects of absence of this constraint can be found in the supplementary materials. Final input-output data and the resulting CAD models for both 4UC and 2UC designs are presented in the table of Fig. 26 and in Fig. 21 , only provided as examples.

[0089] FE simulations were conducted on a single unit cage (UC) model from each configuration of scaffold device 10 to assess mechanical performance under compression. CAD models were meshed in ABACUS using 10-node quadratic tetrahedral elements (C3D10). Material properties were defined for Flexa Gray (Sinterit, Poland), a thermoplastic polymer used in SLS printing, with a bulk elastic modulus of 13.78 MPa and Poisson’s ratio of 0.35. These mechanical properties are comparable to high molecular weight polycaprolactone (PCL), a biomaterial widely used in tissue engineering scaffolds.

[0090] To capture the nonlinear deformation of the struts, the arc-length method with the Riks solver was implemented. Boundary conditions fixed the bottom disc in all degrees of freedom, while the top disc was allowed to translate in all directions and rotate about the z-axis but restricted from tilting. A displacement-controlled loading was applied until full strut contact. Local strain distribution was extracted from the simulation results (Fig. 12a).

[0091] For permeability evaluation, computational fluid dynamics (CFD) simulations were performed in ANSYS. Each simulation was based on two UC geometries: one undeformed (with manufacturing imperfections included) and one deformed under 25% compressive strain, as derived from the FE output. The fluid phase was generated by subtracting the volume of the struts and discs from a hollowed cylinder surrounding the scaffold (Fig. 12a).

[0092] The outer radius (Ro) of the cylinder was selected slightly larger, and the inner radius (Rt) slightly smaller than the UC’s outer radius, to minimize boundary condition effects. The fluid was assumed to be Dulbecco’s Modified Eagle Medium (DMEM), a cell culture medium commonly used to simulate body fluid in vitro. It was modeled with a viscosity of 1 mPa s and an inlet velocity of 1 mm / s. Non-slip boundary conditions were applied to the top and bottom disc interfaces.

[0093] To reduce computational cost, a quarter-cylinder region of interest (ROI) was extracted, assuming symmetry. The calculation of the volumetric flow rate through this region is presented in Fig. 10e, where ^quarter is the curved outer surface area of the ROI, L is the height of the fluid phase, and Vtis the inlet velocity. Radial permeability and equivalent permeability were calculated based on Darcy’s law in cylindrical coordinates and are presented in Fig. 10f and Fig. 10g, respectively, where q is the dynamic viscosity of DMEM and AP is the pressure drop.

[0094] The STL files of the optimized designs of the OISDS were imported into Sinterit Studio (Version 3.6, EOS GmbH) and sliced into layers of 50 pm, at a laser scanning speed of 1500 mm / s. Six samples per selected design (2UC and 4UC) were fabricated using selective laser sintering (SLS) with a Lisa Pro 3D printer (Sinterit Sp. z o.o., Krakow, Poland). The printing material was Flexa Gray, a commercially available thermoplastic polymer powder suitable for SLS.

[0095] Complete scaffold assemblies were fabricated for the 2UC and 4UC models, i.e., respectively two devices 10 and four devices 10. For mechanical testing, each sample was fixed at the top disc in all degrees of freedom using double-sided tape, while the bottom disc was mounted using a custom revolute joint with a thrust bearing to allow free rotation about its central axis.

[0096] Samples were subjected to quasi-static uniaxial compression up to 70% apparent strain in the table of Fig. 26 at a crosshead speed of 2 mm / min using a Mach-1 micromechanical testing system (Biomomentum Inc., Laval, QC, Canada). Loaddisplacement data were recorded with a 100 N load cell.

[0097] Apparent stress (oa) and strain (e) were calculated by dividing the measured force by the cross-sectional area of the scaffold and the actuator displacement by the scaffold height. Stress-strain curves were segmented into three characteristic regions based on the typical behavior of porous structures under compression:

[0098] • Linear region, used to determine the effective elastic modulus (Eetf);

[0099] • Plateau region, characterized by reduced or negative stiffness due to buckling;

[0100] Densification region, where stiffness increases as struts come into contact. The calculated effective elastic modulus is displayed in Fig. 10h, where Fris the reaction force at the bottom disc, A is area of the hollowed discs, E is the apparent strain of the UC, A / i is the height change of UC, ho is the initial height of the UC, and routand rtnare the outer and inner radius of the discs, respectively. Eett was then nondimensionalized by dividing by the bulk modulus (Eb).

[0101] Strain energy density (U) was computed as the area under the stress-strain curve up to 25% strain as in Fig. 10i.

[0102] Experimental data was described as mean ± standard deviation. All mechanical measurements were conducted with a minimum of three replicates per design to ensure reliability and reproducibility.

[0103] Gaussian Process (GP) classification (Fig. 13) highlighted the substantial influence of the normalized cell diameter (dceii / h) on the foldability of unit cells, i.e. , unit devices 10. For the 4UC designs, nearly the entire design space was foldable, with exceptions where strut thickness (d_strut / h) was minimal and the number of struts (n) was high, while <p stayed within 10° to 50°. As the layers changed from outer layer to the inner layer, i.e. from dstrut / h = 4 to dstrut / h = 2.8, the unfoldable space slightly spreaded towards smaller n and wider range of <p. In contrast, 2UC designs exhibited larger unfoldable regions. Specifically, designs with dstrut / h between 0.007 and 0.02 and orientation angles <p between -5° and 40° were more likely to be unfoldable under axial compression. Increasing n slightly expanded these regions.

[0104] Gaussian Process Regression (GPR) models trained on 800 foldable samples demonstrated high predictive accuracy (R2> 0.98) across five key outputs: porosity (P), surface area-to-volume ratio (S / V), maximum rotation (ip), normalized elastic modulus (Eetf / Ebuik), and strain energy density (U). These outputs were analyzed over the full 5D design space (dstrut / h, <p, n, dceii / h, and h), with fixed values of h = 5 mm for 4UC and h = 10 mm for 2UC designs (Fig. 14).

[0105] For porosity, values spanned from 35% to 99% across both scaffold types. In both 4UC (dceii / h = 4) and 2UC (dceii / h = 2) models, decreasing dceii / h shifted P to lower values, while preserving the wide range of P achievable for each layer. Among the input parameters, dstrut / h had the strongest influence on porosity, with larger struts decreasing pore volume. S / V ranged from 0.06 mm-1to 2.6 mm-1in 4UC and nearly doubled in 2UC, reaching up to 4 mm-1. Unlike porosity, S / V was most sensitive to dcei / h and n, while dstrut / h had only a marginal effect. Lower dceii / h and higher n increased surface complexity, leading to higher S / V. Maximum rotation (ip) ranged from 5° to 30° in 4UC and from 20° to 60° in 2UC designs, confirming that taller unit cells (h = 10 mm) increase rotational response. Among the input variables, dceii / h (strut orientation) exerted the greatest effect, followed by <p. In contrast, dstrut / h and n had minimal impact.

[0106] For the normalized elastic modulus (Eetf / Ebuik), values reached as high as 1.28* 10“3in 4UC and upto 0.7* 10“3in 2UC. This reduction in elastic modulus for taller scaffolds is consistent with increased compliance due to longer struts. Eeff / Ebuik was found to be strongly dependent on both dstrut / h and <p, with thicker struts and smaller rotation angles improving load resistance, while the effect of n was comparatively minor. Strain energy density (U) spanned from near-zero values to 33 kJ / m3in 4UC, dropping to 15 kJ / m3in 2UC. Similarly to Eeti / Ebuik, U was predominantly influenced by dstrut / h and <p. Designs with thicker struts and lower orientation angles exhibited more energy absorption. Comparing U with Eeff / Ebuik shows, they have similarities in the trend of values. However, the influence of each input differed when switching from E to U.

[0107] The multi-objective optimization revealed a diverse set of non-dominated solutions within both the 4UC and 2UC design spaces. As shown in Fig. 15, the resulting Pareto fronts illustrate the trade-offs between the four competing objectives for the two defined goals. In scenario A, designs achieving maximum Eeff / Ebuik and minimum ip while maintaining porosity near 80% and maximizing S / V were identified. Scenario B prioritized maximizing U instead of Eeff / Ebuik under similar constraints, leading to alternative optimal configurations. In both cases, red markers represented Pareto- optimal solutions and black stars denoting the compromised solutions closest to the Utopia point (Fig. 15). These points achieved balanced performance across all four objectives and served as the basis for the inverse design in each scaffold layer.

[0108] The Pareto fronts identified a broad spectrum of non-dominated designs where no objective could be improved without compromising others. Fig. 15 illustrates these fronts, with red markers representing Pareto-optimal solutions and black stars denoting the compromised solutions closest to the Utopia point. These points achieved balanced performance across all four objectives and served as the basis for the inverse design in each scaffold layer.

[0109] The inverse design process successfully translated Pareto-optimal solutions into specific input parameters for each scaffold layer. To accommodate the mechanical constraint of a shared rotation angle ip across layers, alternative optimal points were strategically selected for the intermediate and inner layers. This adjustment ensured coherent folding behavior and uniform deformation throughout the scaffold structure. The resulting CAD models were fabricated using SLS, yielding scaffolds with high geometric fidelity and symmetry, as shown in Fig. 16. The approach also demonstrated strong scalability (Fig. 17), with consistent structural features maintained across varying scaffold sizes — highlighting the method’s adaptability to diverse anatomical requirements.

[0110] FE analysis was performed on single unit cage (UC) models from each optimized design group (Models A and B, with both 4UC and 2UC configurations), subjected to 25% compressive strain. The simulations revealed distinct buckling and bending behaviors of the struts, aligning with the expected deformation modes. Compressive loading consistently induced rotation of the top disc, confirming the presence of intrinsic kinematic coupling within the scaffold’s architecture.

[0111] Asymmetric deformation patterns emerged as a result of imposed geometric imperfections. Strut behavior was dependent on length: shorter struts exhibited outward buckling, whereas longer struts deflected inward. These deformation modes were visually distinct in model A with 2UC and in model B (both 2 and 4 UC), and less evident in model A with 4UC, where strut lengths were more uniform.

[0112] Strain distribution analyses revealed localized concentrations at mid-strut regions and disc junctions. Models with 4UC generally exhibited higher peak local strains. For instance, model A with 4UC reached compressive and tensile strain peaks of 28% and 23%, respectively, under 25% global compression — compared to 20% and 12% in the corresponding 2UC model. A similar trend was observed in model B. In all cases, outer scaffold layers experienced greater local strain magnitudes than intermediate and inner layers, highlighting the influence of radial positioning on mechanical response.

[0113] Normalized stress-rotation plots (Fig. 20) revealed that 4UC models underwent less rotation than 4UC models at the same apparent strain level, indicating foldability with less shear stimuli due to shorter unit height. These trends validate the scaffold’s rotational mechanics and support design adaptation for variable deformation behavior.

[0114] CFD simulations were conducted for all four scaffold configurations, using both undeformed and 25% compressed geometries. Velocity streamlines showed maximum flow near strut intersections, especially in narrowed regions under 25% compression (Fig. 21 b and Fig. 22b). These geometric bottlenecks resulted in elevated local velocities and disrupted flow uniformity.

[0115] The 25% compressive deformation altered the geometry of the unit cages, creating new flow paths while constricting others. This led to localized velocity peaks and pronounced shifts in flow direction. Notably, in Model B with 4UC, the deformed configuration caused major flow disturbances near the top and bottom of the scaffold (where struts crossed), as well as in the mid-region, where flow paths were partially obstructed (Fig. 22b). Across all models, compression consistently amplified these disturbances, emphasizing the sensitivity of internal flow dynamics to scaffold deformation.

[0116] Pressure drop analysis revealed that 4UC scaffolds exhibited higher baseline flow resistance compared to 2UC configurations. Following 25% compression, pressure drops increased markedlly by 92.9%, 85.2%, 91.9%, and 122.5% in models A (4UC), A (2UC), B (4UC), and B (2UC), respectively indicating a substantial reduction in fluid conductivity (Fig. 21c and Fig. 22c). Correspondingly, equivalent permeability dropped by approximately 48-53% across all models (Fig. 23). For example, permeability in model A (4UC) decreased from 2.7*10“8to 1.4*10“8m2

[0117] Wall shear stress contours showed localized shear concentration near strut tips and junctions. In the undeformed configurations, maximum shear was primarily observed at the top and bottom surfaces of the struts. Upon deformation, new shear hotspots emerged along the mid-strut regions, attributed to redirected flow paths (Fig. 21d and Fig. 22d). Azimuth-facing strut sides experienced higher shear stress comp9ared to radial-facing ones.

[0118] Fig. 24 shows the bulk porosity and surface area-to-volume (S / V) ratio for each scaffold configuration in both undeformed and deformed states. In the reference configuration, porosity ranged from 75.1% (Model B (4UC)) to 89.6% (Model A (2UC)). All models exhibited a decrease in porosity upon 25% compression; for instance, Model A (4UC) showed a 7.3% reduction, while Model B (2UC) exhibited an 11.1% decrease.

[0119] The S / V ratio varied from 0.73 mm’1(Model A (2UC)) to 1.35 mm’1(Model B (4UC)) in the undeformed state. Compression led to a significant increase in S / V across all models: Model A (2UC) showed a 23% increase, whereas Model B (4UC) exhibited the largest increase at 54.3%. When comparing these structural parameters to equivalent permeability values (Fig. 24c), clear trends emerged. For instance, Model B (2UC), which underwent an 11.1% porosity reduction and a 48.1% S / V increase, experienced a pronounced 55.2% drop in equivalent permeability. In contrast, Model A (2UC), with only a 1.8% decrease in porosity and a 23% increase in S / V, showed a moderate permeability reduction of 46.1%. These results suggest that scaffolds with larger simultaneous reductions in porosity and increases in S / V tend to exhibit greater declines in permeability under compression.

[0120] Fig. 25 summarizes the results of the quasi-static axial compression tests conducted on the fabricated scaffolds. The stress-strain curves for all samples are shown in Fig. 25a. All models exhibited a characteristic response consisting of an initial linear elastic region, followed by a nonlinear buckling phase, and culminating with a densification region marked by a rapid increase in stress due to strut contact. The extent of the linear region varied among models, reaching up to 4% in Models A (4UC) and B (4UC), 3% in Model B (2UC), and 2% in Model A (2UC). Across both models, the 4UC scaffolds demonstrated lower linear elastic moduli and higher strain energy densities compared to their 2UC counterparts (Fig. 25b and Fig. 25c). Model A, optimized to maximize normalized elastic modulus (Eetf / Ebuik), showed lower values for both Eetf / Ebuik and strain energy density (U) compared to Model B, which prioritized U over E in its optimization strategy.

[0121] Visual inspection of sample deformation confirmed consistent foldability in Models A (2 and 4 UCs), with each unit cage compressing fully to its minimum height. Model B (2UC) also displayed acceptable folding behavior. In contrast, Model B (4UC) exhibited strut misalignment during compression, leading to premature densification and early termination of the test.

[0122] The present application introduces a novel class of scaffolds — Origami-Inspired Super-Compressible Scaffolds (the device 10) — designed specifically to meet the evolving mechanical and geometrical demands of distraction osteogenesis. An aspect of the device 10 design lies in its Kresling-inspired origami architecture, which enables large compressive deformations while preserving structural coherence. In contrast to traditional TISs, which may fail to adapt to the dynamic elongation of the osteotomy gap, the device 10 accommodates expansion without compromising mechanical integrity. This adaptability stems from the design strategy based on concentric, strutbased layers capable of controlled folding and deformation. While previous products have explored porous scaffolds for bone healing, and more recently leveraged machine learning for scaffold optimization, none have proposed a foldable scaffold architecture specifically adapted to the evolving geometry of distraction osteogenesis. Importantly, this work is the first to implement and validate a comprehensive multi-objective optimization framework for distraction osteogenesis, integrating data-driven machine-learning predictions, mechanical simulations, and experimental validation — marking a significant advance over previous approaches.

[0123] The several analyses conducted highlight the strength of BML to navigate complex scaffold design spaces. The GPR models accurately predicted key structural and geometric properties - including porosity, surface area to volume ratio, maximum rotation, elastic modulus, and strain energy - across a wide range of input combinations. This predictive capacity enabled efficient exploration of foldable regions within the design space and informed the selection of Pareto-optimal configurations. Notably, as shown in Fig. 14, structural parameters such as strut diameter and cell size exhibited non-linear and occasionally counterintuitive effects on performance metrics.

[0124] The multi-objective optimization strategy successfully balanced competing mechanical and biological criteria, identifying designs closest to a theoretical Utopia point in the objective space. As illustrated in Fig. 15, the optimization process revealed the trade-offs between maximizing elastic modulus or strain energy while maintaining porosity near 80% and minimizing rotational deformation. This approach yielded designs that not only met geometric and mechanical requirements but also offered favorable fluid transport characteristics (Figs. 21 and 22), which are critical for bone regeneration in large bone defects.

[0125] The S / V and porosity values of the device 10 designs aligned well with those reported in other high-performing scaffold architectures, such as MFCC scaffolds and TPMS. For instance, MFCC scaffolds exhibited porosities between 70-90%, with optimal designs centered around 80% porosity and S / V values reaching up to 3.71 mm-1for MFCC-1 and 2.21 mm-1for MFCC-2. Similarly, TPMS-based scaffolds such as gyroid and P-lattice structures optimized for porosities in the 80-90% range, yielded S / V values from 0.3 to 2.9 mm-1depending on geometry and lattice type. While direct biological assays were beyond the scope of this work, the device 10 design achieved comparable porosity and S / V ranges known to support the device 10’s nutrient transport and osteogenic cell adhesion. To assess functional performance under physiological loading, CFD simulations on both undeformed and 25% compressed geometries were conducted. While permeability is frequently estimated based solely on porosity, the present application reveals this approach can overlook critical structural effects. In particular, the analyses show that S / V also influences permeability independently of porosity — likely by modifying internal flow resistance and tortuosity. This distinction underscores that porosity and S / V are intrinsic, designable parameters that influence different performance aspects: porosity primarily determines the availability of space for fluid and cell migration, while S / V is closely associated with cell attachment, protein adsorption, and local mechanotransduction. Permeability, in contrast, emerges as a resulting property shaped by these geometric inputs, but is more challenging to optimize directly without trade-offs. Additionally, wall shear stress (WSS), which is derived from local flow velocity and scaffold surface interactions, was shown to decrease markedly under compression. This suggests that scaffold deformation during DO not only reduces permeability but may also diminish the mechanical stimuli experienced by resident cells.

[0126] These observations underscore the need to evaluate scaffold performance under realistic loading, especially in DO where compression evolves during healing. Future CFD-driven models could help identify the minimal osteotomy gap required to sustain effective fluid transport and WSS during the early healing phase — offering a data- informed threshold for scaffold deployment in clinical settings.

[0127] Mechanical testing and FE simulations further characterized the compressive response of the device 10 designs. The mechanical tests confirmed a typical three- phase stress-strain profile: an initial linear region, followed by a plateau associated with strut buckling, and culminating in densification due to contact between folded struts. FE simulations, while limited to 25% compressive strain, captured the transition from linear to buckling regimes and highlighted features such as mid-strut strain concentration and coordinated folding — critical during early distraction. The experimentally observed deformation patterns in elastomeric prototypes were qualitatively consistent with simulation results, supporting the potential of the device 10 to undergo large, recoverable deformations. This robustness opens the possibility of fabrication with clinically relevant biodegradable polymers such as PCL, which has demonstrated similar performance in porous scaffolds. Despite its contributions, the present disclosure has several limitations. Biological performance - the device 10 including cellular response, tissue integration, and degradation - remains to be investigated. The FE and CFD models relied on simplified material properties and boundary conditions that may not fully capture the complexity of in vivo loading during DO. Moreover, scaffold optimization in the present application focused on structural and mechanical criteria without incorporating biological functionalization, which may be taken into account as well.

[0128] The proposed framework, supported by multi-objective optimization and ML, enables the design of foldable scaffolds with tunable mechanical, geometric, and structural properties tailored to the evolving requirements of distraction osteogenesis.

[0129] The above-mentioned analyses demonstrate that the proposed design framework effectively balances competing performance metrics, including porosity, mechanical integrity, surface-to-volume ratio, and deformation adaptability. Through the integration of FE and CFD simulations, key insights were gained into scaffold behavior under large compressive strains. In particular, critical structure-function trade-offs affecting permeability and wall shear stress were identified, emphasizing the importance to evaluate scaffold performance under physiologically realistic loading conditions, especially during early-stage distraction osteogenesis.

[0130] By enabling the design of scaffolds that are both mechanically adaptive and biologically permissive, the device 10 platform opens new avenues for patient-specific scaffold solutions in regenerative medicine. Moreover, the computational methodology developed in the present application holds potential for broader application in contexts requiring dynamic mechanical compliance. Future efforts will focus on in vivo validation and biological functionalization to support osteogenesis and vascularization, with the ultimate goal of enhancing clinical outcomes in the treatment of large and complex bone defects.

[0131] The above description is meant to be exemplary only, and one skilled in the art will recognize that changes may be made to the embodiments described without departing from the scope of the invention disclosed. Still other modifications which fall within the scope of the present invention will be apparent to those skilled in the art, in light of a review of this disclosure, and such modifications are intended to fall within the appended claims.

Claims

CLAIMS1. A scaffold device for allowing distraction movement between segments of a bone, the scaffold device comprising: a first platform, a second platform, and struts extending from the first platform to the second platform, wherein the scaffold device is configured to be deformed into position between segments of a bone, in a deformed state, from an undeformed state, and wherein, when in the deformed state, the struts buckle and / or flex to bias the first platform and the second platform away from one another toward the undeformed state.

2. The scaffold device according to claim 1 , wherein the device has a monolithic body of biocompatible material.

3. The scaffold device according to claim 2, wherein the monoblock is made of a biodegradable material.

4. The scaffold device according to claim 3, wherein the biodegradable material is polycaprolactone.

5. The scaffold device according to any one of claims 1 to 4, wherein the struts are arranged into at least a first annular distribution of interconnected struts.

6. The scaffold device according to claim 5, including at least a second annular distribution of interconnected struts.

7. The scaffold device according to any one of claims 1 to 6, wherein the first platform and / or the second platform is an annular disk.

8. The scaffold device according to any one of claims 1 to 6, wherein the first platform and / or the second platform is a polygon.

9. The scaffold device according to any one of claims 1 to 8, wherein the struts are arranged into a zig-zag pattern, in which adjacent ones of the struts merge at or adjacent to the first platform or the second platform.

10. The scaffold device according to claim 9, wherein the adjacent ones of the struts merging emulate clamped joints.11 . The scaffold device according to claim 9, wherein the adjacent ones of the struts merging emulate hinged joints.

12. The scaffold device according to claim 9, wherein the struts in the zig-zag pattern include longer struts and shorter struts, wherein, in the deformed state of the scaffold device, a majority of the longer struts is in tension and a majority of the shorter struts is in compression.

13. The scaffold device according to claim 12, wherein in the undeformed state of the scaffold device, a majority of the longer struts is in compression and a majority of the shorter struts is in tension.

14. The scaffold device according to claim 12 or claim 13, wherein, in the zig-zag pattern, the struts alternate in sequence between longer strut and shorter strut.

15. The scaffold device according to any one of claims 1 to 14, wherein the struts include longer struts and shorter struts, wherein, in the deformed state of the scaffold device, a majority of the longer struts is in tension and a majority of the shorter struts is in compression.

16. The scaffold device according to any one of claims 1 to 15, wherein the struts all have an equal length.

17. The scaffold device according to claim 6, wherein the struts of the first annular distribution all have an equal length.

18. The scaffold device according to claim 17, wherein at least some of the struts of the second annular distribution have a length differing from the equal length of the struts of the first annular distribution.

19. An assembly for use in distraction osteogenesis comprising: at least two of the scaffold device according to any one of claims 1 to 18, wherein the at least two scaffold devices are deformed into position between segments of a bone in distraction osteogenesis into the deformed state, and wherein, when compressed to the deformed state between the segments of the bone, adjacent ones of the at least two scaffold devices contact one another withthe first platform of one of the scaffold devices being against the second platform of the other of the scaffold devices.

20. The assembly according to claim 19, wherein the platforms in contact with the segments of the bone have at least one attachment part projecting from the circumference of the platform to the bone outer surface.

21. The assembly according to claim 20, wherein the attachment part has a hole.

Images