Designing a custom medical device

Abstract

A computer implemented method 10 of designing a custom medical device comprising a biomechanical interface for interaction with an external body segment of a patient comprising receiving input data co

Description

The present invention relates to a computer implemented method, a computer readable storage medium, and an apparatus. Custom medical devices for interaction with an external body segment of a patient may be designed specifically for the patient according to one or more measurements of the external body segment. According to a first aspect of the present techniques there is provided a computer implemented method of designing a custom medical device comprising a biomechanical interface for interaction with an external body segment of a patient, the method comprising: receiving input data comprising one or more measurements of the external body segment; generating, using a computational model, an initial design of the biomechanical interface based on the input data, the initial design defined by a plurality of interface parameters selected based on at least one design criterion; and in response to receipt, by the computational model, of user feedback information indicating at least one modification to be applied to the initial design, generating, using the computational model, a modified design of the biomechanical interface, wherein generating the modified design comprises identifying at least one relationship between the at least one modification and the plurality of interface parameters, and re-generating at least one of the plurality of interface parameters based on the at least one design criterion and the at least one relationship. According to a second aspect of the present techniques there is provided a computer readable storage medium configured to store program code comprising instructions which, when executed by a processor, cause the processor to perform the computer implemented method according to the first aspect. In some configurations, the computer readable storage medium according to the second aspect is a non-transitoiy computer readable storage medium. According to a third aspect of the present techniques there is provided an apparatus for designing a custom medical device comprising a biomechanical interface for interaction with an external body segment of a patient, the apparatus comprising: receiving circuitry configured to receive input data comprising one or more measurements of the external body segment; computation circuitry configured to generate an initial design of the biomechanical interface based on the input data, the initial design defined by a plurality of interface parameters, wherein the computation circuitry is configured to select the plurality of interface parameters based on at least one design criterion evaluated at the biomechanical interface, wherein the computation circuitry is configured: to receive user feedback information indicating at least one modification to be applied to the initial design; in response to receipt of user feedback information identifying at least one relationship between the at least one modification and the plurality of interface parameters, to generate a modified design of the biomechanical interface by regenerating at least one of the plurality of interface parameters based on the at least one design criterion and the at least one relationship. The present techniques will be described further, by way of example only, with reference to configurations thereof as illustrated in the accompanying drawings, in which: Figure 1 schematically illustrates a computer implemented method according to some configurations of the present techniques; Figure 2 schematically illustrates an apparatus according to some configurations of the present techniques; Figure 3 schematically illustrates a computer implemented method according to some configurations of the present techniques; Figure 4 schematically illustrates a computer implemented method according to some configurations of the present techniques; Figure 5a schematically illustrates a database storing a plurality of previous designs according to some configurations of the present techniques; Figure 5b schematically illustrates a database storing a plurality of previous designs according to some configurations of the present techniques; Figure 6 schematically illustrates a computer implemented method according to some configurations of the present techniques; Figure 7 schematically illustrates a sequence of steps carried out according to some configurations of the present techniques; and Figure 8 schematically illustrates a sequence of steps carried out according to some configurations of the present techniques. The design of biomechanical interfaces between an external body segment of a patient (e.g., that are in close contact with the skin of a patient) and a custom medical device has traditionally been achieved by taking a mould (e.g., a plaster cast) of the external body segment and using that to create an initial model. The designer for example, a clinician specialising in the particular custom medical device) may then use their expertise and knowledge of the patient (for example, the patient’s condition, lifestyle and / or preferences) to modify that design, e.g., by adding or removing material from the mould, a process known as “modification” or “rectification”, to create a final design that can then be fabricated to form the finalised biomechanical interface. These modifications may be focused at particular locations, or gross changes to size whilst maintaining shape (e.g., scaling) which can be variable in extent across the shape. This process requires a number of manual steps and relies heavily on the expertise of the designer. An initial design of a biomechanical interface could, theoretically, be derived using computational methods by taking detailed measurements of the external body segment of the patient, in order to determine an approximate shape and size of the external body part. An initial design could then be produced based on that approximate shape. However, generally as in the case with physical moulds used to design such interfaces, the initial design would then be significantly modified by the designer (for example, a clinician specialising in the particular custom medical device) to produce a final design. Such modifications may be based on the clinician’s knowledge of the patient (for example, the patient’s condition, lifestyle and / or preferences), the medical history of the patient, prior experience that the clinician has had producing custom medical devices for other patients, and / or knowledge of what use the patient would like to make of the custom medical device. According to some configurations of the present techniques there is provided a computer implemented method of designing a custom medical device comprising a biomechanical interface for interaction with an external body segment of a patient. The method comprises receiving input data comprising one or more measurements of the external body segment. The method further comprises generating, using a computational model, an initial design of the biomechanical interface based on the input data, the initial design defined by a plurality of interface parameters selected based on at least one design criterion. The method also comprises, in response to receipt, by the computational model, of user feedback information indicating at least one modification to be applied to the initial design, generating, using the computational model, a modified design of the biomechanical interface. Generating the modified design comprises identifying at least one relationship between the at least one modification and the plurality of interface parameters, and re-generating at least one of the plurality of interface parameters based on the at least one design criterion and the at least one relationship. The user as referred to herein is any user of the computer implemented method. The user may therefore be a clinician specialising in custom medical devices. The user may also be referred to as a clinician or a designer. The patient as referred to herein is the person for whom the custom medical device is being designed. In some use cases, the patient and the user may be the same person, although this will typically not be the case. In some use cases, the user feedback may be provided based on discussion with the patient. The inventors of the current techniques have recognised that regardless of the level of detail and information that is provided as a starting point for the initial design, there are likely to be many factors known to the designer that could not be trivially factored into the initial design and that, even with significant refinements to the techniques used to create the initial design, there may still be details, experiences, and indeed preferences of the patient that are known to the clinician and that are not present or easy to represent in a model used to generate an initial design. The computer implemented method therefore provides a method that seeks to best utilise the skills of the clinician when generating a modified design. In particular, the inventors have realised that typically, modifications made by the clinician to the initial design are unlikely to be single isolated modifications and that, instead, where a clinician makes one modification, it is likely that further interconnected modifications to the design may be required elsewhere to account for the initial modification. As an illustrative example, a clinician may be provided with an initial design of a biomechanical interface for a custom medical device. However, the clinician may know that for the specific patient using the custom medical device in a specific way the particular design may cause the biomechanical interface to exert pressure on the patient at a specific location in a manner that could cause discomfort, pain and / or injury. The clinician may therefore seek to modify the design to reduce the pressure exerted on the patient at that location. Dependent on the specific nature of the custom medical device (examples of which will be described in further detail below), this may cause the custom medical device to exert pressure unevenly on other portions of the external body segment and / or may cause the device to fit / perform poorly. Therefore, in general the clinician seeking to make use of their knowledge for the specific patient using the custom medical device in a specific way, would be required to make a range of further modifications in order to compensate for unwanted effects resulting from the initial modification. As a result, the overall design process can be burdensome in terms of time and expertise required. The computer implemented method therefore uses a computational model that generates the initial design based on input data identifying one or more measurements (e.g., anatomical measurements of size and / or shape) of the external body segment. The initial design is defined by plural interface parameters which are chosen by the computational model based on at least one design criterion. The at least one design criterion may be any requirement from which an initial design can be generated. For example, the design criterion may specify a degree of conformity of the initial design to a shape of the external body segment, a specification of one or more areas for which there should be space provided between the biomechanical interface and the body segment, an overall size criterion indicating an amount of the external body segment that should interact with the biomechanical interface, and / or a similarity of the biomechanical interface to a pre-existing design for a custom medical device of the same type for which the one or more measurements are similar to the one or more measurements used to generate the initial design. In some configurations the at least one design criterion may also include a specification relating to a pressure measurable (e.g., calculated) at the biomechanical interface and / or a shear stress measurable at the biomechanical interface. The at least one design criterion may relate to a pressure gradient or a shear stress gradient measured or calculated at the interface. The computer implemented method is responsive to user feedback information identifying at least one modification to be made to the initial design. The modification may be, for example, a rectification, a change, or an amendment to the initial design. The computational model is configured to identify relationships between the plurality of interface parameters. Such relationships may be stored in a lookup table, stored as one or more functions, and / or may be based on one or more physical rules. Further details relating to the relationships will be described below. The computational model is responsive to receipt of the user feedback to identify a relationship (or plural relationships) that identify links between the user specified modifications and the one or more interface parameters. Continuing the above example, the computational model may identify a relationship between a modification seeking to reduced pressure on the patient at an identified location and one or more other parameters that would typically be modified to account for this change. The relationships may, for example, comprise a relationship between two of the interface parameters. For example, a volume change (e.g., a volume increase or a volume reduction), and an offloading parameter defining a change in shape of the biomechanical interface. Considering an example in which the custom medical device is a prosthetic and the biomechanical interface is a socket, a total reduction in volume (i.e., a reduction in size of the socket) would typically apply an increased pressure on all portions of the external body part of the patient. In general, a prosthetic limb is fitted to a residual portion of a limb which is not a homogeneous structure. Rather, the residual portion of the limb contains bony prominences such as the fibular head and the anterior tibial crest for a transtibial prosthetic socket. Therefore, a reduction in volume would increase the pressure that the socket exerts and may cause discomfort, pain and / or injury at these bony prominences. A typical relationship may therefore identify that a volume reduction should be accompanied by a change in shape to offload the pressure from the bony prominences. Alternatively, a relationship may identify that a volume reduction should result in a greater amount of space provided at the distal end of the residual limb. In addition, the relationships may be relationships between the input data and design parameters. For example, the one or more measurements may indicate whether the body segment is wide or slender. For some types of customised medical devices, particularly those that are press fitted to the patient, a greater pressure may need to be applied to a wider limb because a wider limb is more likely to be composed of a thicker layer of soft tissue. It will be readily apparent to the person skilled in the art that these relationships have been indicated purely by way of example and that there may be numerous different relationships that between the interface parameters that could be provided. Having identified the relationships, the computational model then generates a modified (e.g., a revised, or changed) design in which one or more of the plurality of interface parameters are re-generated. The re-generation of the one or more interface parameters does not require that the one or more interface parameters change. For example, in some use cases, the user may make a modification to the design that does not affect any other parameters. In such a situation, the regenerated model may comprise the initial design as modified by the user but with no further modifications. The provision of a computer implemented method that receives user feedback on an initial design and that, using the initial design and the modification provides a modified design taking into account relationships between different interface parameters in the model allows an improved design to be created whilst reducing the burden on the user in terms of the amount of time that the user has to spend refining a design, and by supporting the experience and expertise of the user. As a result, the modification or rectification stage of the design process can be made more efficient and provides a computerised method enabling improved design and, subsequently, fabrication of custom medical devices. The at least one relationship may be derived in any manner and may be different between different relationships. For example, the at least one relationship may be based on one or more physical relationships between parameters. In some configurations the at least one relationship is derived using one or more statistical models. The statistical models may, for example, indicate a likelihood that a particular type or size of modification, either to the global size and shape of the biomechanical interface or on a particular portion of the biomechanical interface will require one or more further modifications to be made. The statistical model may form part of the computational model along with one or more other statistical or deterministic models comprised in the computational model. Alternatively, the statistical model may be provided separate to the computational model. The statistical models as defined herein are used to generate the one or more relationships. The computational model makes use of the relationships to generate the initial and modified designs. The computational model may make use of statistical or deterministic modelling techniques to generate the designs. Using the relationships in combination with the computational model, a modified design may be generated that takes the modification and the relationships derived from by the statistical model into account. In some configurations, the modified design may be one of a plurality of modified designs that are presented to the user. For example, it may be that the modification to the design could result in multiple different options for how the interface parameters should be modified. In such a configuration, the user may be presented with the plurality of modified designs and an indication of a most likely or best choice design according to the relationships derived by the statistical models. The statistical models may comprise one or more precomputed models derived based on observation of existing biomechanical interfaces. In some configurations the at least one relationship is derived from a database comprising a plurality of entries, each of the plurality of entries identifying a corresponding previous design. Each of the plurality of entries may have a plurality of interface parameters with the one or more relationships developed based on correlations between different ones of the plurality of interface parameters. For example, different ones of the one or more relationships may relate a first parameter (e.g., a measure of conformity of the biomechanical interface to a corresponding external body segment, an overall size difference between the external body segment and the biomechanical interface, a length of a defined sub-portion of the biomechanical interface, a shape of a defined sub-portion of the biomechanical interface, etc.) to a second parameter (e.g., a measure of conformity of the biomechanical interface to a corresponding external body segment, an overall size difference between the external body segment and the biomechanical interface, a length of a defined sub-portion of the biomechanical interface, a shape of a defined sub-portion of the biomechanical interface, etc.). Deriving the relationships from the database of previous designs provides the advantage that the modified design can be generated, in response to the user feedback information, such that the modified design takes into account the knowledge of previous designs providing an updated design that is more likely to align with the user’s knowledge of biomechanical interfaces of the particular type of custom medical device that is being designed. In some configurations, for each of the plurality of entries, the corresponding previous design comprises at least one of: a corresponding initial design; and a corresponding modified design identifying one or more modifications applied when generating the modified design. The corresponding initial designs may be added to the database as they are generated. Alternatively, in some configurations only initial designs which result in an accepted final design may be added to the database. In some configurations the corresponding modified design may be stored as an initial design stored in combination with the modifications required to generate the corresponding modified design. In some configurations the corresponding modified design may be stored as the final modified design only, or the final modified design in combination with the initial design or a set of modifications required to generate the final modified design. In some configurations each modified design may be stored in the database regardless as to whether it is accepted as a final design, rejected, or further modified. In some configurations each entry of the database comprises previous input data received for generation of that one of the previous designs, wherein the computational model is configured to generate the initial design through comparison of the input data against the previous input data for the plurality of entries. The database may comprise a first database for storing data in a form appropriate for generation of the initial design, and a second database for storing data in a form appropriate for generation of the modified design. In some configurations a single database may be provided that stores data in an appropriate form for both the generation of the initial design and the generation of the modified designs. In some configurations the database may be a static database that provided as part of the computational model, in other configurations the database may be a dynamically varying database that is updated through the generation of designs. In some configurations the user may be given the option to add a new design to the database of designs. The statistical models may then be dynamically regenerated to account for the new designs stored in the database. In some configurations the computational model is configured to generate the initial design through interpolation between the input data for the plurality of entries. The term interpolation as used herein refers to the determination of a set of values from one or more closest entries of the plurality of entries. The closest entry or entries of the plurality of entries may be determined by comparing the input data against the corresponding input data for each of the plurality of entries and determining the entry or entries that have corresponding initial data that is closest to those entries. The comparison may be a straightforward shortest distance comparison with each of the one or more measurements for the input data stored in a vector Y with entries Yi representing the i-th one of the one or more measurements. The corresponding entries in the database may also store the corresponding input data in a vector where the superscript j identifies the particular entry in the database and the entry X / representing the i-th one of the one or more measurements. The closest entry may then be determined by calculating the distance between the vectors = li(Yt — X / )2 and selecting the entry j for which is smallest as the closest entry. Alternatively, a weighted metric may be used to select a closest design, for example the distance Dj may be calculated as = Xi 1¾(Yt — X / )2 where Wt is the i-th entry of a vector W of weights used to alternately weight each of the one or more measurements when performing the interpolation. The interpolation may then be performed based on the closest entry, e.g., by selecting the design having the closest values as the initial design or as a starting point for one or more further modifications when generating the initial design. Alternatively, the interpolation may be performed by selecting all previous designs that are close to input data (e.g., the group of entries that are within a given tolerance of the closest entry) and, for a given one or more of the plurality of interface parameters, performing a mathematical interpolation between the corresponding given parameters in each of the one or more entries. It will be readily apparent to the person skilled in the art that alternative methods of selecting the closest entry or entries may be provided. As discussed, in some configurations the database may be fully static, or may be dynamically updated as new designs are added with the one or more relationships remaining static or being updated in response to an updating action taken by the user. In some configurations the method comprises, in response to a determination that the updated design is accepted, updating the database to store a new entry; and updating each of the one or more relationships based on the new entry. This approach leads to a self-teaching model which can allow the models to be improved for each successful design. In some configurations the user may be provided with the option to add a design to a database. Alternatively, or in addition, in some configurations the method may comprise providing a user with a choice of whether or not to generate the initial design and / or the model design based on a predefined library of designs only or based on a combination of the predefined library of designs and those designs added by the user. As discussed the at least one design criterion can vary dependent on the particular implementation. However, in some configurations the at least one design criterion comprises at least one of: a preferred size criterion specifying a preferred size difference between the external body segment and the biomechanical interface; and a predefined conformity criterion specifying a preferred degree of conformity between a shape of the biomechanical interface and a shape of the external body segment. Dependent on the particular type of custom medical device, the biomechanical interface may be an attachment interface configured to attach the custom medical device to the external body segment. For example, in some configurations the biomechanical interface may be provided as a socket having a particular circumferential length (perimeter length). Alternatively, the custom biomedical device may be designed to allow pressure to be applied to the segment of the body in a distributed manner so as not to provide undue pressure on a given area. The size criterion may and / or the conformity criterion may therefore directly relate to the load distribution that the user experiences and, in some configurations, relates to the attachment strength between the custom medical device and the body segment of the user. Hence, the provision of a model that generates the one or more designs according to one or more of these design criteria allows the user to ensure a satisfactory initial fit of the biomechanical interface. In some configurations the computational model is configured to generate a plurality of possible initial designs; selecting the initial design from amongst the plurality of possible initial designs comprises evaluating each of the plurality of possible designs against the at least one design criterion and selecting one of the plurality of possible designs having a highest evaluation. As discussed, the initial design may be based on interpolation between different previous designs. Each of the previous designs may be provided with a score dependent on how close the corresponding one or more measurements of that design are to the one or more measurements of the user. Alternatively, or in addition, a statistical approach may be taken based on one or more statistical models. For example, the one of more statistical models may be based on a Naive Bayes approach, Bayesian Belief Networks, Markov Chains, Decision Trees, Markov Networks (Markov Random Fields), Hidden Markov Models (HMMs), Gaussian Processes, Dynamic Bayesian Networks (DBNs), Conditional Random Fields (CRFs), Probabilistic Graphical Models (PGMs), and / or Neural Network based models. The one or more statistical models may provide an estimate of a probability that each of the plurality of possible initial designs will be appropriate based on the set of input measurements. Once the plurality of possible initial designs has been assigned a probability, the designs may be evaluated against the design criterion based, for example, starting from the possible initial design having the greatest probability of being appropriate based on the set of input measurements. It will be readily apparent to the skilled person that alternative methods of generating an initial design could be provided. In some configurations the measurements Yt comprise coefficients of a first few components of a principal component analysis applied on a canonical sampling of the limb shape. An initial design comprising a plurality of interface parameters Pk could be derived using a generalised linear relationship between the one or more measurements Yi and the interface parameters: Yt = AikPk such that for a given set of measurements, the interface parameters Pk can be generated by solving Y: = AikPk. In general, this linear system of equations will be over defined, i.e., there is a greater number of measurements than interface parameters. An approximate solution of the system may be determined, for example, using QR factorisation, singular value decomposition, a Moore-Penrose pseudo inverse, or another mathematical inversion technique as would be known to the person skilled in the art. The matrix Aik may be derived for example from the input measurements for each of plurality of stored designs X- and the corresponding interface parameters for those designs Qk for all entries in the database j, for example, the entries of the matrix Aik could be derived using one or more linear fitting techniques and / or one or more statistical techniques. In other words, given a database of stored designs having measurements stored designs X- for entry j, where the measurements comprise coefficients of a first few components of a principal component analysis applied on a canonical sampling of the corresponding limb shape (the limb shape used for the j-th entry in the database) and corresponding parameters Qk for that entry, a linear fit can be performed to identify components Aik that result in a best fit of the data over all entries in the database. Existing parameter sets in the database Qk that correspond to a given set of measurements X. may be derived, prior to implementation of the method, through statistical shape modelling techniques and / or analysis by an clinical expert in the field. Subsequently, the matrix Aik can be used to approximate Pk for a given set of measurements Yj. Subsequently, when the user modifies and / or fixes one or more of the interface parameters Pk by fixing Pk equal to Pk for a given set of k, a new set of those interface parameters that have not been fixed can be obtained by solving a modified linear system in which a subset of the parameters Pk are fixed. The modified linear system, which in general would be over defined when all Pk are variable, is further constrained by these additional requirements resulting in a further over definition of the system. As discussed above, the over defined system may not be uniquely solvable. However, these systems can still be approximately solved, e.g., they can be solved assuming that some error is tolerable and that the error in the solution is minimised. Alternatively, once an initial design has been provided, the modified design can be approximated based on observed correlations between different groups of the variables. As discussed, the observed correlations may be derived using a Naive Bayes approach, Bayesian Belief Networks, Markov Chains, Decision Trees, Markov Networks (Markov Random Fields), Hidden Markov Models (HMMs), Gaussian Processes, Dynamic Bayesian Networks (DBNs), Conditional Random Fields (CRFs), Probabilistic Graphical Models (PGMs), and / or Neural Network based models. The observed correlations may be used, in some configurations, in place of re-computing the solution to the linear system. For example, relations between parameters could be derived in which, for each interface parameter, a Naive Bayes score is computed against all other interface parameters and the highest Naive Bayes score is used to determine the suggested value. In other words, the Naive Bayes score identifies, for a given modification to a parameter, which other interface parameter should also be modified. In some configurations evaluating each of the plurality of possible designs comprises determining a probability that each of the plurality of possible designs will meet the at least one design criterion. The at least one design criterion may be generally defined in terms of size and or conformity of the biomechanical interface. Alternatively, or in addition, the at least one design criterion may include an indication that the design is suitable for a patient having at least one condition not incorporated in the initial set of measurements. For example, in some use cases the patient may require multiple custom medical devices, or have one or more other health conditions, such that specific customisation of the biomechanical interface is required. These additional constraints may not, as standard, be incorporated in the one or more measurements that comprise the input data. The user may therefore set, as the at least one design criterion, that the biomechanical interface should be suitable for a patient having the additional constraints. The suitability of the possible design may be evaluated based on comparison with metadata associated with initial designs, i.e., the evaluation may involve determining how previous users accounted for the additional constraints. Alternatively, the evaluation may be entered as a physical criterion, e.g., the biomechanical interface should avoid conforming to the external body segment in a particular region. In some configurations the user feedback comprises an indication of at least one of the plurality of interface parameters that is to remain fixed when generating the modified design. Whilst in some use cases the user may want the computational model to have a greater degree of input, for example, in the early stages of refining an initial design. However, in some configurations, the user may not want to allow the computational model to alter some of the interface parameters. The user may therefore provide an indication that particular parameters are to remain fixed when providing the user feedback. The modification may be any modification that affects the biomechanical properties of the biomechanical interface. In some configurations the at least one modification comprises one or more of: a modification to the at least one design criterion; a modification to one of the plurality of interface parameters, and a displacement or deletion of one or more portions of the biomechanical interface identified in the initial design. In general, a change to the design criterion or the addition of a further design criterion may require a regeneration of the one or more interface parameters. This, in turn, may result in a change in the design. It is noted that, in some use cases, a modification of the design criterion may not result in any modifications to the design. For example, if an additional design criterion were added, it may be the case that the initial design is already the design most likely to be appropriate once the additional design criterion has been added. In other words, in some use cases the modified design may be identical to the initial design but with the added assurance that the additional design criterion is satisfied. The modification may alternatively, or in addition, comprise a change to one of the one or more parameters. The modified design may therefore incorporate this change along with changes to one or more other parameters that are determined, by the computational model, to be appropriate based on the previous designs which make up the computational model. In addition, or as an alternative, the modification may include a displacement or a deletion of one or more portions of the biomechanical interface. For example, the user may choose to modify the initial design by physically displacing a portion of the initial design or removing the portion. Such freeform modifications may be equivalent to changing one or more of the plurality of interface parameters. In some configurations, the freeform modifications may be incorporated by making one or more modifications to the vector of measurements. Alternatively, the computational model may be responsive to such freeform changes by determining for the plurality of input designs, the probability that those designs would be appropriate given the freeform modification and re-generating the interface parameters based on designs that are most likely to be appropriate. Alternatively, or in addition, the computational model may determine a combination of modifications to interface parameters that, if implemented, would result in the freeform modification and base the modified design on those freeform modifications. In some configurations, freeform modifications may be implemented as a final subsequent step to the modifications to the interface parameters. Alternatively, or in addition, a record of freeform modifications may be stored in the database for subsequent inclusion in future instances of the method. For example, if a user commonly makes a same freeform modification then multiple designs comprising those modifications may be incorporated in the database. These can then be used to update the one or more relationships using, for example, statistical shape modelling techniques. In some configurations the method comprises providing an alert in response to at least one of: a determination that the modified design violates one or more physical constraints; and a determination that one or more of the interface parameters falls outside of a predefined range of expected interface parameters. For example, an alert may be provided to the user if the modification will result in a biomechanical interface that may not be structurally stable, or that could result in an incompatibility between the biomechanical interface and the external body segment. The alert may also be provided in the event that the user chooses to modify the initial design in such a way that the interface parameters that make up the modified design are identified as falling outside of the expected range. The expected range may be derived, for example, through a statistical analysis of all previous designs to determine an expected range for each single interface parameter and / or expected ranges for combinations of parameters. The alert may be provided when one or more single interface parameters fall outside of the expected range and / or when the combination (or a sub-combination) of the interface parameters fall outside of an expected range for that combination (or sub-combination) of parameters. The provision of such an alert allows the user to identify when they are working in an expected range of interface parameters and when they have, either intentionally or unintentionally, entered into a previously unused region of the interface parameter space. In addition to the provision of the one or more measurements, in some configurations the input data comprises patient metadata indicative of one or more patient specific requirements of the custom medical device. The metadata may be compared against corresponding metadata stored in the database in combination with each previous design. As an example, the metadata may indicate a potential use of the custom medical device, e.g., where the custom medical device is a prosthetic limb, the metadata may indicate that the prosthetic limb is for use in a particular sport. As an alternative example, the metadata may indicate a health condition of the patient that could affect the design, e.g., that the patient has a health condition, beyond the need for the custom medical device, that will affect the way in which they interact with that device. The provision of metadata can therefore improve the likelihood that the initial design will not require significant modification and can improve the likelihood that the modified design will be appropriate without further modification. Whilst in some configurations a single modification may be provided by the user resulting in a modified design that is appropriate for the patient, in some configurations the method comprises iteratively generating a plurality of modified designs, each comprising a corresponding plurality of interface parameters, wherein generating each of the plurality of modified designs comprises: receiving corresponding user feedback information identifying a corresponding modification to the preceding one of the plurality of modified designs; identifying at least one corresponding relationship between the corresponding modification and the corresponding plurality of interface parameters; and re-generating at least one of the corresponding plurality of interface parameters based on the at least one design criterion and the at least one corresponding relationship. In other words, the computational model is responsive to receipt of iterative feedback to refine the modified design through repeated re-generation of at least some of the plurality of interface parameters. At each stage the user feedback is received as an input to the computational model that generated the initial design. This therefore ensures that each of the plurality of modified designs is generated taking into account the relationships built into the computational model resulting in a more efficient design process for the user that accounts for prior knowledge of biomechanical interfaces developed for the particular custom medical device. In some configurations the one or more measurements comprise at least one of: a measurement of a shape of the external body segment; a measurement of a size of at least one part of the external body segment; and a computational mesh conforming to a surface of at least a portion of the external body segment. The measurements may be provided as numerical inputs, e.g., in terms of lengths, angles, and or distances measured by a user in advance, or may be derived from one or more images taken of the external body segment. The computational mesh may be derived from the one or more images. Alternatively, or in addition, the computational mesh may be derived by scanning the external body segment to determine both a shape and size of the external body segment. Whilst the external body segment may be any external body segment, in some configurations the external body segment is a residual limb. The residual limb may, for example, be a residual limb remaining subsequent to an amputation, e.g., a lower limb amputation. As discussed, the custom medical device may be any custom medical device. However, in some configurations the custom medical device is one of: a prosthetic device; an orthotic device; and an external fracture stabilisation device. In some configurations the prosthetic device may be one of a transtibial, a transradial, a transhumeral, or transfemoral prosthetic limb, an ankle disarticulation prosthetic limb, or knee disarticulation prosthetic limb. In such configurations the biomechanical interface may comprise a socket configured to interface with the external body segment which, as discussed, may be a residual limb. In some configurations the orthotic device may be one of a spinal brace, an ankle-foot-orthosis (AFO), a knee-ankle-foot-orthosis (KAFO), a cranial helmet, insoles, or other custom diabetic footwear. In some configurations the external fracture stabilisation device may be an external fracture stabilisation splint. In some configurations the method comprises outputting control data for controlling one or more fabrication devices to perform at least one fabrication step to fabricate the biomechanical interface. The control data may be output in any form suitable for feeding onto a fabrication device (e.g., a computer controlled manufacturing device) used in fabrication (manufacturing) of the biomechanical interface. In some configurations the output data is in a form other than a form used for visualising the biomechanical interface, e.g., the control data is specifically tailored for the at least one fabrication step. For example, the control data may be provided as a control data file that is in a form suitable to be interpreted by a fabrication device that produces the biomechanical interface itself and / or that is in a form suitable to be interpreted by a device that fabricates a mould from which the biomechanical interface is fabricated. In some configurations multiple sets of control data may be output, each suitable for controlling a fabrication device during a different stage of the fabrication process. In some configurations the control data may take the form of a surface mesh file, e.g., an stl file or other mesh file. For example, the surface mesh file may be generated directly from the one or more interface parameters. The control data may alternatively, or in addition comprise Computer Numerical Control (CNC) cutting paths for caving at least one portion of the biomechanical interface, or at least one portion of a mould for forming the biomechanical interface. In some configurations the at least one fabrication step is a step of manufacturing the biomechanical interface. In some configurations the at least one fabrication step is a step of manufacturing a mould for the biomechanical interface. The biomechanical interface may be manufactured from the mould through lamination of a material over the mould. Alternatively, a sheet of plastic may be heated and draped over the mould before air is sucked from the mould (i.e., vacuum forming). In some configurations the at least one fabrication step comprises a three dimensional printing step. Whilst in some configurations the control data may be provided in a form suitable for storing, for subsequent fabrication of the biomechanical interface, in some configurations the method comprises transmitting the control data to at least one of the one or more fabrication devices. Transmitting the control data may comprise transmitting the control data to a local fabrication device to trigger fabrication. Alternatively, or in addition, transmitting the control data may comprise transmitting the control data over a local area network, or a wide area network, e.g., the internet. In some configurations the method comprises triggering at least one of the one or more fabrication devices to perform the at least one fabrication step. In some configurations the method comprises manufacturing at least one part of the biomechanical interface. In some configurations the method comprises manufacturing the biomechanical interface. In some configurations the method comprises manufacturing the custom medical device including manufacturing the biomechanical interface. In some configurations the method comprises displaying the plurality of interface parameters on a graphical user interface; providing, for at least one parameter of the plurality of interface parameters, an input interface allowing the user to modify, as the at least one modification, the at least one parameter. The graphical user interface therefore may provide an alternative shortcut to allow the user to directly set one or more parameters of the at least one parameter. The graphical user interface may also allow the user to fix (e.g., to lock) each parameter. A fixed or locked parameter will be left unchanged when the modified design is generated. The provision of such shortcuts and the ability to fix or lock a parameter allows a user to quickly and efficiently explore a parameter space of the design and provides an improved way for the user to interact with the computational model in order to generate the modified design. According to some configurations of the present technique there is provided a computer readable storage medium configured to store program code comprising instructions which, when executed by a processor, cause the processor to perform the computer implemented method of any preceding claim. In some configurations the computer readable storage medium is a non-transitory computer readable storage medium. The instructions may be provided in any machine readable format. According to some configurations of the present technique there is provided an apparatus for designing a custom medical device comprising a biomechanical interface for interaction with an external body segment of a patient, the apparatus comprising: receiving circuitry configured to receive input data comprising one or more measurements of the external body segment; computation circuitry configured to generate an initial design of the biomechanical interface based on the input data, the initial design defined by a plurality of interface parameters, wherein the computation circuitry is configured to select the plurality of interface parameters based on at least one design criterion evaluated at the biomechanical interface, wherein the computation circuitry is configured: to receive user feedback information indicating at least one modification to be applied to the initial design; in response to receipt of user feedback information identifying at least one relationship between the at least one modification and the plurality of interface parameters, to generate a modified design of the biomechanical interface by re-generating at least one of the plurality of interface parameters based on the at least one design criterion and the at least one relationship. The apparatus is provided with circuitry configured to implement the method described in the above claims and may comprise specific circuitry configured to implement each of the above-described configurations. The circuitry may be provided as specific circuitry configured specifically to perform the described function or may be provided as one or more blocks of circuitry each configured to contribute to the described function. The circuitry may be optimised to perform the specified function or may be provided using one or more custom computer parts. In some configurations the apparatus comprises fabrication hardware configured to support fabrication of at least one portion of the biomechanical interface and arranged to be triggered to perform one or more stage of fabrication of the biomechanical interface in response to an indication that a user has accepted a modified design. The apparatus may, in some configurations, take the form of a general purpose computer. Some configurations of the present techniques will now be described with reference to the accompanying figures. Figure 1 schematically illustrates a computer implemented method 10 according to some configurations of the present techniques. The computer implemented method 10 combines a computational model 12 with user interface 14 to generate a modified design of a custom medical device comprising biomechanical interface that is suitable for interaction with an external body segment of a patient. In particular, the computational model receives input data which comprises one or more measurements of the external body segment. On receipt of the input data, the computational model 12 generates an initial design for the custom medical device based on the input data and a at least one design criterion. The initial design is defined by a plurality of interface parameters. The initial design is provided from the computational model 12 to the user interface 14 to receive user input. A user of the computer implemented method specifies user input including feedback information specifying at least one modification to the initial design using the user interface 14. The computational model 12 receives the user feedback from the user interface 14 and generates a modified design based on at least one identified relationship between the at least one modification and the plurality of interface parameters. Generation of the modified design comprises re-generating at least one of the plurality of interface parameters based on the input data, and the user feedback supplied in response to the initial design. Once the computational model 12 has generated the modified design, the modified design is output. Figure 2 schematically illustrates an apparatus 20 according to some configurations of the present techniques. The apparatus 20 is provided with receiving circuitry 22 configured to receive input data comprising one or more measurements of an external body segment. The receiving circuitry 22 passes the one or more measurements to the computation circuitry 24 which generates an initial design of the biomechanical interface from the input data and according to at least one design criterion. The computation circuitry 24 is configured to receive user feedback information indicating at least one modification to be applied to the initial design 26. The computation circuitry 24 is responsive to receipt of the user feedback to identify at least one relationship between the at least one modification and the plurality of interface parameters and to generate a modified design of the biomechanical interface by regenerating at least one of the plurality of interface parameters based on the at least one design criterion and the at least one relationship. The computation circuitry 24 is configured to output the modified design. Whilst the receiving circuitry 22 and the computation circuitry 24 have been illustrated as individual blocks of circuitry, it will be readily apparent to the person skilled in the art that, in alternative configurations, a single block of circuitry could be provided that performs the function of the receiving circuitry 22 and the computation circuitry 24. Figure 3 schematically illustrates steps of the computer implemented method 30 according to some configurations of the present techniques. In particular, the computer implemented method 30 receives input data and outputs data suitable for one or more fabrication steps. In the illustrated configuration the external body segment of the patient is a limb of the patient. Generation of the input data comprises performing a limb scan 32. The limb scan 32 may comprise any scanning technique to identify a three dimensional shape of the limb or to identify one or more two dimensional images of the limb. For example, the scanning may comprise laser scanning or structured white light scanning to capture a physical form or shape of the limb, a Magnetic Resonance Imaging (MRI) scan of the limb, photogrammertry techniques to determine a 3D shape of the limb from a sequence of 2D photographs, a Computer Tomography (CT) scan of the limb, and / or CAD / CAM by measures to determine a limb shape. The limb scan 32 may comprise generation of a three dimensional mesh indicative of a surface shape of the limb. Once the limb scan 32 is complete, the generation of the input data comprises extracting one or more measurements 34 from the data generated in the limb scan. The extraction of measurements may comprise steps of identifying one or more features in the limb and fitting one or more shapes to the limb in order to generate an artificial representation of that limb having the same basic size and shape properties as the limb that has been scanned. Once the step of extracting the measurements 34 has been completed, the extracted measurements are combined with metadata 36 associated with the patient. The metadata may include, for example, patient characteristics, Statistical Shape Model (SSM) inputs, clinician preferences, geography, and activity level. SSM inputs may include a mean shape along with principal components derived from a principal component analysis. Where the custom medical device being designed is a prosthetic device following amputation, the metadata may also include, for example, time since amputation, cause of amputation, areas of tissue sensitivity, pressure sensors, and pressure predictions. The metadata and the measurements are then categorised 38 before being provided to the computational model 30 as input data comprising the extracted measurements and the metadata. The input data is passed to a predictive model 40 which generates a design 42 of the biomechanical interface based on the input data and at least one design criterion. The design 42 is presented to the user, e.g., a clinician, who has the option to provide feedback information indicating one or more modifications, e.g., one or more rectifications, to be applied to the design. For each design a determination 44 is made as to whether the user accepts the design or whether the user is to provide at least one modification 46. When the user provides at least one modification 46, the at least one modification is provided as an input back into the predictive model 40 which generates a new design based on the modification, the input data and the at least one design criterion. This process may repeat iteratively until the user accepts the determination as to whether the user accepts the design 44 indicates that the user has accepted the design. The design is then output from the model 30 and one or more CAD steps 48 may be applied before the customised medical device is manufactured 50. The model 30 outputs the design for the biomechanical interface of the custom medical device in the form of control data that is suitable for being read by a fabrication device that causes at least one manufacturing step of the biomechanical interface to be performed. Figure 4 schematically illustrates a further example of the computational method 64 according to some configurations of the present techniques. The computational method 64 comprises a predictive model 72 and a user interface 76. The computational model 64 receives shape and size measurements from an initial statistical shape model 62 which is based on a limb scan of a patient 60. The computational model 64 also receives metadata identifying one or more details relating to the patient as described above. The computational model 64 receives this input data and provides it to the predictive model 72. The predictive model generates an initial model state (state 0) 68 which defines an initial design for the biomechanical interface. The initial design is provided from the predictive model 72 to the user interface 76 and may be presented to the user as a CAD model 78. The CAD model 78 is able to receive an indication of modifications from a user 74. The modifications including changes to parameters and an indication of locked and unlocked parameters are provided back to the predictive model 72. The predictive model 72 is responsive to receipt of the modifications including the changes to parameters and any changes to which parameters are locked and unlocked and generates a new model state (state N+l) 70 based on the previous model state (state N). The updated parameters are passed back from the predictive model 72 to the user interface 76 where the updated design is presented to the user 74. The user is then able to apply one or more further modifications or accept the design which is output to a manufacturing interface 66. In the illustrated configuration the output design is output as a mesh file suitable for triggering the manufacturing interface 66 to perform one or more manufacturing stages of the design. It will be readily apparent to the person skilled in the art that whilst model state N+l could be derived from state N along with any modifications to the interface parameters at state N, in some configurations model state N+l may be derived from state 0 along with a current set of interface parameters including any modifications made in response to model state N. Generating state N+l may involve setting locked parameters to the specific value at which they are currently set and allowing any unlocked parameters to vary according to the one or more relationships. Figures 5a and 5b schematically illustrate further details of the computational method. Figure 5a schematically illustrates details of the generation of an initial design 84. The method uses a statistical model 82 and a database 80. The database 80 stores a plurality of previous designs 81. Each of the previous designs 81 includes a record of the input data, the metadata, an initial design and a final design. The statistical model 82 is responsive to receipt of input data 88 comprising a plurality of measurements and metadata 86 as described above, to determine which of the previous designs 81 include input data and metadata that are closest to the input data 88 and the metadata 86 that have been received by the statistical model. As discussed, this may be performed using one or more statistical models designed to determine the probability that, given the differences between the input data 88 and the input data stored in each of the previous designs 81, and the differences between the metadata 86 and the metadata stored in each of the previous designs 81, what is the probability of that one of the previous designs being a suitable candidate for the initial design 84. The statistical model 82 identifies plural possible designs 90 each having a probability 92 that the corresponding possible design will be a suitable candidate for the initial design 84. In the illustrated configuration, the first possible design 90(A) has a probability 92(A) that it is a suitable candidate; the second possible design 90(B) has a probability 92(B) that it is a suitable candidate; the third possible design 90(C) has a probability 92(C) that it is a suitable candidate; and the fourth possible design 90(D) has a probability 92(D) that it is a suitable candidate. The statistical model 82 evaluates each of the possible designs 90 having a probability 92 that is within a given proximity of the possible design 90 having a highest probability 92 against the design criterion and selects an initial design 84 to be output. For example, if the possible design having the highest probability has a probability P that is a suitable design, then the statistical model may evaluate all designs having a probability falling in the range P-5P to P where 5P is a predetermined probability range. The statistical model 82 then outputs the initial design 84. Figure 5b schematically illustrates how the computer implemented method responds to receipt of at least one modification. The statistical model outputs a suggested design 85. The suggested design may be the initial design 84 or it may be a modified initial design that has been modified in response to a previous modification suggested by the user. The statistical model 82 is responsive to receipt of the at least one modification from the user, e.g., a change in one of the plurality of interface parameters and or an indication that one or more of the plurality of interface parameters is to be locked (fixed to its current value). The statistical model then retrieves a plurality of relationships 96 between modifications In the illustrated configuration, the statistical model has retrieved relationship 96(A), relationship 96(B), and relationship 96(C). These relationships may indicate how one or more other parameters of the plurality of interface parameters should be modified in response to the suggested at least one modification. The relationships may be derived through a statistical analysis of the previous designs 81 stored in the database 80 and may indicate correlations between different groups of the one or more interface parameters and the metadata. For example, the relationships may determine that for patients having specific metadata, there is a given relationship, e.g., a linear or non-linear relationship between a subset, or all, of the plurality of interface parameters. Alternatively, the relationships may determine that, there are distinct groups of potential designs, e.g., multiple styles of design, that may be appropriate for given input data 88 and given metadata 86. The modification may cause the particular style of design to change significantly resulting in the new suggested design 85 being output based on a different one of the previous designs 81. Once the suggested design 85 that is output may be modified further by the user or may be accepted by the user and may become a final design 94. The final design 94 may be output as control data suitable for fabrication and may (optionally) be fed back to the database 80 be added to the group of previous designs 81. Figure 6 schematically illustrates a user interface 100 that may be provided according to some configurations of the present techniques. The user interface 100 provides an indication of a current design of the biomechanical interface 102 having a patellar tendon modification 103, a fibular head modification 105, and a lateral paratibial modification 107. In the illustrated configuration, the current design of the biomechanical interface 102 is a socket for a prosthetic limb. The user interface also provides a series of sliders 104 to allow the user to modify each of the plurality of interface parameters. Each slider 104 is provided with a corresponding lock button 106 to prevent the interface parameter corresponding to that lock button 106 from being modified when the modified design is generated. Each slider 104 is also provided with a corresponding value indicator to indicate a current modification to that interface parameter. In the illustrated configuration a volume modification slider 104(A) is provided with a corresponding lock button 106(A), and a corresponding current value indication 108(A); a length modification slider 104(B) is provided with a corresponding lock button 106(A), and a corresponding current value indication 108(A); a patellar tendon modification slider 104(B) is provided with a corresponding lock button 106(A), and a corresponding current value indication 108(A); a fibular head modification slider 104(B) is provided with a corresponding lock button 106(A), and a corresponding current value indication 108(A); and a medial paratibial modification slider 104(B) is provided with a corresponding lock button 106(A), and a corresponding current value indication 108(A). Each of these sliders indicates a modification to the corresponding parameter which, when a user makes the modification, is provided as feedback information to the computational model along with information identifying a current set of locked parameters. The volume modification slider 104(A) modifies an overall volume of the biomechanical interface 102. The length modification slider 104(B) modifies an overall length (e.g., a length in the vertical direction of the image) of the biomechanical interface 102. The patellar tendon modification slider 104(C) modifies a size of the patellar tendon modification 103. The fibular head modification slider 104(D) modifies a size of the fibular head modification 105. The lateral paratibial modification slider modifies a size of the lateral paratibial modification 107. The biomechanical interface 102 is updated once the model has generated a modified design and the corresponding modifications to the unlocked parameters other than the modified one are indicated to the user. The user may also enter a value for the modification directly by editing the value indication 108. It will be readily apparent to the person skilled in the art that the modifications may be represented as percentage modifications, dimensional modifications, for example having units of length, area or volume, or in terms of dimensionless units scaled to a typical unit of length, area or volume. Furthermore, the parameters indicated in figure 6 are but a subset of possible interface parameters that could be provided in designing a prosthetic limb having a socket as a biomechanical interface. Alternative parameters may be present in the case of different types of custom medical device as described above. In the illustrated configuration, a modified design is generated in response to any modification to any one of the interface parameters. It will be readily apparent to the person skilled in the art that, in alternative configurations, an update button may be provided and the steps of generating the modified design may occur when the update button is pressed. Figure 7 schematically illustrates a sequence of steps carried out according to some configurations of the present techniques. Flow begins at step S70 where input data is received. The input data comprises one or more measurements of an external body segment of the patient. The input data may also comprise metadata as described above. Flow then proceeds to step S72 where an initial design of the interface is generated based on the input data and at least one design criterion. Flow then proceeds to step S74 where user feedback is received indicating at least one modification to the initial design. Flow then proceeds to step S76 where at least one relationship between the modification and the plurality of interface parameters is identified. Flow then proceeds to step S78 where a modified design is generated based on the modification, the design criterion and the at least one relationship. Figure 8 schematically illustrates a sequence of steps carried out according to some configurations of the present techniques. Flow begins at step S80 where the computational method is waiting to receive input data. Flow then proceeds to step S82 where measurements of the body segment are received as part of the input data. The input data may also comprise metadata. Flow then proceeds to step S84 where the received measurements are compared against previous designs comprised in a database. Flow then proceeds to step S86 where previous designs are identified that had, as their input data, measurements that are close to the measurements received in step S82. Flow then proceeds to step S88 where, for each of the input designs, a probability is calculated that the previous design meets the design criteria. Flow then proceeds to step S90 where the suggested design having the highest probability is suggested as an initial design. Flow then proceeds to step S92 where it is determined if user feedback suggesting a modification has been received. If, at step S92, it is determined that user feedback suggesting a modification has been received, then flow proceeds to step S94 where at least one relationship between the suggested modification and one or more further parameters is identified. Flow then proceeds to step S96 where it is determined if user feedback indicates that any parameters are fixed. If, at step S96, it is determined that the user feedback identifies that there is at least one parameter that is fixed, then flow proceeds to step SI08 where the value of the parameters indicated by the user as fixed are held at their current value. Flow then proceeds to step S98. If, at step S96, it was determined that none of the values are fixed, then flow proceeds to step S98 where one or more modifications are applied to the non-fixed parameters according to the identified relationships to generate a modified design. Flow then returns to step S92. If, at step S92, it was identified that no user feedback has been provided, then flow proceeds to step SI 00 where it is determined if the user accepts the current design. If, at step SI00, it is identified that the user does not accept the design, then flow returns to step S92. If, at step SI00, it is determined that the user accepts the design, then flow proceeds to step SI02 where control data is output for controlling one or more fabrication devices. Flow then proceeds to step SI04 where the database is updated to comprise the new entry including modifications. Flow then ends at step SI06. In brief overall summary there is provided a computer implemented method of designing a custom medical device comprising a biomechanical interface for interaction with an external body segment of a patient. The method comprises receiving input data comprising one or more measurements of the external body segment. The method further comprises generating an initial design of the biomechanical interface based on the input data, the initial design defined by a plurality of interface parameters selected based on at least one design criterion. The method further comprises, in response to receipt of user feedback information indicating at least one modification, generating a modified design of the biomechanical interface. Generating the modified design comprises identifying at least one relationship between the at least one modification and the plurality of interface parameters, and re-generating at least one of the plurality of interface parameters based on the at least one design criterion and the at least one relationship. In the present application, the words “configured to..are used to mean that an element of an apparatus has a configuration able to carry out the defined operation. In this context, a “configuration” means an arrangement or manner of interconnection of hardware or software. For example, the apparatus may have dedicated hardware which provides the defined operation, or a processor or other processing device may be programmed to perform the function. “Configured to” does not imply that the apparatus element needs to be changed in any way in order to provide the defined operation. In the present application, lists of features preceded with the phrase “at least one of’ mean that any one or more of those features can be provided either individually or in combination. For example, “at least one of: [A], [B] and [C]” encompasses any of the following options: A alone (without B or C), B alone (without A or C), C alone (without A or B), A and B in combination (without C), A and C in combination (without B), B and C in combination (without A), or A, B and C in combination. Although illustrative configurations of the invention have been described in detail herein with reference to the accompanying drawings, it is to be understood that the invention is not limited to those precise configurations, and that various changes, additions and modifications can be effected therein by one skilled in the art without departing from the scope of the invention as defined by the appended claims. For example, various combinations of the features of the dependent claims could be made with the features of the independent claims without departing from the scope of the present invention.

Claims

1. A computer implemented method of designing a custom medical device comprising a biomechanical interface for interaction with an external body segment of a patient, the method comprising:receiving input data comprising one or more measurements of the external body segment;generating, using a computational model, an initial design of the biomechanical interface based on the input data, the initial design defined by a plurality of interface parameters selected based on at least one design criterion; andin response to receipt, by the computational model, of user feedback information indicating at least one modification to be applied to the initial design, generating, using the computational model, a modified design of the biomechanical interface, wherein generating the modified design comprises identifying at least one relationship between the at least one modification and the plurality of interface parameters, and re-generating at least one of the plurality of interface parameters based on the at least one design criterion and the at least one relationship.

2. The computer implemented method of claim 1, wherein the at least one relationship is derived using one or more statistical models.

3. The computer implemented method of claim 1 or claim 2, wherein the at least one relationship is derived from a database comprising a plurality of entries, each of the plurality of entries identifying a corresponding previous design.

4. The computer implemented method of claim 3, wherein for each of the plurality of entries, the corresponding previous design comprises at least one of:a corresponding initial design; anda corresponding modified design identifying one or more modifications applied when generating the modified design.

5. The computer implemented method of claim 3 or claim 4, wherein each entry of the database comprises previous input data received for generation of that one of the previous designs,wherein the computational model is configured to generate the initial design through comparison of the input data against the previous input data for the plurality of entries.

6. The computer implemented method of claim 5, wherein the computational model is configured to generate the initial design through interpolation between the input data for the plurality of entries.

7. The computer implemented method of any of claims 3 to 6, comprising:in response to a determination that the updated design is accepted, updating the database to store a new entry; andupdating each of the one or more relationships based on the new entry.

8. The computer implemented method of any preceding claim, wherein the at least one design criterion comprises at least one of:a preferred size criterion specifying a preferred size difference between the external body segment and the biomechanical interface; anda predefined conformity criterion specifying a preferred degree of conformity between a shape of the biomechanical interface and a shape of the external body segment.

9. The computer implemented method of any preceding claim, wherein:the computational model is configured to generate a plurality of possible initial designs;selecting the initial design from amongst the plurality of possible designs comprises evaluating each of the plurality of possible initial designs against the at least one design criterion and selecting one of the plurality of possible designs having a highest evaluation.

10. The computer implemented method of claim 9, wherein evaluating each of the plurality of possible designs comprises determining a probability that each of the plurality of possible designs will meet the at least one design criterion.

11. The computer implemented method of any preceding claim, where the user feedback comprises an indication of at least one of the plurality of interface parameters that is to remain fixed when generating the modified design.

12. The computer implemented method of any preceding claim, where the at least one modification comprises one or more of:a modification to the at least one design criterion;a modification to one of the plurality of interface parameters; anda displacement or deletion of one or more portions of the biomechanical interface identified in the initial design.

13. The computer implemented method of any preceding claim, comprising providing an alert in response to at least one of:a determination that the modified design violates one or more physical constraints; anda determination that one or more of the interface parameters falls outside of a predefined range of expected interface parameters.

14. The computer implemented method of any preceding claim, wherein the input data comprises patient metadata indicative of one or more patient specific requirements of the custom medical device.

15. The computer implemented method of any preceding claim, comprising iteratively generating a plurality of modified designs, each comprising a corresponding plurality of interface parameters, wherein generating each of the plurality of modified designs comprises:receiving corresponding user feedback information identifying a corresponding modification to the preceding one of the plurality of modified designs;identifying at least one corresponding relationship between the corresponding modification and the corresponding plurality of interface parameters; andre-generating at least one of the corresponding plurality of interface parameters based on the at least one design criterion and the at least one corresponding relationship.

16. The computer implemented method of any preceding claim, wherein the one or more measurements comprise at least one of:a measurement of a shape of the external body segment;a measurement of a size of at least one part of the external body segment; anda computational mesh conforming to a surface of at least a portion of the external body segment.

17. The computer implemented method of any preceding claim, wherein the externalbody segment is a residual limb.

18. The computer implemented method of any preceding claim, wherein the custom medical device is one of:a prosthetic device;an orthotic device; andan external fracture stabilisation device.

19. The computer implemented method of any preceding claim, comprising outputting control data for controlling one or more fabrication devices to perform at least one fabrication step to fabricate the biomechanical interface.

20. The computer implemented method of claim 19, comprising transmitting the control data to at least one of the one or more fabrication devices.

21. The computer implemented method of claim 19 or claim 20, comprising triggering at least one of the one or more fabrication devices to perform the at least one fabrication step.

22. The computer implemented method of any preceding claim comprising: displaying the plurality of interface parameters on a graphical user interface; providing, for at least one parameter of the plurality of interface parameters, an input interface allowing the user to modify, as the at least one modification, the at least one parameter.

23. A computer readable storage medium configured to store program code comprising instructions which, when executed by a processor, cause the processor to perform the computer implemented method of any preceding claim.

24. The computer readable storage medium of claim 23, wherein the computer readable storage medium is a non-transitory computer readable storage medium.

25. An apparatus for designing a custom medical device comprising a biomechanical interface for interaction with an external body segment of a patient, the apparatus comprising:receiving circuitry configured to receive input data comprising one or more measurements of the external body segment;computation circuitry configured to generate an initial design of the biomechanical interface based on the input data, the initial design defined by a plurality of interface parameters, wherein the computation circuitry is configured to select the plurality of interface parameters based on at least one design criterion evaluated at the biomechanical interface,wherein the computation circuitry is configured:to receive user feedback information indicating at least one modification to be applied to the initial design;in response to receipt of user feedback information identifying at least one relationship between the at least one modification and the plurality of interface parameters, to generate a modified design of the biomechanical interface by regenerating at least one of the plurality of interface parameters based on the at least one 5 design criterion and the at least one relationship.

Images