Method and system for predicting the condition of a person's spine

Abstract

The invention relates to a method for predicting the condition of a person's spine, the method comprising: - making a first estimate (E1) of personalised information about the spine on the basis of first input data (d1); - making a second estimate (E2) of mechanical loads that may be applied to the spine on the basis of second input data (d2); - performing a calculation (E3) to predict the condition of the spine on the basis of the first estimate (E1) and the second estimate (E2).

Description

DESCRIPTION Title of the invention: Method and system for predicting the condition of an individual's spine

[0001] Technical field

[0002] The present invention relates to a method for predicting the condition of an individual's spine. The present invention also relates to a system for predicting the condition of an individual's spine. The invention further relates to a computer program implementing such a method. The invention finally relates to a recording medium on which such a program is recorded.

[0003] Previous Art

[0004] The spine plays a mechanical role in maintaining the human body and a neurological role in protecting the spinal cord within the vertebral canal. A healthy spine painlessly supports the body's weight, ensures good mobility, and a balanced posture. With increasing life expectancy, maintaining spinal health is increasingly important to ensure a good quality of life for individuals. The condition of the spine is a key parameter in an individual's overall health.

[0005] Solutions exist to detect the presence of pathology during an imaging examination of an individual's spine, for example, an X-ray examination. The prescribing physician and / or radiologist use the data from the examination to propose a diagnosis of the condition of the patient's spine at the time of the examination.

[0006] Furthermore, solutions exist to refine a surgeon's surgical approach during spinal surgery. However, the long-term effectiveness of treating a pathology remains uncertain, and the relapse rate approaches 10%. Indeed, the initial surgery may fail and / or negatively affect neighboring areas of the spine, which could in turn become pathological and painful for the patient.

[0007] Existing solutions do not allow for the estimation of changes in the health of the spine over time, particularly for patients at risk due to their lifestyle or previous surgical treatment.

[0008] The invention aims to overcome at least in part the aforementioned drawbacks of the prior art. More particularly, the invention aims to provide a method and a system for predicting the state of a spine of an individual which is simple, efficient, reliable and accurate.

[0009] Summary of the invention

[0010] An object of the invention is therefore a method for predicting the state of an individual's spine, said method comprising: - a first estimation of personalized information of the spine from first input data; - a second estimate of mechanical loads likely to be applied to the spine from second input data; - a calculation for a prediction of the state of the spine from the first estimate and the second estimate.

[0011] “Personalized information” includes information that is dependent on or specific to each individual.

[0012] Personalized information may be biochemical and / or structural information of an individual's spine.

[0013] The first input data can be obtained by imaging techniques and possibly image processing.

[0014] The second input data can be derived from data concerning an individual's lifestyle.

[0015] One advantage of such a method is that it is based on a multi-scale, multi-physical model of the spine, which allows for the description of physical, biological, and chemical phenomena at the heart of matter. Such a method thus allows for accurate prediction of the progression of degeneration of intervertebral discs and vertebrae, as well as changes in the shape of the spine. Such a method can improve the prevention and treatment of spinal problems, opening up new perspectives in the field of spinal health.

[0016] One advantage of such a method is that it allows predicting the condition of the spine in the future in order to inform an individual about risks concerning their spine before they develop a pathology.

[0017] An advantage of such a method is that predictions can be obtained based on different scenarios. Such a method thus makes it possible to estimate the effects of different treatments, including drug and / or surgical treatments and / or those based on lifestyle changes.

[0018] In an alternative embodiment, the first input data is determined from imaging data obtained by magnetic resonance imaging.

[0019] In an alternative embodiment, the first input data are determined from a spin echo signal of the magnetic resonance imaging according to the equation: k: a proportionality constant depending on the sensitivity of the receiver; p: a proton density (or DP); T R : a rehearsal time; T E : an echo time; i: a longitudinal relaxation rate; T2: a transverse relaxation rate.

[0020] In an alternative embodiment, the longitudinal relaxation rate is based on an identification of the spin echo signal as a function of the repetition time of a plurality of pixels of a first type of images obtained by magnetic resonance imaging and in which the transverse relaxation rate is based on an identification of the spin echo signal as a function of the echo time of a plurality of pixels of a second type of images obtained by magnetic resonance imaging.

[0021] In an alternative embodiment, the spinal column comprising at least one intervertebral disc comprising a nucleus pulposus and a fibrous ring, said method comprises a step of determining a local grade of said nucleus pulposus and a step of determining a local grade of said fibrous ring, said determination steps being carried out by an artificial intelligence module from the longitudinal relaxation rate and / or the transverse relaxation rate.

[0022] In an alternative embodiment, the method comprises a step of determining an overall grade of the intervertebral disc from the equation: in which - Grade NP corresponds to an overall grade in the nucleus pulposus (NP), such that - Grade AF corresponds to an overall grade in the annulus fibrosus (AF), such as - f NPcorresponds to a volume fraction of the nucleus pulposus (NP), such as a volume fraction of annulus fibrosus (AF), such as - Volume NP corresponds to a volume of the nucleus pulposus (NP) - Volume AF corresponds to a volume of the annulus fibrosus (AF) - Volume disque corresponds to a volume of the intervertebral disc, said volume Volume NP of the nucleus pulposus, said volume Volume AP of the annulus fibrosus, said volume Volume disque of the intervertebral disc being determined from imaging data obtained by magnetic resonance imaging (MRI).

[0023] In an alternative embodiment, the artificial intelligence module is capable of using a support vector regression approach.

[0024] In an alternative embodiment, the second input data may comprise information about the individual's lifestyle, for example their occupation and / or their physical habits and / or their sleep and / or their diet.

[0025] In an alternative embodiment, the method comprises obtaining a digital twin of the spine.

[0026] In an alternative embodiment, the method further comprises providing at least one message, in particular: - at least one personalized recommendation; and / or - at least one guideline intended to help a practitioner choose at least one treatment, for example from the group including drug treatments, physical or bodily treatments, surgical treatments; and / or - at least one directive for personalized surgical planning purposes.

[0027] Another object of the invention also relates to a system for predicting the state of a spinal column of an individual, the system comprising a first storage unit capable of storing first input data, a second storage unit capable of storing second input data and at least one hardware processor capable of implementing: - a first estimation of personalized spine information from the first input data; - a second estimate of mechanical loads likely to be applied to the spine from the second input data; - a calculation for a prediction of the state of the spine from the first estimate and the second estimate.

[0028] Another object of the invention also relates to a computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the method defined previously.

[0029] Another object of the invention also relates to a data recording medium, readable by a computer, on which a computer program product defined previously is recorded.

[0030] Description of the figures

[0031] Figure 1 is a flowchart of one embodiment of a method for predicting the condition of a spine;

[0032] Figure 2 is a plan view schematically showing an intervertebral disc;

[0033] Figure 3 shows an embodiment of a spinal condition prediction system;

[0034] Figure 4 illustrates a mapping of a longitudinal relaxation rate over a region of interest in the spine;

[0035] Figure 5 illustrates a mapping of a transverse relaxation rate over a region of interest in the spine.

[0036] Detailed description of the invention

[0037] The invention proposes a method for predicting the state of an individual's spine in the future, in particular from imaging data of a spine or spine.

[0038] An embodiment of a method for predicting the condition of an individual's spine is described below with reference to Figure 1.

[0039] The method comprises a first estimation E1 (ESTIM1) of personalized information of the spine, in particular biochemical and / or structural, from first input data d1.

[0040] “Personalized information” includes information that is dependent on or specific to each individual.

[0041] The first input data d1 can be obtained by imaging techniques and possibly image processing.

[0042] The first input data d1 may include personal information related to the individual, for example the individual's age, gender, height, weight, ethnic origin.

[0043] An individual's age is a measure of the time elapsed since birth. An individual's sex is a biological classification of an individual as male or female. An individual's weight is a measure of an individual's body mass.

[0044] The first input data d1 may further include demographic data. Demographic data can be obtained from scientific literature.

[0045] The first input data d1 may comprise imaging data. The imaging data may be obtained by at least one medical imaging technique.

[0046] Imaging data can be obtained from magnetic resonance imaging (MRI). MRI uses radio waves and a strong magnetic field to create detailed images of the spine. MRI is useful for assessing the soft tissues of the spine such as intervertebral discs, ligaments, and spinal cord.

[0047] Imaging data can be obtained from a medical imaging technique other than an MRI.

[0048] Imaging data can be obtained from an X-ray. An X-ray can provide an overview of the bone structures of the spine and help detect possible fractures or deformities.

[0049] Imaging data can be obtained from a computed tomography (CT) scan. CT uses X-rays to create detailed cross-sectional images of the spine. This allows for precise information about bones, intervertebral discs and surrounding structures.

[0050] For all the variants described above for obtaining imaging data, the imaging data may undergo processing, in particular implemented via software. Such processing of the imaging data makes it possible in particular to estimate physical parameters, describing for example the state of at least one intervertebral disc.

[0051] An anatomical structure of an intervertebral disc 1 is shown in Figure 2.

[0052] The intervertebral disc 1 consists of a central part called the nucleus pulposus or nucleus pulposus NP surrounded by a ring called the annulus fibrosus or annulus fibrosus AF.

[0053] The annulus fibrosus AF of the intervertebral disc 1 comprises 3 lamellar layers, 5 oriented collagen fibers, and 7 interlamellar layers.

[0054] The annulus fibrosus AF of intervertebral disc 1 consists of a fluid portion and a solid portion. The fluid portion consists of water. The solid portion consists of extracellular matrix and collagen.

[0055] The intervertebral disc 1 comprises an Anterior-Internal region A1, an Anterior-External region A0, a Posterior-Internal region P1, and a Posterior-External region P0.

[0056] The volume fractions of the different constituents of the annulus fibrosus AF of the intervertebral disc 1 depend on the region Al, AO, PI, PO of the intervertebral disc 1.

[0057] Oriented collagen fibers 5, for example, have dimensions of the order of 250 micrometers.

[0058] The intervertebral disc 1 also includes elastic fibers of nanometric dimensions, for example of the order of 150 nanometers.

[0059] The first input data d1 may include intrinsic chemomechanical properties of the spine constituents.

[0060] By "chemo-mechanical properties" we mean in particular properties linked to chemical and mechanical phenomena.

[0061] The first input data d1 may include biochemical information of the spine constituents, including the densities of water, collagen, and extracellular matrix.

[0062] The first input data d1 may include intrinsic information of the intervertebral discs, including biochemical information, linked to the individual.

[0063] The first input data d1 may include fiber orientations and distributions of the collagen network of intervertebral discs.

[0064] The first input data d1 may include a definition of the multi-scale distribution, from the nanoscale, of spine collagen networks.

[0065] The imaging data may include at least one dimension of the intervertebral discs. The imaging data may include the thickness (height) of at least one intervertebral disc and / or the surface area of ​​at least one intervertebral disc and / or the position of at least one intervertebral disc.

[0066] By thickness or height of an intervertebral disc is meant a dimension of an intervertebral disc in a direction A parallel or substantially parallel to the main direction in which the spinal column extends.

[0067] Imaging data may include the amount of water in at least one intervertebral disc.

[0068] A mathematical model can be used to exploit the first input data d1 to obtain the first estimate E1.

[0069] Advantageously, in the first estimate E1, the imaging data can be coupled with personal information related to the individual and / or demographic data deemed relevant according to the scientific literature, to estimate the personalized information, in particular biochemical and / or structural, of the spine. This results in a first estimate having high precision.

[0070] The first estimate E1 may include an assessment of the condition of an individual's spine at an initial time, in particular from a result of a medical imaging technique of the spine.

[0071] Assessment of an individual's spinal condition may include determination of the grade of the intervertebral discs and / or the grade of the vertebrae of the spine at the initial time.

[0072] Intervertebral discs can be classified according to their condition and degeneration.

[0073] As soon as growth stops and under the effects of everyday life, the intervertebral discs are the site of degenerative phenomena, corresponding in particular to cracks in the periphery of the disc (fibrous ring AF) and dehydration of its center (nucleus pulposus NP).

[0074] An example of intervertebral disc grade classification is described below.

[0075] According to a first grade, called Grade I, the intervertebral disc is considered normal, with no significant signs of degeneration. The structure of the intervertebral disc is intact, with good hydration and normal height.

[0076] In a second grade, called Grade II, the intervertebral disc shows mild degeneration, often characterized by minor signs of hydration loss or slight structural alteration. The height of the intervertebral disc may be slightly reduced.

[0077] A third grade, called Grade III, indicates moderate degeneration of the intervertebral disc. The intervertebral disc shows more obvious signs of loss of hydration, a decrease in height, and more pronounced structural changes, such as cracks or deformations.

[0078] According to a fourth grade, called Grade IV, the degeneration of the intervertebral disc is severe. The intervertebral disc is significantly dehydrated, with a significantly reduced height. There may be significant structural deformities, herniated discs, or disc protrusions.

[0079] According to a fifth grade, called Grade V, degeneration of the intervertebral disc is at its most advanced stage. The intervertebral disc is completely dehydrated, with a total loss of height. Adjacent structures, such as the vertebrae and surrounding tissues, may be affected and show signs of deterioration.

[0080] It should be noted that this classification may vary slightly depending on the classification systems used by healthcare professionals, such as the Pfirrmann classification or the Modic classification. These classification systems aim to assess the condition of the intervertebral discs and provide useful information for the diagnosis and management of spinal problems.

[0081] The vertebrae of the spine can also be classified based on various criteria, such as their shape, position, and condition.

[0082] An example of vertebral grade classification is described below.

[0083] Vertebral bone quality (VBQ) classification is related to the assessment of vertebral bone mineral density. The VBQ system quantifies vertebral mineral density using a medical imaging technique. The VBQ system assigns vertebrae grades based on their mineral density, reflecting their health and fracture risk.

[0084] According to a first grade, called Grade 1, the bone mineral density of the vertebra is normal, indicating a healthy vertebra.

[0085] According to a second grade, called Grade 2, the bone mineral density of the vertebra is slightly reduced, but without signs of fracture.

[0086] According to a third grade, called Grade 3, the bone mineral density of the vertebra is moderately reduced, presenting an increased risk of fracture.

[0087] According to a fourth grade, called Grade 4, the bone mineral density of the vertebra is significantly reduced, indicating a high risk of fracture.

[0088] The method may include a prior step E01 (MODEL) of biomechanical modeling of the spine.

[0089] Biomechanical modeling of the spine may include a geometric representation of the spine.

[0090] The geometric representation of the spine is, for example, a three-dimensional representation of the series of intervertebral discs and vertebrae of the spine.

[0091] Geometric representation can take into account the shape of the intervertebral discs, the fact that they include several layers, the variable thickness (height) of the intervertebral discs.

[0092] The geometric representation can take into account the multi-scale distribution of intervertebral discs.

[0093] The geometric representation can take into account the regional variation in the morphology of the intervertebral disc, in particular according to the different regions described in relation to figure 2.

[0094] Biomechanical modeling of the spine may include discretizing the geometry of intervertebral discs and vertebrae into material points to represent these anatomical structures in a simplified manner.

[0095] A geometric representation of an intervertebral disc can comprise a set of material points that are distributed along the surface of the intervertebral disc. The general shape of an intervertebral disc can be considered as a circular surface of variable diameter relative to the center of the intervertebral disc. A number of points equidistant from the perimeter of the intervertebral disc can be determined. The material points geometrically represent, in particular, the approximate positions of the lamellae of the annulus fibrosus AF as well as the area that separates the annulus fibrosus AF from the nucleus pulposus NP.

[0096] Biomechanical modeling of the spine may include association of biochemical information of intervertebral disc constituents, including water and / or collagen and / or extracellular matrix, with material points of the intervertebral discs. Biomechanical modeling of the spine may take into account the multi-scale distribution, from the nanoscale, of collagen networks.

[0097] The number of material points in the biomechanical modeling of the spine can be chosen to optimize a compromise between precise discretization and reduced calculation time.

[0098] A geometric representation of a vertebra may include a set of material points that are distributed along the contact surfaces of the vertebra with the intervertebral discs. The contact surfaces of a vertebra with the intervertebral discs may be represented by a set of points regularly spaced, for example in the form of a grid of points. A discrete representation of the geometry of a vertebra can thus be obtained. A set of material points can be added along the contours of the vertebrae to represent the complex shape of the vertebrae.

[0099] Biomechanical modeling of the spine may include association of biochemical information to the material points of the vertebrae.

[0100] An advantage of such biomechanical modeling is that it is simple to implement because it allows taking into account the interactions between the three-dimensional geometric representations of the intervertebral discs and vertebrae comprising material points and the intrinsic bio-chemo-mechanical properties of the intervertebral discs and vertebrae.

[0101] By "intrinsic bio-chemo-mechanical properties" we mean in particular properties linked to biological, chemical and mechanical phenomena or aspects inherent to biological tissues.

[0102] Once the three-dimensional geometric representations of the intervertebral discs and vertebrae have been created, biomechanical simulations can be implemented using the material points to calculate the forces, deformations and movements of the intervertebral discs and vertebrae. The simulations are particularly intended to study the evolution of the microstructural behavior of the different components of the spine and the effects of this evolution on the health of the spine.

[0103] Such biomechanical modeling using discretization of three-dimensional geometric representations of intervertebral discs and vertebrae simplifies calculations while preserving the essential characteristics of anatomical geometry. This results in an improved understanding of mechanical behaviors and interactions within the spine.

[0104] The first estimate E1 may include a definition of the relationship between microstructure and intrinsic bio-chemo-mechanical properties for each material point including regional variation in the initial state.

[0105] The method may include obtaining a digital twin of an individual's spine, including using digital means to represent the spine. The digital twin is, for example, created from an MRI.

[0106] The first estimate E1 can be made using a mathematical model, in particular using equations integrated into the digital twin of the individual's spine.

[0107] The first estimate E1 may include substeps to obtain the first input data d1 .

[0108] The first estimation E1 may include a sub-step of collecting the first input data d1.

[0109] In the collection sub-step, first input data d1 can be collected via an analysis of at least one image, for example by replacing each material point.

[0110] In the collection sub-step, initial input data d1 can be collected, particularly systematically, from existing scientific literature. Initial input data d1 can be extracted from journal articles, experimental studies, and specialized databases.

[0111] Advantageously, automatic data extraction techniques can be used, for example by focusing on specific key terms and search criteria. This results in a simple sub-step of collecting initial input data.

[0112] The first estimate E1 may include a sub-step of normalization of the first input data d1 .

[0113] Once the first input data d1 is collected, the first input data d1 can be normalized. The normalization substep is intended to ensure the consistency, uniformity, and comparability of the first input data d1.

[0114] Advantageously, the normalization sub-step may include a definition of formats and structures of the first input data d1 .

[0115] Optionally, the first estimate E1 may comprise a sub-step of converting units of measurement of at least part of the first input data d1 when required.

[0116] Optionally, the first estimate E1 may include a sub-step using standardized references.

[0117] The first estimation E1 may include a sub-step of integrating the first input data d1 into a first centralized database.

[0118] The definition of formats and structures of the first input data d1 during the normalization sub-step described above is intended in particular to facilitate the integration of the first input data d1 into the first database during the integration sub-step.

[0119] The first database is for example a relational database or other storage system suitable for organizing the first input data d1 , in particular in such a way that it can be queried efficiently.

[0120] The first estimation E1 may include a sub-step of validation of the first input data d1 . The validation sub-step is intended to guarantee the quality and reliability of the first input data d1 .

[0121] The validation sub-step may include comparison with other reliable data sources and / or consultation with domain experts and / or the use of specific validation methods for each type of first input data d1 .

[0122] The first estimate E1 may include a sub-step of updating the first input data d1 .

[0123] Advantageously, the update sub-step can be carried out several times over time, particularly regularly.

[0124] The update sub-step is intended to ensure the quality and relevance of the first input data d1 over time.

[0125] The update sub-step may include monitoring of new scientific publications, research advances and additional discoveries. The most recent information can thus be taken into account to make the first E1 estimate.

[0126] Optionally, the first estimate E1 can be performed using an artificial intelligence AU. This results in a first estimate with increased accuracy.

[0127] The first estimate E1 can notably use a “random forest” type learning algorithm.

[0128] The first estimation E1 may include training an artificial intelligence model. The first input data d1 may be used to train at least one artificial intelligence model, such as neural networks. Artificial intelligence models may be developed to capture the complex relationships between the different variables corresponding to the first input data d1 .

[0129] The method comprises a second estimation E2 (ESTIM2) of mechanical loads likely to be applied to the spine from second input data d2.

[0130] Second input data d2 can be used to perform the second estimation E2, in particular to predict mechanical loads, in particular the most severe ones, likely to affect the spine, as well as movements, in particular the most repetitive ones, of the spine.

[0131] The second input data d2 can be derived from data concerning an individual's lifestyle.

[0132] The second input data d2 may include information about an individual's lifestyle, e.g., an individual's occupation, physical habits, diet, sleep, factors and / or stress level.

[0133] The second input data d2 may include information about an individual's occupation. Some occupations may expose the spine to specific stresses. For example, construction workers or movers are often subjected to heavy and repetitive lifting movements, which can put additional strain on the spine. People who work in offices may sit for long periods, which can lead to poor posture and back problems. By taking the occupation into account, it is possible to assess the mechanical loads and stresses imposed on the spine.

[0134] The second input data d2 may include information about an individual's physical habits. Physical activities, including sports and regular exercise, can have an impact on spinal health. By For example, regularly participating in high-impact sports like rugby or weightlifting can increase the risk of spinal injuries. On the other hand, muscle strengthening and flexibility exercises can help maintain a healthy spine.

[0135] The second input data d2 may include information about an individual's posture. Poor posture can put excessive pressure on the spine and lead to problems such as back pain or spinal misalignments.

[0136] The second input data d2 may include data relating to an individual's environment.

[0137] The second estimate E2 may comprise an estimate, from the second input data d2, of modes or types of mechanical loadings which will be applied in the future to the spinal column, for example mechanical loadings in compression and / or traction and / or torsion and / or flexion and / or extension.

[0138] The second estimate E2 may include a definition of the loading conditions, in particular monotonous and / or cyclic, for different mechanical loading modes.

[0139] The second estimate E2 may include a determination of cyclic numbers of mechanical loads applied to the vertebrae and intervertebral discs.

[0140] The second estimate E2 may comprise an estimate, from the second input data d2, of the amplitude of the mechanical loads which will be applied in the future to the spine, that is to say the level of the forces which will be applied in the future to the spine.

[0141] The second estimate E2 may include an estimate, from the second input data d2, of the frequencies of the mechanical loads which will be applied in the future to the spine.

[0142] A mathematical model can be used to exploit the second input data d2 to obtain the second estimate E2.

[0143] In the second estimation E2, an algorithm can use the second input data d2 to estimate the types of mechanical loads and force levels applying locally to each intervertebral disc.

[0144] A mathematical model can be used to estimate interactions within matter, including mechanical constraints, from the forces applied to the spine. Such a model can be defined to account for microstructural phenomena that cause the spine's behavior to change over time.

[0145] In the second estimation E2, an estimation of the biomechanical stress and damage state within the spine can be performed using the second input data d2.

[0146] The mathematical model makes it possible to describe the biomechanical fatigue behavior of intervertebral discs to predict the future state of the spine, in particular the grades of degeneration in the future of the intervertebral discs and vertebrae as well as their geometric evolutions thanks to a behavior law.

[0147] The second estimate E2 allows to predict the mechanical damage of the spine over time.

[0148] The second estimate E2 can use a mechanical protocol to mimic the mechanical loads applied to an individual's spine.

[0149] The second estimation E2 may include a sub-step of collecting the second input data d2.

[0150] Advantageously, the second input data d2 can be collected by qualified healthcare professionals, such as spine doctors or physiotherapists.

[0151] The second input data d2 can be collected using a questionnaire, especially a generic one, to gather information about an individual's lifestyle, including their occupation, physical habits, diet, sleep, stress factors, especially from the initial time to the time of the second estimate.

[0152] The second input data d2 can be obtained from a questionnaire. Such a questionnaire is intended to collect relevant information about an individual's lifestyle and to determine a personalized mechanical loading for each individual. The answers to the questionnaire questions can be used as a starting point to better understand an individual's context and to provide more specific information.

[0153] Examples of questions in such a questionnaire regarding an individual's occupation might include: • What is your profession? • How long have you been practicing this profession? • What are your main responsibilities and tasks in your job?

[0154] Examples of questions from such a questionnaire regarding an individual's physical habits might include: • How often do you exercise? • What type of physical activity do you do? (example: running, swimming, weight training, etc.) • How much time do you spend on physical activity each week? • Do you have any hobbies or leisure activities that involve regular physical activity?

[0155] Examples of questions in such a questionnaire regarding an individual's diet might include: • Briefly describe your general diet. Are you vegetarian, vegan, do you eat everything, etc.? • How often do you eat home-cooked meals versus eating out? • Do you consume specific foods or dietary supplements to support your health and well-being?

[0156] Examples of questions from such a questionnaire regarding an individual's sleep might include: • How many hours of sleep do you get on average each night? • Do you have a regular sleep routine? • Do you experience sleep problems such as insomnia or sleep disturbances?

[0157] Examples of questions from such a questionnaire regarding an individual's stressors might include: • Identify the main stressors in your life (work, family, finances, etc.). • How do you manage stress on a daily basis? • Do you have any stress management practices, such as meditation, yoga, or other relaxation techniques?

[0158] The second input data d2 makes it possible to determine the mechanical loads, in particular the most severe ones, which can affect the spine of an individual, and possibly the most repetitive movements of the spine.

[0159] The second estimate E2 may include a determination of the ultimate mechanical loads supported by the vertebrae and intervertebral discs.

[0160] The second input data d2 can be collected from various sources, such as lifestyle surveys, occupational activity records, physical activity tracking devices, medical studies, existing literature searches.

[0161] The second estimation E2 may comprise a sub-step of integrating the second input data d2 into a second centralized database.

[0162] The second database is, for example, a relational database or other storage system suitable for organizing the second input data d2, in particular in such a way that it can be queried efficiently.

[0163] Optionally, the second estimate E2 can be performed using AI2 artificial intelligence. This results in a second estimate with increased accuracy.

[0164] The second estimate E2 may include training an artificial intelligence model.

[0165] Machine learning models can be trained from the second input data d2, particularly using algorithms capable of detecting complex patterns and correlations between the data.

[0166] In the case where the first estimation and the second estimation use artificial intelligence, the processing of the first input data d1 and the second input data d2 can be carried out independently of each other.

[0167] The second estimate E2 can help determine different types of loading or mechanical loads that will be applied in the future to the spine based on different scenarios concerning an individual's lifestyle.

[0168] The method comprises a calculation E3 (CALC) for a prediction of the state of the spine from the first estimate E1 and the second estimate E2.

[0169] The process makes it possible, in particular, to precisely model the evolution of the degradation of the mechanical, biochemical and geometric properties of the spine as a function of an individual's age.

[0170] The E3 calculation allows for a prediction of the evolution of bio-chemo-mechanical properties, including kinematics, hydration and pressure of the spine, geometry of the spine and grade in the future.

[0171] The E3 calculation allows a prediction of the evolution of the level of damage to the spine to be obtained.

[0172] The E3 calculation can take into account the mechanical, chemical and physiological behavior of tissues.

[0173] The E3 calculation makes it possible to predict the temporal evolutions of the degradation of intervertebral discs and vertebrae thanks to relationships between bio-chemo-mechanical properties (resistance, rigidity, ultimate stress-strain, hydration, osmotic pressure) and grades.

[0174] E3 calculation can implement segmentation reconstruction.

[0175] The E3 calculation can be performed on each material point of the biomechanical modeling of the spine, preferably on the entire spine.

[0176] In the E3 calculation, the degradation coefficients of each material point can be calculated.

[0177] The E3 calculation can be based on physical, mechanical, fluidic, mechanical / fluid interaction, biological / chemical interaction equations that represent the behavior of an individual's spine.

[0178] Advantageously, the E3 calculation can be multi-physical, that is to say that it takes into account different phenomena, in particular biological, mechanical, biochemical, for example osmolarity, the behavior of soft tissues, the non-homogeneous nature of the tissues, the plurality of elements which interact (fluid, matrix, fibers, micro-fibers).

[0179] Advantageously, the E3 calculation can be multi-scale, that is, it concerns dimensions ranging from millimeter dimensions to nanometer dimensions via micrometer dimensions. The E3 calculation can in particular take into account fiber networks at different scales, for example micrometer and nanometer.

[0180] The E3 calculation can be carried out from a behavior law between the deformation and the stress of the spine.

[0181] The applicant carried out tests and experiments in order to determine the constants of the law of behavior and the adequate resistance of the intervertebral tissues.

[0182] Volumetric change effects can be taken into account in the E3 calculation via hydro-chemo-mechanical coupling.

[0183] The total strain gradient can be taken into account in the E3 calculation as a multiplication of the mechanical strain gradient of the solid phase and the anisotropic volumetric strain gradient induced by chemically induced fluid phase transfer.

[0184] The volumetric deformation gradient can be expressed mathematically by a nonlinear function taking into account the chemical expansion at equilibrium state and its time evolution.

[0185] The strain energy of the extracellular matrix can depend on both a strain measure, intrinsic characteristics and a damage variable to describe the failure of the extracellular matrix.

[0186] The strain energy of the collagen network can be expressed mathematically by at least one relationship between a strain measure, intrinsic characteristics, the orientation of the network constituents, their realignment, and a damage variable to describe the failure of each network constituent.

[0187] The hydrochemical strain energy of the fluid part, resulting from the chemically induced volumetric change, can be evaluated by means of a nonlinear function depending on the volumetric variation, the osmotic property, as well as the amount of damage resulting from failure events of the collagen network and the extracellular matrix.

[0188] The stress-strain response can be defined by a micromechanical approach dependent on densities, first derivatives of strain energies with respect to a strain measure, as well as layer fractions.

[0189] The continuous variation of the layer fraction across the areas of the AF annulus fibrosus can be accounted for by a particular function depending on both the thickness and the volume change of the AF annulus fibrosus layers.

[0190] The stress-strain response can take into account the fatigue behavior by a mathematical relationship depending on both the number of cycles, the intensity of the stress, as well as the grade of the intervertebral disc 1 .

[0191] Hydration level can be calculated from grayscale via an MRI of the spine.

[0192] Collagen level and orientation can be estimated from multiple MRI mapping and machine learning.

[0193] Cellular activity can be estimated.

[0194] Spinal conditions can include biological damage to the spine. Biological damage occurs when cells can no longer produce sufficient collagen due to a decrease in nutrient intake, such as during a sedentary lifestyle.

[0195] The condition of the spine may include mechanical damage to the spine.

[0196] The E3 calculation may include an estimation of mechanical damage under monotonic loading and / or under cyclic loading.

[0197] The condition of the spine may include the grade of at least one intervertebral disc and / or the grade of at least one vertebra. The grade may be determined in particular according to the Pfirrmann classification and / or the Modic classification and / or according to the VBQ system.

[0198] The condition of the spine may include Pfirrmann grade.

[0199] The Pfirrmann grade is used to estimate the level of degeneration of the spine.

[0200] The Pfirrmann grade can be calculated, in particular automatically, from two parameters: - hydration via MRI mapping with distinction between hydration of the nucleus pulposus NP and the annulus fibrosus AF after MRI image processing; - disc height, that is to say the ratio between the height of the intervertebral disc and the height of the vertebra to take into account the size of the individual.

[0201] An advantage of a method of the type described above is that it is based on a multi-scale, multi-physical model of the spine, which allows for the description of physical, biological, and chemical phenomena at the heart of matter. Such a method thus makes it possible to obtain an accurate prediction of the progression of degeneration of intervertebral discs and vertebrae, as well as changes in the shape of the spine. Such a method makes it possible to improve the prevention and treatment of spinal problems, thus opening up new perspectives in the field of spinal health.

[0202] The method may further comprise an E4 (SUP) provision of at least one message.

[0203] The at least one message may be at least one personalized recommendation, in particular intended to be transmitted to the individual.

[0204] The at least one message may be at least one directive intended to help a practitioner choose at least one treatment, for example from the group comprising drug treatments, physical or bodily treatments, surgical treatments.

[0205] The at least one message may be at least one directive for personalized surgical planning purposes.

[0206] In the provision E4 of at least one message, a report can be obtained corresponding to different mechanical load scenarios which will apply in the future to the spine depending on the second input data d2.

[0207] The provision E4 of at least one message, in particular at least one personalized recommendation, advantageously makes it possible to offer an individual solutions and advice intended to keep their spine healthy for as long as possible.

[0208] A process of the type described above can be used to assess potential risks related to an individual's lifestyle, occupation, and physical habits. Such a process can help identify harmful postures, excessive repetitive movements, or heavy loads that could lead to spinal injuries.

[0209] Specific recommendations could be provided to an individual who spends long hours sitting in front of a computer to adopt correct posture in order to prevent possible problems with their spine.

[0210] Based on the results of the E3 calculation, at least one personalized recommendation could be provided to an individual upon providing E4, e.g., ergonomic adjustments, strengthening exercises, regular breaks to avoid mechanical overload. Tailored strategies could be recommended to maintain an individual's spinal health and prevent injuries. Personalized recommendations may include appropriate restraint and movement limits.

[0211] Providing E4 with at least one personalized recommendation would enable individuals to take proactive steps to reduce the risk of chronic spinal injuries and disorders, thereby helping to improve their well-being and quality of life.

[0212] One advantage of a method like the one described above is that it can predict the condition of the spine in the future in order to inform an individual about risks to their spine before they develop pathology.

[0213] An advantage of a method of the type described above is that it helps to prevent a pathological condition of a spine by providing recommendations to an individual.

[0214] An advantage of a method of the type described above is that predictions can be obtained based on different scenarios. Such a method thus makes it possible to estimate the effects of different treatments, including drug and / or surgical treatments and / or treatments based on lifestyle changes.

[0215] Figure 3 represents a system 10 for predicting the state of a spinal column of an individual according to an embodiment of the invention. A system 10 for predicting the state of a spinal column of an individual is capable of implementing a method for predicting the state of a spinal column of an individual of the type described in relation to Figure 1.

[0216] The system 10 may comprise hardware and / or software elements configured to implement a method of the type described previously.

[0217] The system 10 may comprise at least one hardware processor 11.

[0218] The at least one hardware processor 11 may be capable of implementing: - a first estimate E1 of personalized information, in particular biochemical and / or structural, of the spine from first input data d1; - a second estimate E2 of mechanical loads likely to be applied to the spine from second input data d2; - an E3 calculation for a prediction of the state of the spine from the first estimate E1 and the second estimate E2.

[0219] The system 10 may include at least one storage unit.

[0220] The system 10 may comprise a first storage unit 13. The first storage unit 13 is capable of storing the first input data d1.

[0221] The system 10 may comprise a second storage unit 15. The second storage unit 15 is capable of storing the second input data d2.

[0222] The first storage unit 13 and / or the second storage unit 15 may be formed by any suitable means capable of storing data in a computer-readable manner.

[0223] The system 10 may comprise a delivery unit 17. The delivery unit 17 is capable of generating at least one message, in particular at least one personalized recommendation, from the calculation provided by the processor 11. The delivery unit 17 may provide at least one message intended to be transmitted to a health information system and / or to a practitioner, in particular a doctor, and / or directly to an individual. The at least one message may be a notification.

[0224] Although the method has been described above in the case where first input data and second input data are used, the method could be implemented using a single database. The first database and the second database could be combined.

[0225] The first estimate and the second estimate can be carried out in any order or simultaneously.

[0226] The present invention integrates, in addition to the analysis of MRI medical images, a big data-driven approach to feed physically based models. The objective is to predict, over the long term, the evolution of the grade of degeneration of intervertebral discs by taking into account individual specificities as well as biochemical and biomechanical parameters.

[0227] A first estimate E1 of personalized spine information is determined from first input data d1 .

[0228] These first input d1 data are obtained from imaging data obtained by magnetic resonance imaging MRI.

[0229] More specifically, these first input data d1 are obtained from a spin echo signal S of magnetic resonance imaging MRI according to the equation: k: a proportionality constant depending on the sensitivity of the receiver; p: a proton density (or DP); T R : a rehearsal time; T E : a time of echo; T-: a longitudinal relaxation rate; T2: a transverse relaxation rate.

[0230] From these data, it is possible to determine a longitudinal T1 relaxation rate and a transverse T2 relaxation rate from the pixels of the images obtained by MRI.

[0231] From the MRI images, it is thus possible to determine a spin echo signal S according to the following equation:

[0233] k: a proportionality constant depending on the receiver sensitivity;

[0234] : a proton density (or DP);

[0235] T R : a rehearsal time;

[0236] T E : a time of echo;

[0237] Ti: a longitudinal relaxation rate;

[0238] T2 : a transverse relaxation rate.

[0239] The longitudinal relaxation rate is based on an identification of the spin echo signal S as a function of the repetition time T R of a plurality of pixels of a first type of images obtained by magnetic resonance imaging.

[0240] More specifically, the longitudinal relaxation rate 7^ is based on the identification of the formula of the echo signal as a function of the repetition time T R of each pixel of the first image type according to the formula: / (T R ) = 1(T R = oo) - and from the multi-echo intensities / (T R ) of the same pixel. It is possible to determine a mapping of this longitudinal relaxation rate from multi-echo intensities 7(T R ) of the same pixel at different repetition times. Figure 4 illustrates a mapping of this longitudinal relaxation rate on a region of interest. From this mapping, it is possible to determine a vertebral bone quality score called a VBQ score.

[0241] The transverse relaxation rate T2 is based on an identification of the spin echo signal S as a function of the echo time T E of a plurality of pixels of a second type of images obtained by magnetic resonance imaging.

[0242] More specifically, the transverse relaxation rate T2 is based on the identification of the formula of the echo signal as a function of the echo time T E of each pixel of the second type of image according to the formula: 1(T E ~) = 1 O (T E = 0) exp f-— ) and from the multi-echo intensities T Eof the same pixel at different repetition times. Figure 5 illustrates a mapping of this transverse relaxation rate T2 on a region of interest. From this mapping, it is possible to determine a hydration rate of the region of interest.

[0243] The spine comprising at least one intervertebral disc comprising a nucleus pulposus NP and a fibrous ring AF. The method thus comprises a step of determining a local grade Graded" 11 of the nucleus pulposus NP and a step of determining a local grade Grades 1 of said AF fibrous ring. The determination steps are carried out by an artificial intelligence module from the longitudinal relaxation rate and / or the transverse relaxation rate T2.

[0244] This intelligence module estimates the grade by a biochemical route and uses a support vector regression approach called SVR (for "Support Vector Regression" in English) with a non-linear kernel. This model generates a relationship (curve) correlating the grade and the local fluid density of the disc (Grade = f(fluid density)), taking as input data such as: vertebral bone quality score, sex, ethnic origin. The grade estimation by the biochemical route is particularly suitable when the disc presents a mild to moderate degeneration, i.e. an overall grade less than 3.

[0245] In the case where the degeneration is more marked (visible on the obtained imaging data), it is possible to estimate the grade by a geometric route, based on a deep neural network (called "Deep Neural Network" in English). Such a network was designed to generate a relationship (curve) representing a function Grade = f (disc / vertebra height ratio). This model takes as input personalized characteristics of the patient such as sex, height, ethnic origin, disc level. To optimize this model and improve its generalization, transfer learning and cross-validation were used. Grade estimation by the geometric route is particularly suitable when the disc presents a strong degeneration, that is to say an overall grade greater than the value 3.

[0246] The generation of these relationships required a thorough analysis of the model outputs. Post-hoc analysis and data visualization techniques were used, including the following:

[0247] - feature space sampling: a large number of data points covering the entire input feature space were generated for each model;

[0248] - bulk predictions: models were trained to make predictions on these sampled data points;

[0249] - sensitivity analysis: sensitivity analysis techniques were applied to understand how variations in inputs affect model outputs;

[0250] - local regression: weighted local regression techniques of the LOWESS type (called “LOcally Weighted Scatterplot Smoothing”) were used to generate smooth curves from the discrete predictions;

[0251] - principal component analysis (PCA). This technique made it possible to reduce the dimensionality of the data and identify the main trends in the relationships between variables;

[0252] - data visualization: advanced visualization techniques, such as contour plots and heat maps, were used to represent the complex relationships between grade, fluid density and disc-to-vertebra height ratio;

[0253] - clustering: clustering algorithms, such as K-means, have been applied to identify natural groups in the data and help characterize the curves.

[0254] By combining these advanced machine learning and data analysis approaches, curves can be generated and interpreted that effectively correlate grade with fluid density and disc-to-vertebral height ratio. Both models were trained using a batch learning approach, and their performance was evaluated using metrics such as mean square error (MSE) and coefficient of determination (R 2 ).

[0255] From these models, it is thus possible to determine the local grade Grade p al of the nucleus pulposus NP and the local grade Grade l AF al of the annulus fibrosus AF.

[0256] Calculation of the overall grade Grade 9 d l ° sq “ e l of each intervertebral disc uses a mixing law based on volume fractions and local volumes.

[0257] This mixing law weights the contributions of each region from the volume fractions of the nucleus pulposus and annulus fibrosus of each disc (f NP , f AF ) and disk volumes (Volume disque , Volume NP , Volume AF ).

[0258] The overall grade of the intervertebral disc (Grade 9l ° q ^) is thus determined from the equation: in which - Grade NP corresponds to an overall grade in the nucleus pulposus (NP), such that - Grade AF corresponds to an overall grade in the annulus fibrosus (AF), such as - f NP corresponds to a volume fraction of the nucleus pulposus (NP), such as a volume fraction of annulus fibrosus (AF), such as - Volume NP corresponds to a volume of the nucleus pulposus (NP) - Volume AFcorresponds to a volume of the annulus fibrosus (AF) - Volume disque corresponds to a volume of the intervertebral disc, said volume Volume NP of the nucleus pulposus, said volume Volume AF of the annulus fibrosus, said volume Volume disque of the intervertebral disc being determined from imaging data obtained by magnetic resonance imaging (MRI).

Claims

CLAIMS 1. A method for predicting the condition of a spine of an individual, said method comprising: - a first estimate (E1) of personalized information of the spine from first input data (d1); - a second estimate (E2) of mechanical loads likely to be applied to the spine from second input data (d2); - a calculation (E3) for a prediction of the state of the spine from the first estimate (E1) and the second estimate (E2).

2. Method according to claim 1, wherein the first input data (d1) are determined from imaging data obtained by magnetic resonance imaging (MRI).

3. Method according to claim 2, wherein the first input data (d1) are determined from a spin echo signal (S) of magnetic resonance imaging (MRI) according to the equation: k: a proportionality constant depending on the sensitivity of the receiver; p: a proton density (or DP); T R : a rehearsal time; T E : a time of echo; T: a longitudinal relaxation rate; T2: a transverse relaxation rate.

4. The method of claim 3, wherein the longitudinal relaxation rate is based on an identification of the spin echo signal (S) as a function of the repetition time (T R) of a plurality of pixels of a first type of images obtained by magnetic resonance imaging and in which the transverse relaxation rate (T2) is based on an identification of the spin echo signal (S) as a function of time echo (T E ) of a plurality of pixels of a second type of images obtained by magnetic resonance imaging.

5. Method according to claim 4, wherein the spinal column comprises at least one intervertebral disc comprising a nucleus pulposus (NP) and a fibrous ring (AF), said method comprises a step of determining a local grade (Grade l p al of said nucleus pulposus (NP) and a step of determining a local grade (Grade AF al ) of said fibrous ring (AF), said determination steps being carried out by an artificial intelligence module from the longitudinal relaxation rate (TJ) and / or the transverse relaxation rate (T2).

6. Method according to claim 5, wherein said method comprises a step of determining an overall grade of the intervertebral disc (Grade^^) from in which - Grade NP corresponds to an overall grade in the nucleus pulposus (NP), such that - Grade AF corresponds to an overall grade in Fibreboard (AF), such as - f NP corresponds to a volume fraction of the nucleus pulposus (NP), such as a volume fraction of annulus fibrosus (AF), such as - Volume NP corresponds to a volume of the nucleus pulposus (NP) - Volume AF corresponds to a volume of the annulus fibrosus (AF) - Volume disque corresponds to a volume of the intervertebral disc, said volume Volume NP of the nucleus pulposus, said volume Volume AP of the annulus fibrosus, said volume Volume disqueof the intervertebral disc being determined from imaging data obtained by magnetic resonance imaging (MRI).

7. Method according to one of claims 5 or 6, in which the artificial intelligence module is capable of using a support vector regression approach.

8. Method according to any one of claims 1 to 7, wherein the second input data (d2) comprises information on the lifestyle of the individual, for example his profession and / or his physical habits and / or his sleep and / or his diet.

9. Method according to any one of claims 1 to 8, wherein said method comprises a step of obtaining a digital twin of the spine.

10. Method according to any one of claims 1 to 9, further comprising a provision (E4) of at least one message, in particular: - at least one personalized recommendation; and / or - at least one guideline intended to help a practitioner choose at least one treatment, for example from the group including drug treatments, physical or bodily treatments, surgical treatments; and / or - at least one directive for personalized surgical planning purposes.

11. System for predicting the state of a spine of an individual, the system comprising a first storage unit (13) capable of storing first input data (d1), a second storage unit (15) capable of storing second input data (d2) and at least one hardware processor (11) capable of implementing: - a first estimate (E1) of personalized information of the spine from the first input data (d1); - a second estimate (E2) of mechanical loads likely to be applied to the spine from the second input data (d2); - a calculation (E3) for a prediction of the state of the spine from the first estimate (E1) and the second estimate (E2).

12. Computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the method according to any one of claims 1 to 10.

13. Computer-readable data storage medium on which a computer program product according to claim 12 is recorded.

Images