System for assisting in the diagnosis of neurodevelopmental disorders and related mental health disorders in child or adolescent users

Abstract

The present invention also relates to a system for assisting in the diagnosis of neurodevelopmental disorders and / or associated mental health disorders in a child or adolescent user, the system comprising an interface module, an acquisition module, an extraction module, and a classification module, wherein the interface module is configured to present an interactive program to the user, preferably the program being a game, and the acquisition module is configured to simultaneously receive a plurality of temporal biometric signals resulting from the user's interaction with the interface module as the user interacts with the program, the extraction module is configured to extract or calculate from each temporal biometric signal at least one feature related to the assessment of at least one symptom associated with at least one neurodevelopmental disorder or associated mental health disorder in the child or adolescent, and the classification module is configured to assign the user one or more scores related to one or more symptoms associated with one or more neurodevelopmental disorders or associated mental health disorders based on the features extracted or calculated by the extraction module and a machine learning model pre-trained on a set of annotated data collected from the child user.

Description

[Technical Field]

[0001] The present invention relates to a system, and related computer-implemented methods, for assisting in the diagnosis of neurodevelopmental disorders and related mental health disorders in child or adolescent users.

[0002] The present invention may correspond to a medical device used to automatically determine at least one score for symptoms associated with one or more neurodevelopmental disorders and related mental health disorders in a child or adolescent through a combination of biometric and machine learning techniques. [Background technology]

[0003] Globally, one in six children and adolescents suffers from neurodevelopmental and mental health disorders that are associated with long-term negative consequences, including loss of self-esteem, failure in school, and suicidal thoughts. An estimated 166 million young people are affected in their daily lives, whether academically, behaviorally, or socially.

[0004] Neurodevelopmental disorders (NDDs) can begin in childhood and become lifelong chronic conditions, reducing life expectancy by 10-15 years. However, the earlier these disorders are detected, the earlier appropriate in-school and out-of-school interventions and support can be implemented, demonstrating significant improvements on multiple levels (e.g., reduced symptoms, improved neurodevelopment).

[0005] Unfortunately, currently, receiving a diagnosis takes 2-4 years and multiple visits to different medical institutions. Furthermore, psychiatry has little experience with clinical adjudication. Currently, computerized patient records do not provide this functionality. Finally, clinical data resources are largely unstructured.

[0006] Furthermore, up to 10% of children have one or more neurodevelopmental disorders. For example, learning disabilities are present in 44% of children with attention-deficit / hyperactivity disorder (TDAH), 65–85% of children with autism, and 30–70% of children with co-occurring TDAH and autism. Finally, neurodevelopmental disorders and related mental health disorders have complex clinical phenotypes that are often associated with trans-nosological aspects such as emotion dysregulation and executive function disorders. These aspects impair adaptive functioning and often lead to negative outcomes such as peer rejection, bullying, harsh parenting, or severe negligence. Furthermore, the phenotypes of neurodevelopmental disorders have a high degree of overlap, making diagnosis and treatment planning difficult in everyday clinical practice.

[0007] Therefore, there is a real need to use powerful and meaningful tools to understand the complex set of characteristics of each child on a large scale, thereby enabling us to understand the underlying mechanisms and therefore provide effective support to those who need it. However, currently, there is no single consensus-based digital tool that facilitates the initial assessment and diagnosis of patients and their referral to the most appropriate treatment. The purpose of this tool is to provide a more agnostic and comprehensive (transdiagnostic) understanding of patients' symptoms, monitor the progression and severity of their symptoms, and provide indicators for individualizing treatment. The present invention proposes to improve this shortcoming.

[0008] Finally, it should be noted that the brain structure and social behavior of children and adolescents differ from those of adults, making it essential to have tools adapted to these particularities. Indeed, applying tools developed for adults to children and adolescents, whose cognitive, behavioral, and biological development is not yet complete and whose physiological responses are not comparable, will not produce results that can be used and interpreted appropriately. Summary of the Invention

[0009] Thus, to address these various issues, an object of the present invention is a system for assisting in the diagnosis of neurodevelopmental disorders and / or associated mental health disorders in a child or adolescent user, the system comprising an interface module, an acquisition module, an extraction module, and a classification module, wherein the interface module is configured to present an interactive program to the user, preferably the program being a game, the acquisition module is configured to simultaneously receive a plurality of temporal biometric signals resulting from the user's interaction with the interface module as the user interacts with the program, the extraction module is configured to extract or calculate from each temporal biometric signal at least one feature that is relevant to the assessment of at least one neurodevelopmental disorder or associated mental health disorder in the child or adolescent, and the classification module is configured to assign to the user one or more scores related to one or more neurodevelopmental disorders or associated mental health disorders based on the features extracted or calculated by the extraction module and a machine learning model pre-trained on a set of annotated data collected from the child or adolescent user.

[0010] Preferably, the score(s) assigned by the classification module relate to one or more symptoms, which themselves relate to one or more neurodevelopmental or related mental health disorders, allowing each symptom to be quantified individually, the presence and severity of which may vary depending on the disorder being diagnosed.

[0011] The annotated training data collected from the child or adolescent user is pre-annotated to determine the presence or absence of at least one symptom associated with at least one neurodevelopmental disorder and / or at least one associated mental health disorder, and the annotated data includes at least one severity associated with the symptom of the diagnosed disorder.

[0012] The various temporal biometric signals (i.e., containing the user's information characteristics) recorded by the acquisition module while the user interacts with the interface module produce accurate measurements of the user's response to one or more external stimuli. These multimodal signals are processed by the extraction module, and characteristics are extracted or calculated from these signals.

[0013] The characteristics can be obtained directly through mathematical calculations or through the use of machine learning models. Advantageously, the characteristics are selected in relation to a neurodevelopmental disorder(s) or related mental health disorder. These characteristics may constitute all or part of the user's digital phenotype.

[0014] A final classification is performed by a classification module, which uses the extracted features to assign one or more scores to one or more disorders to aid in diagnosing the disorder in question in the child or adolescent user. Based on the resulting score(s), appropriate treatment responses can be suggested to the user.

[0015] Until now, a system that analyzes multiple types of signals simultaneously and over time and automatically assigns scores to facilitate transdiagnosis of neurodevelopmental disorders in children or adolescents and subsequent individualized treatment referral, as proposed in the present invention, has not been possible in the prior art.

[0016] Preferably, the neurodevelopmental disorder may be selected from attention deficit hyperactivity disorder, communication disorder, specific learning disorder, and / or movement disorder. According to the ICD-10 classification definition, the codes for neurodevelopmental disorders are F-80 to F-98.

[0017] Preferably, the associated mental health disorder may be selected from depression, anxiety disorders, sleep disorders, social interaction disorders, and / or emotion regulation disorders, which correspond to codes F-20 to F-59 as defined in the ICD-10 classification.

[0018] Preferably, the plurality of temporal biometric signals may be selected from video signals including tracking the user's eyes or face, position signals of the user's body parts, audio signals of the user's voice, an electroencephalogram, and / or physiological signals of the user.

[0019] These biometric signals are particularly relevant for the assessment of neurodevelopmental or related mental health disorders in children or adolescents.

[0020] Preferably, the acquisition module may comprise a camera and is configured to receive a video signal, including tracking the user's eyes, and the characteristics extracted from the video signal are selected from gaze position on a surface, a gaze heatmap on the surface, the user's emotional profile, facial action units, the user's blinks, the user's gaze acceleration, the angle formed by the user's gaze relative to a normal to the surface, saccades along the gaze path, the number of fixations on a particular item, and the time spent looking at a particular item.

[0021] These characteristics are advantageously relevant to assessing neurodevelopmental disorders or related mental health disorders based on video signals of a user's eyes and / or face.

[0022] Preferably, the acquisition module may comprise a touch surface or a specific acquisition device and is configured to receive position signals from at least one finger or hand of the user, and the characteristics extracted from the position signals are selected from finger or hand trajectory, finger or hand velocity, finger or hand acceleration, finger or hand pressure, a heat map of finger or hand position, number of screen touches by the user.

[0023] These characteristics are advantageously relevant to assessing neurodevelopmental disorders or related mental health disorders based on signals obtained from the position of a user's fingers or hands on the surface, such as trajectory, velocity, acceleration, finger pressure, or hand gyroscopic data. Preferably, the surface is a touchscreen and the interface module displays an interactive program on the touchscreen.

[0024] Preferably, the acquisition module may comprise a microphone and is configured to receive a signal, and the characteristics extracted from the audio signal are selected from a transcription of the user's speech, frequency variability of the user's voice, acoustic characteristics of the speech signal such as instability (also known as jitter), fluctuation, tremor, harmonic to noise ratio, frequency fluctuation ratio, amplitude fluctuation ratio, peak slope, mean first formant (F1), mean second formant (F2), F1 variability, F2 variability, F1 range, vowel space, linear predictive coding coefficients, Mel-frequency cepstral coefficients (MFCC), mean fundamental frequency (F0), F0 variability, F0 range, mean intensity, intensity variability, energy variability, energy rate, maximum utterance duration, speaking rate, articulatory rate, mean speech unit duration, mean pause duration, pause rate, and total number of pauses.

[0025] These characteristics are advantageously relevant to assessing neurodevelopmental disorders or related mental health disorders based on audio signals including a user's voice or acoustic responses.

[0026] Preferably, the acquisition module may comprise at least one of a sensor for measuring physiological signals, a touch screen, a mouse, a microphone, a camera, a keyboard, an accelerometer, and a gyroscope.

[0027] For example, physiological signals may include an electroencephalogram, an electrocardiogram, heart rate, blood pressure, and the like.

[0028] Preferably, the system may be contained in a tablet equipped with a touch screen, camera, and microphone.

[0029] Such a system is compact and easy to use for a child or adolescent user.

[0030] Preferably, the extraction module may use at least one machine learning algorithm that has been pre-trained on a set of annotated signals collected from child or adolescent users.

[0031] In particular, the signals may be annotated by a medical professional using a questionnaire.

[0032] Preferably, the machine learning model of the classification module may include at least one artificial neural network.

[0033] Preferably, the artificial neural network may consist of a single layer.

[0034] Such neural networks improve the interpretability of the resulting classifications, making it possible to determine how different features contribute to the final classification.

[0035] Preferably, the system may comprise a comparison module configured to compare the score(s) assigned to the user with one or more reference values.

[0036] Preferably, the system may further comprise a display module including a display for displaying the obtained score(s) to the user or professional.

[0037] Preferably, the interface module comprises a display device, and the acquisition module, extraction module and classification module are included in one or more processing units each comprising a processor.

[0038] According to another embodiment, the present invention provides a method for assisting in the diagnosis of neurodevelopmental disorders and / or associated mental health disorders in child or adolescent users. a human-machine interface configured to present an interactive program to a user, preferably a game, and to simultaneously receive a plurality of temporal biometric signals resulting from the user's interactions with the interface module as the user interacts with the program; at least one processor, extracting or calculating from each temporal biometric signal at least one characteristic that is relevant to the assessment of at least one neurodevelopmental disorder or related mental health disorder in the child or adolescent; assigning one or more scores associated with one or more neurodevelopmental disorders or associated mental health disorders to the user based on the characteristics extracted or calculated by the extraction module and a machine learning model pre-trained on a set of annotated training data collected from the child or adolescent user that has been annotated to determine the presence or absence of at least one neurodevelopmental disorder and / or at least one associated mental health disorder, the annotated data including at least one severity level associated with the diagnosed disorder; a processor configured to: at least one output interface (i.e., at least one output) configured to provide the score(s) assigned to the user, and optionally, the at least one output interface configured to transmit the score(s) assigned to the user to a healthcare professional who can diagnose at least one neurodevelopmental disorder and / or associated mental health disorder in the user; The present invention also relates to a system comprising:

[0039] Advantageously, the at least one processor is also configured to identify a probability that the child or adolescent user has at least one symptom associated with a neurodevelopmental disorder and / or associated mental health disorder.

[0040] The present invention also relates to a computer-implemented method for assisting in the diagnosis of neurodevelopmental disorders and / or associated mental health disorders in a child or adolescent user, the method comprising at least the steps of presenting an interactive program to the user; simultaneously acquiring a plurality of temporal biometric signals resulting from the user's interaction with the interactive program; extracting or calculating, for each temporal biometric signal, at least one feature that is relevant to the assessment of neurodevelopmental disorders or associated mental health disorders in children or adolescents; and assigning one or more scores associated with one or more neurodevelopmental disorders based on the extracted or calculated features and a machine learning model pre-trained on a set of annotated data collected from the child user.

[0041] The present invention also relates to a computer-implemented method for assisting in the diagnosis of neurodevelopmental disorders and / or associated mental health disorders in a child or adolescent user, the method comprising at least the steps of presenting an interactive program to the user; simultaneously acquiring a plurality of temporal biometric signals resulting from the user's interaction with the interactive program; extracting or calculating, for each temporal biometric signal, at least one feature that is relevant to assessing symptoms related to neurodevelopmental disorders or associated mental health disorders in the child or adolescent; and assigning one or more scores related to one or more symptoms of the neurodevelopmental disorder based on the extracted or calculated features and a machine learning model pre-trained on a set of annotated data collected from the child user.

[0042] The preferred and advantageous properties associated with the systems described above are also applicable to the methods described below.

[0043] definition For purposes of the present invention, the following terms are defined as follows:

[0044] "Neurodevelopmental disorders" are characterized by disturbances in a child's cognitive, behavioral, or emotional development that impair adaptive functioning in school, social, and home environments.

[0045] A "multimodal signal" refers to a set of components simultaneously represented or transmitted through multiple sensory modalities or communication channels. These sensory modalities may include vision, hearing, touch, taste, and smell. The components of a multimodal signal allow a receiver to extract a variety of information. By combining components from different sensory modalities, a multimodal signal provides richer and more complex information that reflects the diversity of sensory stimuli present in the environment. For example, a video may contain both images and sound, which are two different modalities.

[0046] In the present invention, the terms "adapted" and "configured" are used broadly to encompass the initial configuration, adaptation, or subsequent supplementation of the device by hardware or software means (including firmware), or any combination of these elements.

[0047] The term "processor" should not be construed as being limited to hardware capable of executing software, but generally refers to a processing device that may comprise, for example, a computer, a microprocessor, an integrated circuit, or a programmable logic device (PLD). A processor may also include one or more graphics processing units (GPUs), whether used for computer graphics and image processing or other functions. Furthermore, instructions and / or data enabling the performance of the associated and / or resulting functions may be stored on any medium readable by the processor, such as, for example, an integrated circuit, a hard disk, an optical disk such as a CD (compact disc), a DVD (digital versatile disc), a RAM (random access memory), or a ROM (read-only memory). Instructions may be stored in hardware, software, firmware, or any combination thereof, among others. [Brief explanation of the drawings]

[0048] [Figure 1] FIG. 1 illustrates a system according to one embodiment of the present invention. [Figure 2] 2 illustrates another embodiment of a system according to the present invention. [Figure 3] 3 is a flowchart illustrating various steps of a computer-implemented method according to one embodiment of the present invention. [Figure 4A] Here is an example. [Figure 4B] Here is an example. DETAILED DESCRIPTION OF THE INVENTION

[0049] The present description illustrates the principles of the present disclosure. It should be noted that those skilled in the art will be able to design various arrangements which, although not explicitly described or shown herein, embody the principles of the invention and are within the scope of the invention.

[0050] All examples and conditional language cited in this document are intended to aid the reader in understanding the principles of the present invention and the concepts contributed to the advancement of the art by the inventors, and should not be construed as being limited to the specifically cited examples and conditions.

[0051] Moreover, all descriptions of principles, aspects, and embodiments of the invention, as well as specific examples thereof, are intended to encompass both structural and functional equivalents thereof. Additionally, these equivalents are intended to include both currently known equivalents and equivalents developed in the future (i.e., any elements developed that perform the same function, regardless of structure).

[0052] Thus, for example, those skilled in the art will understand that the block diagrams presented herein may represent conceptual views of circuitry illustrating the principles of the invention. Similarly, it will be appreciated that all flowcharts, step diagrams, etc. illustrating various processes, whether or not a computer or processor is explicitly shown, may inherently reside within a computer-readable medium and operating system.

[0053] The functions of the various elements shown in the figures may be performed through the use of dedicated hardware and hardware capable of executing software in association with appropriate software. If performed by a processor, the functions may be performed by a single dedicated processor, a single shared processor, or a plurality of individual processors, some of which may be shared.

[0054] It will be understood that the elements shown in the figures may be implemented in various forms of hardware, software, or a combination thereof. Preferably, these elements are implemented in a combination of hardware and software on one or more appropriately programmed multi-purpose devices, which may include a processor, memory, and input / output.

[0055] system A system 100 according to one embodiment of the present invention is shown in Figure 1. The system 100 is configured to aid in the diagnosis of neurodevelopmental disorders or mental health disorders associated with these neurodevelopmental disorders in children or adolescents.

[0056] It should be noted that neurodevelopmental disorders, such as autism spectrum disorders, and / or related mental health disorders, can be detected within the first few months of life and may be diagnosed as early as age 2. Scientists estimate that in adolescents, biological brain development continues until approximately age 20. Therefore, system 100 is preferably targeted at children from about age 2 years old (i.e., the age at which a child can interact with system 100) and older, through adolescents about age 20.

[0057] Neurodevelopmental disorders may be selected from attention deficit hyperactivity disorder, communication disorders, specific learning disorders, and / or movement disorders, among others. According to the ICD-10 classification, neurodevelopmental disorders are coded F-80 through F-98. ICD corresponds to the International Classification of Diseases, a medical classification system developed by the World Health Organization (WHO), and 10 corresponds to the 10th edition of this classification.

[0058] The associated mental health disorder may be selected from depression, anxiety disorders, sleep disorders, social interaction disorders, and / or emotion regulation disorders, which correspond to codes F-20 to F-59 as defined in the ICD-10 classification.

[0059] The system 100 comprises, inter alia, an interface module 110 , an acquisition module 120 , a processing unit 130 including an extraction module 131 , a classification module 133 and, optionally, a comparison module 135 .

[0060] The interface module 110 is configured to present the user 1 with an interactive program, i.e., one or more tasks that stimulate the user and require a response from the user. Preferably, the interactive program is a game, which increases the likelihood of participation from children and adolescents. The type of game may be advantageously adapted to detect specific symptoms. The interface module 110 may comprise a display screen and means for the user to interact with the interactive program (e.g., a touch surface, a mouse, and / or a keyboard).

[0061] Advantageously, if the interactive program is a game, it can include progression events and challenges and can assess different levels of response from the user. For example, the game can be configured to define expected behavioral trajectories from the user based on biometric signals and characteristics targeted according to neurodevelopmental disorders or related mental health disorders. The game can include the concept of failure or success at each level played by the user.

[0062] Advantageously, the duration of events is also recorded, such as the time spent on a level, the time between two game actions, etc. As part of recording the position signal of at least one finger, touch events can be calculated to build a complete trace of all digital input collected on the touch surface, including the velocity, length and duration of gestures.

[0063] The acquisition module 120 is configured to simultaneously receive multiple different temporal biometric signals 121 resulting from the user's interaction with the interface module 110 as the user interacts with the program. The temporal biometric signals are signals that are indicative of and unique to the user. Such signals may be acquired in a variety of ways, such as using the user's means of interaction with the interface module (e.g., a touch surface, a mouse, and / or a keyboard) or using another specific sensor (e.g., a camera, a microphone, or a biometric sensor).

[0064] Advantageously, each temporal biometric signal is unimodal (i.e., auditory, visual, tactile, etc.), e.g., a temporal biometric signal corresponds to a sound made by a user, a temporal biometric signal corresponds to a user's interaction with a touch surface, a temporal biometric signal corresponds to a video recording of a user, etc.

[0065] The plurality of temporal biometric signals 121 may be selected from video signals including tracking the user's eyes or face, position signals of the user's body parts, audio signals, electroencephalograms, and / or physiological signals.

[0066] In one embodiment, the acquisition module 120 comprises a camera. The camera may be configured to receive a video signal containing a series of images of a user. In particular, the video signal may enable tracking of the user's eyes. Features extracted from the video signal are selected from gaze position on a surface, a gaze heatmap on a surface, an emotional profile of the user, facial action units, user blinks, a user's gaze acceleration, the angle formed by the user's gaze relative to a normal to the surface, saccades along the gaze path, the number of fixations on a particular item, and the time spent looking at a particular item.

[0067] In an exemplary embodiment, the acquisition module 120 comprises a touch surface or specific acquisition device and is configured to receive position signals from at least one finger or hand of a user, and the characteristics extracted from the position signals are selected from the acceleration of the user's finger or hand, a heat map of the finger or hand position, and the number of screen touches by the user.

[0068] In an exemplary embodiment, the acquisition module 120 further comprises at least one microphone. The microphone is configured to receive an audio signal, particularly audio including sounds produced by a user's voice while interacting with a program. The characteristics extracted from the audio signal are selected from a transcription of the user's speech, frequency variability of the user's voice, acoustic characteristics of the speech signal, such as jitter, fluctuation, tremor, harmonic-to-noise ratio, frequency fluctuation ratio, amplitude fluctuation ratio, peak slope, mean first formant (F1), mean second formant (F2), F1 variability, F2 variability, F1 range, vowel space, linear predictive coding coefficients, Mel-frequency cepstral coefficients (MFCCs), mean fundamental frequency (F0), F0 variability, F0 range, average intensity, intensity variability, energy variability, energy rate, maximum utterance duration, speaking rate, articulatory rate, average speech unit duration, average pause duration, pause rate, and total number of pauses.

[0069] Table 1 lists the various acoustic properties that can be used. [Table 1] TIFF2026501743000003.tif173165

[0070] In an exemplary embodiment, the acquisition module 120 may include at least one of a sensor for measuring physiological signals, a touchscreen, a mouse, a microphone, a camera, a keyboard, an accelerometer, and a gyroscope. The acquisition module 120 may alternatively use a touch surface, a touchscreen, or a mouse of an interface module.

[0071] The processing unit 130 may consist of one or more different processors and comprises an extraction module 131 and a classification module 133 .

[0072] The extraction module 131 is configured to extract or calculate at least one characteristic from each temporal biometric signal that is relevant to assessing at least one symptom related to a neurodevelopmental disorder or related mental health disorder in a child or adolescent. The extraction module 131 can perform a calculation (e.g., a mathematical transformation, application of a specific mathematical function) on the biometric signal or use a machine learning model. If this is a machine learning model, the model is pre-trained on a training database containing a set of annotated data collected from child or adolescent users and produces better results for these users than known methods.

[0073] For example, if each temporal biometric signal is unimodal, such as an acoustic signal, the extraction may include, for example, extracting features, such as the acoustic features described above (see Table 1), to each associated temporal signal. Thus, the machine learning model may advantageously include multiple artificial intelligence models already trained, for example, using deep learning networks already trained to detect face and eye points, head position, emotion recognition, and gaze vector estimation.

[0074] Advantageously, the machine learning model(s) used are equipped with an explainability module that extracts features to make them interpretable based on human experience and useful for prediction.

[0075] Since different temporal biometric signals can be acquired at different sampling frequencies (e.g., video at 28 frames per second, audio signals at approximately 16 kHz, haptic signals at approximately 100 Hz), it is advantageous to align the acquired data to allow comparisons between the same or different users. Preferably, features are extracted for each modality and aggregated at the smallest sampling rate (typically that of the video) using an average or summation function depending on the feature in question. Thus, the temporally aligned multimodal data advantageously consists of n features per frame (video), where n is an integer greater than or equal to 1.

[0076] In some cases, temporal biometric signals may have long or short durations depending on the user interaction, so it is advantageous to introduce the concept of grouped data, dividing each user's signal into any number of time intervals (e.g., p time intervals, where p is an integer greater than or equal to 1, for example, but not limited to, p equal to 100). Thus, since the number of frames varies depending on the participant and the interactive program used, intervals can group a variable number of frames per participant, but the measured data is distributed over the same number of intervals. Within each interval, the characteristic can be aggregated across frames, for example, using an average or sum function, depending on the characteristic being investigated.

[0077] Thus, the extraction model 131 takes as input an acquired biometric signal corresponding to a time series containing a set of characteristics and converts it into a matrix of size nxp containing n characteristics to be extracted or calculated and divided into p intervals.

[0078] The classification module 133 is further configured to assign the user one or more scores for one or more symptoms of at least one neurodevelopmental disorder or related mental health disorder based on the features extracted or calculated by the extraction module and a machine learning model (pre-trained on a set of annotated data collected from the child or adolescent user) to represent the user's mental state as a network of symptoms. These scores should be consistent with scores on scales shown to have clinical significance. Indeed, each scale relates to at least one symptom of at least one disorder, and the score can determine the degree of presence or severity of the symptom.

[0079] For example, one of the most widely used assessment tools for determining autism spectrum disorder is the Diagnostic and Statistical Manual of Mental Disorders (DSM-V, 5th Edition) published by the American Psychiatric Association. The DSM-V provides the official diagnostic criteria for autism spectrum disorder (ASD). The DSM-V, which replaced DSM-IV in 2013, introduced the category of autism spectrum disorder, encompassing several existing disorders, including childhood autism, Asperger syndrome, and pervasive developmental disorder not otherwise specified. The DSM-V diagnostic criteria for ASD are based on two main areas: persistent deficits in social communication and restricted and repetitive behaviors. Each criterion is scored on a scale of 1–3 or 0–3, depending on the specific criteria. Combining these assessments provides an overall assessment of the severity of ASD symptoms. For example, for deficits in nonverbal social communication behaviors, the severity scale is as follows: 0 - no deficits observed, 1 - mild deficits, 2 - moderate deficits, and 3 - severe deficits. These severity ratings are used for each diagnostic criterion and contribute to an overall assessment of ASD based on the presence and severity of symptoms in the areas of social communication, restricted behaviors, and repetitive behaviors.

[0080] The training data collected from the child or adolescent user is pre-annotated to determine the presence or absence of at least one symptom associated with at least one neurodevelopmental disorder and / or at least one associated mental health disorder, and the annotation includes at least one severity associated with the diagnosed disorder symptom. If the child or adolescent does not have symptoms associated with the neurodevelopmental disorder or associated mental health disorder, this severity may be zero. Preferably, healthy children and adolescents are screened to ensure that their parents have not reported any significant psychiatric or neurological illnesses, brain injuries, or other medical histories that may affect brain development. Preferably, premature infants (<36 weeks gestation) who have been significantly exposed to prenatal toxicants, particularly alcohol or drugs, are excluded.

[0081] Advantageously, to perform cross-validation of a machine learning model (also called "k-fold validation" in English), the training database can be divided into k equal parts (k is an integer greater than or equal to 1).

[0082] The machine learning model used may advantageously comprise an artificial neural network, preferably with a single layer, in order to be able to explain in a simplified manner the score(s) obtained, i.e. to determine an estimate of how much different characteristics contribute to obtaining a given score.

[0083] Alternatively, the machine learning model may include a decision tree (single or in a forest), or a k-nearest neighbor algorithm, or may include multiple models of these types and aggregate the results of each of the models (e.g., by averaging the results of each of the models).

[0084] In an advantageous embodiment, the machine learning model may be a Clustering Representation Learning on Incomplete time-series (CRLI) learning model, which allows the synchronization of different acquired biometric signals and identifies specific biomarkers by grouping extracted or calculated features in a non-temporal structured latent space, but has the advantage of not reducing the temporal dimension.

[0085] According to another advantageous embodiment, the machine learning model may be a Masked Hierarchical Cluster-Wise Contrastive Learning model (called in English "Masked Hierarchical Cluster-Wise Contrastive Learning" or MHCCL), which has the advantage of being able to identify specific biomarkers by grouping extracted or calculated features in a non-temporally structured latent space, reducing the temporal dimension, but without the advantage of synchronizing the different acquired biometric signals.

[0086] According to other embodiments, the machine learning model may be a variational autoencoder (also known as VAE), which also involves building a latent network, or a Valence Aware Dictionary and sEntiment Reasoner, or VADER for short, which is a vocabulary and simple rule-based model for analyzing emotions.

[0087] Thus, the classification module 133 takes as input an nxp size matrix containing n extracted or computed features divided into p intervals and transforms it into a reduced vector via a latent space that allows the model's training data to be grouped or "clustered."

[0088] As shown, the machine learning model used may be pre-trained on a dataset collected from child or adolescent users. For example, training data may be collected by recording the child or adolescent user's temporal biometric signals while interacting with a game. Features are then extracted from these signals, and one or more scores related to neurodevelopmental disorders or related issues are obtained using questionnaires completed by experts. Thus, the model is trained based on the features and scores obtained from questionnaires for a set of users. Examples of questionnaires that can be used include the ADHD Scale, MADRS (Montgomery-Asberg Depression Rating Scale), SHAPS (Snaith-Hamilton Pleasure Scale), ADOS (Autism Diagnostic Observation Schedule), WISC (Wechsler Intelligence Scale for Children), CBCL (Child Behavior Checklist), SDQ (Strengths and Difficulties Questionnaire), and K-SADS (Semi-Structured Interview for Childhood Psychiatric Disorders). Advantageously, the choice of questionnaire(s) is made depending on the symptom(s) related to the neurodevelopmental disorder or related mental health disorder being studied. The aim is therefore to optimize the parameters of the machine learning model so that it is able to predict with comparable or even greater precision the scores obtained from clinical assessments traditionally carried out using the various questionnaires mentioned above.

[0089] Furthermore, machine learning models may be advantageously used to identify specific digital biomarkers based on multiple characteristics extracted or calculated from the acquired biometric signals, such as the user's playability (including, for example, all tactile events related to the game or their reactivity, facial expressions, shouts / sounds emitted and their characteristics, etc.), the user's head position (i.e., head orientation), or an estimation of the user's gaze (i.e., gaze direction), including detection of saccades, gaze fixations, blinks, checking whether the user is looking at the screen, etc., or some landmarks on the user's face.

[0090] In one advantageous embodiment, the machine learning model of the interface module 110 and the machine learning model of the acquisition module are common (i.e., the system 100 has a single machine learning model that is common to the two modules), while in an alternative embodiment, these machine learning models are different.

[0091] The interface module 110 may further comprise a display for displaying the score(s) obtained.

[0092] The system 100 may comprise a comparison module 135 configured to compare the score(s) assigned to the user with one or more reference values. Such a comparison may thus enable the user's scores to be mapped onto a chart to identify different symptoms related to neurodevelopmental disorders or related mental health disorders, improve diagnostic support for these disorders, and estimate their prevalence.

[0093] System 100 may be advantageously included in a tablet 200 ( FIG. 2 ) with a touchscreen 210, a front camera 220, and a microphone 230. User 1 may interact with tablet 200 by using one or more fingers 202 when touching or swiping the screen. Tablet 200 may record video of the user's face, including the user's eyes or gaze 2. Tablet 200 may also record audio signals, including noises and speech 3, made by the user using microphone 230. Tablet 200 may also record signals representing interactions between the user's body parts, e.g., hands or fingers 4, and touchscreen 210. Tablet 200 processes the recorded signals locally (i.e., extracts characteristics and assigns a score) or transmits them to a remote unit that performs processing and presents a report to a physician who can make a diagnosis and provide an appropriate treatment response. System 100 thus combines cognitive, emotional, behavioral, and brain measurements to provide a user-friendly, comprehensive tool that can be deployed on a large scale to aid in the diagnosis of neurodevelopmental disorders and / or related health problems.

[0094] method Referring now to the flowchart of Figure 3, a computer-implemented method for assisting in the diagnosis of neurodevelopmental disorders and / or associated mental health disorders in child or adolescent users will be described.

[0095] First, the method includes a step E1 of presenting an interactive program, preferably a game, to a user.

[0096] The method then includes a step E2 of simultaneously acquiring a plurality of temporal biometric signals resulting from the user's interaction with the interactive program, which signals can be stored in a local memory or transmitted to a remote memory, such as a remote server.

[0097] The method then comprises a step E3 of extracting or calculating, for each temporal biometric signal, at least one characteristic relevant for the assessment of a neurodevelopmental disorder or related mental health disorder in children or adolescents.

[0098] Finally, the method includes a step E4 in which the user is assigned one or more scores related to one or more neurodevelopmental disorders based on the extracted or calculated features and a machine learning model pre-trained on a set of annotated data collected from the pediatric user.

[0099] These score(s) are finally compared with one or more reference values ​​in step E5, making it possible to build a network of symptoms specific to the user.

[0100] Example A stimulating game is presented to the user on the tablet, while a video signal is recorded from the front camera. From this video signal (a sequence of images) containing the user's face, a machine learning algorithm trained on children's faces can extract scores for multiple emotions, particularly seven emotions, including "fear." The machine learning algorithm can use, for example, a facial movement unit to compare the positions of multiple marks on the user's face and infer a score for each emotion. The output of this algorithm is a vector (feature vector) for each image in the image sequence, containing a score for each of the tested emotions. A second algorithm, trained on the video signal and questionnaires about attention disorders ("ADHD Scale") and sleep disorders ("PSQI" Pittsburgh Sleep Quality Index), can then derive a score for the "ADHD Scale" or "PSQI Scale" from the set of generated vectors. The second algorithm is advantageously a recurrent neural network (RNN).

[0101] Figure 4A shows the predicted attention disorder ("ADHD") score based on the scores obtained for the emotion "fear" for four different users. Each point corresponds to the score calculated on the ADHD scale for one image in each of the user's video sequences (in the example, each video contains 5,000 images). A good correlation was observed for all four users tested.

[0102] Figure 4B shows the predicted sleep disturbance ("PSQI") scores based on the scores obtained for the "fear" emotion for the same four users. A good correlation was also observed for all users tested.

Claims

1. 1. A system (100) for assisting in the diagnosis of neurodevelopmental disorders and / or associated mental health disorders in a child or adolescent user, comprising: an interface module (110) adapted to present an interactive program to said user, said program being preferably a game; an acquisition module (120) configured to simultaneously receive a plurality of temporal biometric signals (121) resulting from the user's interaction with the interface module (110) as the user interacts with the program; an extraction module (131) configured to extract or calculate from each temporal biometric signal at least one characteristic relevant to the assessment of at least one neurodevelopmental disorder or related mental health disorder in said child or adolescent; a classification module (133) configured to assign one or more scores for one or more symptoms of one or more neurodevelopmental disorders or associated mental health disorders to a user based on the features extracted or calculated by the extraction module (131) and a machine learning model pre-trained on a set of annotated training data collected from a child or adolescent user annotated to determine the presence or absence of at least one symptom of at least one neurodevelopmental disorder and / or at least one associated mental health disorder, wherein the annotated data includes at least one severity level for the symptom of a diagnosed disorder; A system comprising:

2. The system (100) of claim 1, wherein the neurodevelopmental disorder is selected from attention deficit hyperactivity disorder, a communication disorder, a specific learning disorder, and / or a movement disorder.

3. 10. The system (100) of any one of the preceding claims, wherein the associated mental health disorder is selected from depression, anxiety disorders, sleep disorders, social interaction disorders, and / or emotion regulation disorders.

4. 10. The system (100) of any one of the preceding claims, wherein the plurality of temporal biometric signals (121) are selected from video signals including tracking the user's eyes or face, position signals of a user's body parts, audio signals, electroencephalograms, and / or physiological signals.

5. 10. The system of claim 1, wherein the acquisition module comprises a camera and is configured to receive a video signal including tracking the user's eyes, and wherein the characteristics extracted from the video signal are selected from gaze position on a surface, a gaze heatmap on a surface, a user's emotional profile, facial action units, user blinks, a user's gaze acceleration, an angle formed by the user's gaze relative to a normal to the surface, saccades along a gaze path, a number of fixations on a particular item, and a time spent viewing a particular item.

6. 10. The system of claim 1, wherein the acquisition module comprises a touch surface or a specific acquisition device and is configured to receive position signals from at least one finger or hand of the user, and wherein the extracted characteristics from the position signals are selected from finger or hand trajectory, finger or hand velocity, finger or hand acceleration, finger or hand pressure, a heat map of finger or hand position, and number of screen touches by the user.

7. 10. The system of claim 1, wherein the acquisition module comprises a microphone and is configured to receive an audio signal, and the extraction module is configured to extract features extracted from the audio signal selected from a transcript of the user's speech, frequency variability of the user's voice, acoustic characteristics of the speech signal, such as jitter, fluctuation, tremor, harmonic to noise ratio, frequency fluctuation ratio, amplitude fluctuation ratio, peak slope, mean first formant (F1), mean second formant (F2), F1 variability, F2 variability, F1 range, vowel space, linear predictive coding coefficients, Mel-Frequency Cepstral Coefficients (MFCC), mean fundamental frequency (F0), F0 variability, F0 range, average intensity, intensity variability, energy variability, energy rate, maximum phonation time, speaking rate, articulatory rate, average speech unit duration, average pause duration, pause rate, and total number of pauses.

8. 10. The system (100) of any one of the preceding claims, wherein the acquisition module (120) comprises at least one of a sensor for measuring physiological signals, a touchscreen, a mouse, a microphone, a camera, a keyboard, an accelerometer, and a gyroscope.

9. 10. The system (100) of any one of the preceding claims, included in a tablet (200) with a touchscreen (210), a camera (220) and a microphone (230).

10. 10. The system (100) of any one of the preceding claims, wherein the extraction module (131) uses at least one machine learning model pre-trained on a set of annotated signals collected from a child or adolescent user.

11. 10. The system of claim 1, wherein the machine learning model of the classification module (133) comprises at least one artificial neural network.

12. 10. The system of claim 9, wherein the artificial neural network comprises a single layer.

13. 10. The system of any one of the preceding claims, comprising a comparison module (135) configured to compare the score(s) assigned to the user with one or more reference values.

14. 10. The system of any one of the preceding claims, wherein the interface module (110) includes a display for displaying the obtained score(s) to the user or expert.

15. 1. A computer-implemented method for assisting in the diagnosis of neurodevelopmental disorders and / or associated mental health disorders in a child or adolescent user, comprising at least: - presenting (E1) an interactive program, preferably a game, to said user; - a step (E2) of simultaneously acquiring a plurality of temporal biometric signals resulting from the user's interaction with the interactive program; - a step (E3) of extracting or calculating for each temporal biometric signal at least one feature relevant to the assessment of symptoms related to neurodevelopmental disorders or related mental health disorders in said child or adolescent; - assigning to the user one or more scores for one or more symptoms of one or more neurodevelopmental disorders or associated mental health disorders based on the extracted or calculated features and a machine learning model pre-trained on a set of annotated training data collected from pediatric users, the annotated data being annotated to determine the presence or absence of at least one symptom of at least one neurodevelopmental disorder and / or at least one associated mental health disorder, wherein the annotated data includes at least one severity level for the symptom of the diagnosed disorder (E4); A method comprising:

Images