COMPUTER-IMPLEMENTED METHOD AND APPARATUS FOR DETECTING AND PREDICTING THE NEURAL SIGNATURE OF DISORDERS OF EXECUTIVE FUNCTIONS

Abstract

The present relates to a computer-implemented method for detecting and predicting cognitive pathologies of the brain based on a Hold and Release process. The invention also concerns a device related to the claims method.

Claims

1. Computer-implemented method for detecting and/or predicting cognitive pathologies of the brain in a patient using the results of a Hold and Release (HR) process measured by an electroencephalography machine or other neurophysiological and brain imaging equipment and methods, wherein the Hold and Release process comprises a series of trials, each trial consisting in exposing the patient to a pair of successive stimuli, each stimulus meeting or not a specified criteria according to the following steps: a) specifying the patient which criteria shall meet a stimulus, b) exposing the patient to trials of the Hold and Release process in a serial manner, where c) if the 1st stimulus of a trial fulfills said specified criteria, the patient shall hold his/her attention to the subsequent stimulus of the current trial named Hold trial, and, if the 2.sup.nd stimulus also fulfils said specified criteria the trial is considered as a target-related trial whereas if the 2.sup.nd stimulus does not fulfil said specified criteria the trial is considered as a distracted-related trial, c) if the 1st stimulus of a trial does not fulfill said specific criteria, the patient shall release his/her attention and disregard the subsequent stimulus of the current trial named Release trial, wherein the computer-implemented method comprises the steps of d) providing electro-encephalographic event-related potentials (ERP) data recorded from the patient during a plurality of interspersed Hold trials and Release trials e) averaging the signal resulting from the target-related trials and from the distracter-related trials recorded in the Hold trials and in the Release trials respectively, f) establishing an amplitude vs. time pattern of ERP resulting from the signal averaging results, g) comparing the said amplitude vs. time patterns of ERP signal amplitude with an equivalent ERP amplitude vs. time pattern obtained in similar Hold and Release trials in a normative population of healthy participants named normative mean data, and h) determining a probability of cognitive decline and/or diagnosing cognitive pathology in the patient when, for the comparison between Hold and Release trials, the individual differential spatial temporal pattern of the ERP signal, including the global field power, over a continuous duration covering at least 30 ms of the said whole duration, differs significantly from the normative mean data, said significance consisting in individual values of the patient differing from the mean of the normative population by 2 standard-deviations and in at least 50% of the recorded ERP data.

2. Method according to claim 1, wherein steps c) to g) comprises establishing significant difference of ERPs between target-related and distracter-related trials in Hold condition in the patient and detecting as in h) the said difference being statistically set apart from a normative distribution.

3. Method according to claims 1, wherein steps c) to h) comprises repetitively measuring the patient's brain activity using Electroencephalography (EEG) signal other than ERPs, such as time-frequency analyses.

4. Method according to claim 1, wherein the brain imaging equipment and methods comprise MRI, MEG or near infrared spectroscopy.

5. Method according to claim 1, wherein the items are presented in a digitized format such as auditory, visual and the like.

6. Method according to claim 1, further comprising distributing over time both the trials according to named Stimulus Onset Asynchrony SOA, and the stimuli within a given trial according to named Inter Stimulus Interval ISI, as a function of the type of patients and the pathology tested.

7. Method according to claim 1, wherein the criteria are chosen among verbal, for instance a word or a pseudo-word, or non-verbal auditory or visual, or any other type of stimuli suitable to elicit HRA-related brain responses.

8. Device for detecting and predicting cognitive pathologies of the brain in a patient, the device comprising: i) a module for recording electro-encephalographic event-related potentials (ERP) data from the patient during a plurality of interspersed Hold trials and Release trials; ii) a module for averaging the signal resulting from the target-related trials and from the distracter-related trials recorded in the Hold trials and in the Release trials respectively; iii) a module for establishing an amplitude vs. time pattern of ERP resulting from the signal averaging results, iv) a module for comparing the said amplitude vs. time patterns of ERP signal amplitude with an equivalent ERP amplitude vs. time pattern obtained in similar Hold and Release trials in a normative population of healthy participants named normative mean data; v) a module for diagnosing cognitive pathology in the patient when, for the comparison between Hold and Release trials, the individual differential spatial temporal pattern of the ERP signal, including the global field power, over a continuous duration covering at least 30 ms of the said whole duration, differs significantly from the normative mean data, said significance consisting in individual values of the patient differing from the mean of the normative population by 2 standard-deviations and in at least 50% of the recorded ERP data.

9. Device according to claim 8, wherein at least one of said module is chosen amongst a processor, a computation module, display means, and an EEG collecting means.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Further particular advantages and features of the invention will become more apparent from the following non-limitative description of at least one embodiment of the invention which will refer to the accompanying drawings, wherein

[0030] FIG. 1 represents averaged ERPs recorded over centro-parietal region in healthy, elder and young, participants who participated in an experiment using an HRA variant. Important features of the HRA pattern are shown, the negative variation observed after the 1.sup.st item and the P3 observed after the 2nd item both in Hold condition (of note a first P3 is observed in any condition whether Hold or Release);

[0031] FIG. 2 illustrates the similarity of HRA-related effects on ERPs whether target or distracters are words or pseudowords in a variant consisting in a lexical decision task, therefore showing the domain-general nature of HRA.

[0032] FIG. 3 represents averaged ERPs recorded over centro-parietal region in an experiment using another HRA variant using semantic decision (as whether items are semantically related or not); here, the 65 elder subjects (mean age 74) who participated in HRA were divided in two groups as whether they show either stable verbal fluency performance over 2 consecutive assessments 5 years apart (N=28) or worsened performance (N=37). The average ERP curve recorded in the worsening participants shows, relative to the stable group, a striking decrease of the amplitude of the post-item1 Hold-related negative variation (that gets closer to the Release-related curve) and the same is true for the P3 observed in Hold condition following the item2.

DETAILED DESCRIPTION OF THE INVENTION

[0033] The present detailed description is intended to illustrate the invention in a non-limitative manner since any feature of an embodiment may be combined with any other feature of a different embodiment in an advantageous manner.

[0034] The HRA method of the present invention is a generic method as one can generate many different variants following the same principles while using varied type of stimuli (visual, auditory, and other sensory modalities), timing of presentation, type of responses such as manual, oral, oculo-motor; go-no-go versus A-B (obligatory) responses, in conventional or virtual reality environments (in the latter case, Hold responses evolve to gathering responses of virtual objects interesting to retain as a potential component of a target, while Release responses consist in discarding or throwing away irrelevant items), etc. all being within the scope of the present invention.

[0035] Stimuli are in either sensory modality, and HRA variants encompass a number of cognitive domains from various levels of language representations to memory processes as well as visual non-verbal perception (e.g. faces, natural or artefact objects, natural scenes) or other cognitive domains.

[0036] In HRA, target trials are in general defined by a specific perceptual or cognitive dimension of these items according to which items involved in these trials are categorized and unified; for instance, as exemplified above, targets may be pairs in which both the 1st and the 2nd item are real words, to be discriminated in turn from distracters; the latter are experimentally generated pseudo-words and may resemble existing words but do not qualify as an existing entry of the vocabulary of the participant's mother tongue.

[0037] Once instructed about target trials, i.e. those in which each item meets the required criteria, the participant proceeds with any trial in a serial manner, i.e. by categorizing the 1st item, as to the latter either fulfills criterion 1 or not.

[0038] If not (e.g. the 1st item is a pseudo-word), the participant can make readily his/her decision: the current trial, being for sure a non-target, is to be discarded; the participant can therefore release his/her attention and disregard the 2nd item: this case defines the Release condition.

[0039] In the case of Hold condition, the 1st item does meet criterion 1 (e.g. it is a word); focused attention should then be engaged and maintained to the point where the participant can address whether the subsequent item too meets the assigned criterion so that the current trial qualifies as a target.

[0040] Of note, while the below shown examples involve trials made of pairs, the same logic may be extended to longer stimulus sets (e.g. involving 3 or more items in series).

[0041] Coupled with brain responses recording (e.g. using Event-Related Potentials (ERP) from the EEG signal, or other techniques such as functional MRI, MEG, infrared multichannel recording etc.), HRA allows for exploring not only response accuracy and response time (RT), but also the time course and amplitude of processes-related neural signals, especially those relating to focused attention and working memory that either allow for stimulus-specific processing (Hold condition) or are withdrawn from it (Release condition).

[0042] For instance, in the variant of HR experiment involving a lexical decision task (with combinations of words and/or pseudo-words in pairs, as mentioned above) sustained negativity following the 1st item and the P3 wave following the 2nd item (FIGS. 1 and 2) were reliable markers of Hold condition.

[0043] The sustained negativity following the 1st item has shown to witness engagement of selective attention and expectancy of the next stimulus.

[0044] Further recent evidence from the Inventor show that P3 is likely related to updating and registration processes.

[0045] Release trials (or pairs) were instead associated with positivity following the 1st item and the absence of P3 following the 2nd item.

[0046] Effects are independent of the lexical status of items (i.e. the pattern was very similar whether the designated target was pairs of words or pairs of pseudowords), a finding that has been confirmed by further experiments by the Inventor and collaborators using various types of stimuli whether verbal or non-verbal. It illustrates that the main pattern of HRA (Hold-related negative variation and P3) is domain-general and independent from the type of stimuli to be processed.

[0047] These typical features were recorded predominantly in central electrodes over the skull vertex (Pz, Cz, FCz) as well as in the neighboring lateral electrodes in both cerebral hemispheres.

[0048] In a nutshell, HR allows one to investigate the behavioral and neurophysiological correlates of the alternating engagement (Hold condition) and disengagement (Release condition) of attention resources that are allocated to stimulus-specific processing depending on task requirements.

[0049] This alternation has a profound functional significance as it relies to the various functional regimes that the brain develops in an awakened life as we keep switching from one task to another. In daily life, after attention resources have been instantly engaged in a given task, a switch to a different task or target with higher priority may become soon necessary. It may also turn out that the current environment is of low interest and it is preferable to pause for a while with external stimuli and the attention is redirected to inner sources. Such a quieter or internally driven neurofunctional regime has been identified as the brain default mode in which external stimuli tend to be disregarded while mental resources are re-focused to inner representations. The ability to seamlessly shift from one regime to another involves a complex interplay between competing, large-scale functional neural networks e.g. between the default mode network and other networks specialized in processing stimuli from the external world (e.g. speech signal). This apt switching across higher order functions is affected by a number of brain diseases and especially by neurodegeneration in the early stages of dementing conditions such Alzheimer's disease.

[0050] The present HRA invention focus on the ability to recognize, monitor and possibly diagnose this subtle dysfunction regime in the human brain, for instance in the early stages of neurodegeneration in elderly individuals.

[0051] Recently, this framework was used to design HRA tasks suitable for elderly participants using both nonverbal and verbal stimuli.

[0052] By comparison to young adults, consistent results were obtained in ca 100 elder participants; the latter showed delayed reaction times and delayed ERPs relative to results obtained in young adults, while the typical HR pattern is preserved as shown in FIG. 1. For instance, FIG. 1 shows that the young adults have an average reaction time for Release of about 402 ms while for the elder ones average reaction time is 757 ms (see the R dots in the upperpart of FIG. 1). The same happens for the latency of P3 in Hold curve and the corresponding reaction times that are also set apart by ca. 300 ms in young relative to elder participants.

[0053] Further, preliminary evidence show that subtle deficits in attention and executive functions developing over time (i.e. ca 5 years) are reflected in behavioral measurements (Reaction Times) and, much more importantly, in HRA-generated ERPs so that the post-item1 sustained negativity and the P3 differ significantly in amplitude and latency, according to stability versus worsening of performance on attention-related tests such as verbal fluency tasks (see FIG. 3).

[0054] FIG. 3 clearly shows that upon HRA process, participants with stable performance over time show a profound negative variation when the first item meets the criteria, i.e. when they hold their attention, whereas in worsening participants in the same Hold condition, the amplitude of this component in much reduced when the first item meets the criteria, thereby suggesting a diagnosis of incipient brain degeneration.

[0055] More complex changes in ERPs are also associated with decline of other cognitive functions such as episodic memory. The invention covers the clinical use of all possible variants of the HRA paradigm mentioned above, as well as the use of advanced techniques for signal analyses and the use of complementary, multi-dimensional profiling that may increase the sensitivity and specificity of HRA for diagnosis and prognosis purposes.

[0056] This multi-dimensional profile involves

[0057] (i) other HRA-related features derived from changes in oscillatory EEG and MEG signal in a variety of frequency bands (from alpha to gamma),

[0058] (ii) The use of Artificial Intelligence techniques such as machine-learning (deep learning) and virtual-classifiers based analyses of HRA-generated neurophysiological signals

[0059] (iii) other brain imaging features (in the broadest sense) including functional connectivity analyses,

[0060] (iv) behavioral and cognitive phenotypic features that are associated with HRA, especially those related to disorders of attention, executive and memory functions in various brain diseases from developmental to neurodegenerative, traumatic, psychiatric and vascular stroke brain damage, and especially in the domain of the age-related cognitive decline (Alzheimer's disease (AD) and related dementing diseases). As regards cognitive phenotype, HRA involves especially post-hoc long term memory tests; these tests explore especially recognition memory test and may involve any type of item retrievable from any modality (visual, auditory, and others) of long-term memory (e.g. words, faces, artefactual or natural objects, landscapes and other scenes) that are used to form HRA item pairs (or longer stimulus sets in trials). Post-HRA recognition of the involved stimuli is explored as whether participants can distinguish between those stimuli they had to process (thereafter called OLD items) from never presented in HRA before (thereafter called NEW); performance measurements are based on this OLD/NEW recognition paradigm and otherwise derived behavioral features and allow for the exploration of the functional integrity of various parts of the mesial temporal cortex.

[0061] (v) specific genotype features (especially the genotype APOE4 associated with late-onset AD, as well as other genetic traits).

[0062] (vi) other peripheral physiological and biological markers (e.g. circulating biomarkers) that may complement and increase the diagnosis and prognosis accuracy of HRA-based method.

[0063] The HRA diagnosis and prognosis power is likely increased by deriving a unique profile involving HRA-derived characteristics combined with behavioral performance (for instance recognition memory performance) and genotypic features (e.g. including APOE4 for AD).

[0064] Notwithstanding other modalities of on-line brain activities recording (fMRI, functional near-infrared spectroscopy fNIRS), HRA provides mainly the user with typical changes in reaction times and Event-Related Potentials (ERPs) using surface (scalp) EEG recording under the engagement, maintenance and disengagement of attention that will allow the user to diagnose subtle or early disorders of executive, working memory and attention functions and to follow up longitudinal changes related to either clinically relevant, spontaneous changes of cognitive and behavioral abilities, as well as changes induced by any treatment (using drug compound or other therapeutic method) of the considered disorders.

[0065] HRA is a cost-effective screening tool to detect incipient age-related cognitive decline in asymptomatic volunteer participants that may be included in clinical studies including innovative treatments. Especially in the domain of the clinical neuroscience of ageing, by comparison to the existing methods, HRA is likely to be both more sensitive and easier to apply than conventional methods.

[0066] While a number of previous EEG studies pointed to the importance of P3 as a correlate of attention disorders in Mild Cognitive Impairment, these findings most likely do not capture early enough and with enough sensitivity changes that occur even before the MCI stage. We content that, compared to these previous studies, HRA bears additional diagnosis power for earlier deficit affecting attention and working memory capacities, especially the detection of changes in the post-1stitem sustained negativity.

[0067] Besides, PET scan and MRI were also used to tackle early-bird evidence of metabolic, structural or complex functional changes in the ageing brain. However, these techniques are, by far, less accessible, more invasive and dramatically more expensive than EEG. EEG/ERP signal recording necessary to reveal HRA is easy to obtain via only several electrodes (minimally one at vertex, and some in the periphery of the scalp) in a completely harmless way, so that wearable devices may be used for these measurements and the latter can be acquired virtually anywhere and in outside world conditions.

[0068] By comparison MRI and PET are heavy and highly expensive equipment that are in use only in relatively few medical Centers and University Hospitals; they use powerful electromagnetic fields and cannot be used easily in frequent medical conditions such as, for MRI, patients wearing cardiac pacemakers, vascular implants for MRI or for PET cannot be repeated easily owing to the limitation of allowed radioactivity doses. Finally, the cost of a standard MRI examination (ca. 800 ) is, at least, 16 times more expensive than an EEG and for PET the multiplicative factor is about 60 times.

[0069] Aside from its cost-effectiveness, HRA, a versatile experimental method, also allows researchers to explore in more details than with conventional ERPs, more complex neural correlates of attention-related fluctuations induced by Hold and Release conditions as a domain-general phenomenon; for instance, these neural signals consist of high-frequency oscillations and functional connectivity effects in the Alpha, Beta, Theta and Gamma bands explored with EEG or MEG, as well as their counterparts in functional MRI of near-infra red spectroscopy modalities such as changes in BOLD-like signals in task-induced and resting state conditions.

[0070] In addition, these changes in oscillatory regimes are expected to present with various topographical distributions, depending on the domain-specific effects tackled by specific variants of HRA, such as predominantly left-sided temporal distribution for language-related variants, right-sided temporal distribution for famous faces, or right-sided parietal in the case of visual-spatial variants; these and other specific distributions of signal changes are likely related to HRA, especially effects linked to loading of to-be-processed items in working memory and specific buffers depending on the type of contents and processes involved in variants of HRA. Such changes in HRA-induced oscillatory regimes are likely of diagnosis interest for specifically localized brain dysfunction, for instance verbal working memory effects in predominantly left-sided cortical lesions.

[0071] More in general, the invention comprises changes in Electroencephalographic (EEG),

[0072] Magnetoencephalographic (MEG), fMRI or near-infrared spectroscopy signals related to HRA that will be demonstrated using machine learning and deep learning programs. These machine-learning methods allow for deciphering, at the single-individual and the single-event levels, neural responses related to either engagement or release of attention, as well as neural events related to memory load or retrieval associated with HRA. The invention covers both computer-implemented programs and hardware devices that are derived from the above described methods related to HRA; the user can therefore be provided with results significant in each and single subject who undergoes an HRA test following the use of software or hardware that are specific to the Invention. These individual results inform about attention and memory performance that together with other individual results such as genetic and non-genetic biological, cognitive or imaging factors are indicative of risk of brain diseases affecting cognition, e.g. neurocognitive disorders.

[0073] In sum, HRA consists in a new, harmless, cost-effective and widely useable neurofunctional tool to detect early evidence of clinically significant cognitive decline and to follow up these changes including objective evidence of positive effects of innovative treatments. These features make HRA a highly interesting method for any clinical team in the field of clinical neuroscience as well as any industrial company (e.g. pharmaceutical companies) that aim to monitor the effects of experimental treatments in an efficient and affordable way.

[0074] HRA applies to the diagnosis and follow up of disorders of executive functions, attention and working memory in humans in various clinical populations, especially in elder subjects affected by, or complaining about, age-related cognitive disorders. HRA can also be used as a surrogate marker of changes in the brain metabolism as a consequence of therapeutic intervention in the broadest sense (whether pharmacological or not). Other applications may be developed, also for monitoring the neural counterparts of therapeutic interventions, in younger patients be they affected by e.g. TBI, strokes, depression, neurodevelopmental pathologies, such as major psychiatric syndromes (e.g. schizophrenia), and in children affected by developmental learning disabilities, e.g. attention disorder or dyslexia.

[0075] While the embodiments have been described in conjunction with a number of others, it is evident that many alternatives, modifications and variations would be or are apparent to those of ordinary skill in the applicable arts. Accordingly, this disclosure is intended to embrace all such alternatives, modifications, equivalents and variations that are within the scope of this disclosure. This for example particularly the case regarding the different apparatuses which can be used.