method

Abstract

The present invention provides a method for detecting a current or previous respiratory tract infection, the method comprising analysing data collected from oculography of a subject performing one or more of a pro-saccade task, an anti- saccade task, a visual search task and a visual world paradigm test to determine values for said one or more pro-saccade task, anti-saccade task, visual search task and visual world paradigm test for said subject; and based on the values determined for said subject, detecting the presence of a current or previous respiratory tract infection. Computer implemented methods and computing apparatus for detecting a current or previous respiratory tract infection in a subject are also provided.

Description

[0001] Method

[0002] Field of Invention

[0003] The present invention relates to method for detecting a current or previous respiratory tract infection.

[0004] Background to the Invention

[0005] Ocular-motor movements involve the muscles that move the eye in different directions. These movements are controlled by the brain and can be affected by neurological conditions. Tracking of eye gaze is used to study multiple aspects of language processing including language comprehension, language production and individuals' ability to process cognitive-linguistic information.

[0006] Quantitative assessment of video oculography has become a very attractive means to assess neurological and psychological responses because: 1) the method is simple and non-invasive, subjects look at a computer screen while a camera records the eye; and 2) knowledge of cortical and subcortical brain circuits is extensive so that deficits can be mapped to specific brain circuits. These include regions of the visual system, parietal and frontal cortex, basal ganglia, thalamus, superior colliculus, cerebellum, and brainstem. Therefore, the eye movement system can be used as an effective tool to probe sensory, motor, and cognitive function to search for specific biomarkers of disease.

[0007] Summary of the Invention

[0008] Eye tracking technology has been used in the diagnosis of conditions including attention deficit hyperactivity disorder (ADHD), autism spectrum disorder (ASD), glaucoma, macular degeneration, diabetic retinopathy, myasthenia gravis, Parkinson's disease and Alzheimer's disease. The present inventors have realised that neurological conditions affecting ocular-motor movements can sometimes be triggered by infectious diseases. The connection between infectious diseases affecting the respiratory tract and ocular motor movements has been demonstrated herein using eye-gaze technology.

[0009] The present invention provides a method for detecting a current or previous respiratory tract infection, the method comprising measuring performance of one or more of a pro-saccade task, an anti-saccade task, a visual search task and a visual world paradigm test by a subject using oculography; analysing data collected from the oculography to determine values for said one or more prosaccade task, anti-saccade task, visual search task and visual world paradigm test for said subject; and based on the values determined for said subject, detecting the presence of a current or previous respiratory tract infection.

[0010] In embodiments of the invention, the method may comprise analysing data collected from oculography of a subject performing one or more of a pro-saccade task, an anti-saccade task, a visual search task and a visual world paradigm test to determine values for said one or more pro-saccade task, anti-saccade task, visual search task and visual world paradigm test for said subject; and based on the values determined for said subject, detecting the presence of a current or previous respiratory tract infection. In other words, the method does not have to include the active step in which the subject performs the pro-saccade task, antisaccade task, visual search task and / or visual world paradigm test. The method may be a computer implemented method.

[0011] The method of the present invention allows for a quick and non-invasive diagnosis of a current or previous respiratory tract infection. In particular, performing video oculography using a remote eye tracking system allows the method to be performed without requiring any physical contact between the subject, the apparatus and / or any person facilitating the video oculography. Remote eye tracking is also effective for individuals without significant head control, meaning that the method can be used for individuals with neuromuscular disabilities.

[0012] The present invention provides a method for detecting a current or previous respiratory tract infection, the method comprising: measuring performance of one or more of a pro-saccade task, an anti-saccade task, a visual search task and a visual world paradigm test by a subject using oculography; analysing data collected from the oculography to determine values for said one or more prosaccade task, anti-saccade task, visual search task and visual world paradigm test for said subject; and based on the values determined for said subject, detecting the presence of a current or previous respiratory tract infection. Optionally, the method does not include an active step of measuring the subject's performance and instead begins by analysing data previously collected from oculography of a subject performing one or more of a pro-saccade task, an anti-saccade task, a visual search task and a visual world paradigm test to determine values for said one or more pro-saccade task, anti-saccade task, visual search task and visual world paradigm test for said subject; and based on the values determined for said subject, detecting the presence of a current or previous respiratory tract infection.

[0013] The control of ocular motor responses involves several components of the nervous system to achieve precise eye movements. These movements are essential for vision, enabling us to focus on stationary objects while we or our environment are moving, and to swiftly shift our gaze from one object to another.

[0014] The primary types of eye movements include saccades, smooth pursuit, vergence, and vestibulo-ocular reflex, each serving different purposes and being controlled by different neural mechanisms.

[0015] Saccades are rapid, ballistic movements of the eyes that abruptly change the point of fixation. They can be voluntary, as when looking from one object to another, or reflexive, as when reacting to a new stimulus. A pro-saccade task requires a subject to look toward a target and an anti-saccade task requires a subject to look away from a target. The control of saccades involves the frontal eye fields (FEF), parietal eye fields, superior colliculus, and the brainstem saccadic generator, which includes the paramedian pontine reticular formation (PPRF) and the rostral interstitial nucleus of the medial longitudinal fasciculus (riMLF).

[0016] The interaction between the auditory and visual systems in controlling eye movements underscores the brain's ability to integrate sensory information from different modalities to guide behaviour. This multisensory integration is crucial for navigating and understanding complex environments, allowing for more effective responses to a wide range of stimuli. The stimulation of ocular motor responses with sounds involves complex interactions between the auditory and visual systems, primarily mediated through reflexive pathways and higher-order cognitive processes. While sound itself does not directly trigger the classic ocular motor responses involved in tracking visual objects, it can influence eye movements through several mechanisms:

[0017] Auditory-Visual Integration: The brain integrates auditory and visual stimuli to enhance spatial awareness and orientation. Sounds can cue the direction of potential visual targets, prompting anticipatory or reflexive saccades toward the source of the sound. This integration is crucial in environments where visual information might be limited or ambiguous but is complemented by auditory cues.

[0018] Auditory-Triggered Reflexive Saccades: Some studies have shown that unexpected auditory stimuli can trigger reflexive saccades towards the direction of the sound, especially in situations where the sound could signify an important or relevant event. This response likely involves the superior colliculus, which plays a critical role in integrating multisensory information and initiating saccadic eye movements.

[0019] Cognitive Processes and Attention: Sounds can also influence ocular motor responses through cognitive channels, particularly by altering attention and expectation. For example, if a sound signals the appearance of a visual target or the need to shift attention to a different location, it can modulate the timing and accuracy of saccades.

[0020] A visual world paradigm (VWP) test is specific to a time dependent response. A visual search test is similar but does not include the time dependent element. Both are characterised by an auditory stimulus as compared to the visual stimulus in the pro- and anti-saccade tasks.

[0021] Oculography refers to the measurement and recording of the position and movement of the eyes. Oculography can utilise an eye tracking system that is head-stabilised, remote, mobile (head mounted) or embedded (integrated). Preferably the eye tracking system used in the methods described herein is remote or mobile.

[0022] Suitable oculography for use in the methods of the present invention includes video-oculography (VOG). VOG is a non-invasive, video-based method of measuring horizontal, vertical and, optionally, torsional position components of the movements of one or both eyes (eye tracking). A VOG system typically includes a video camera, which communicates images to a computer for image processing. Most VOG systems use infrared light and thus can function in complete darkness. VOG systems may be monocular or binocular. The VOG system may have a frame rate of about 50 Hz to about 250 Hz, or about 50 Hz to about 200 Hz, or about 50 Hz to about 100 Hz. Optionally, the VOG system has a frame rate of about 50 or about 60 Hz. Preferably the video camera and the computer screen can be run at the same frame rate. The oculography may include pupil center corneal reflection and / or infrared oculography. Pupil center corneal reflection captures an image of the eye, which is then used to identify the pupil center and the reflection of the illuminators on the cornea. The pupil's position and the illuminators' reflections are used to calculate the participant's gaze. Infrared oculography (IROG) is a non-invasive method that uses an infrared sensor validated to define time of foveation (time to fixation), an indirect measure of visual acuity.

[0023] The method of the present invention may additionally comprise the subject performing one or more calibration tasks prior to performing the one or more prosaccade task, anti-saccade task, visual search task and visual world paradigm test. For example, the calibration task may comprise the subject tracking movement of a dot across a screen.

[0024] When analysing the data collection from the oculography, the subject's performance of the one or more pro-saccade task, anti-saccade task, visual search task and visual world paradigm test can be compared with values determined for healthy control subjects. In other words, the subject's performance can be compared with the performance of the same one or more activities by a healthy, age and / or gender matched control population. The matched control population can be defined based on, for example, health status, age or any other characteristic that can affect the performance of the one or more tasks. Preferably the subjects within the control population do not have a current or previous respiratory tract infection. Once the performance of the control population is known, the measured performance of the subject can be compared, and the significance of any difference can be determined, for example, by using standard statistical methods.

[0025] The subject is preferably a human and may be of any age or a paediatric or a geriatric subject. In embodiments of the invention the subject may be between the ages of 5 years and 99 years, or between the ages of 7 years and 90 years, or between the ages of 10 years and 80 years. Preferably, the subject does not suffer from significant hearing loss and / or blindness.

[0026] Preferably the subject performs (i) at least one pro-saccade task and / or at least one anti-saccade task; and (ii) at least one visual search task and / or at least one visual world paradigm test. Using a combination of the two types of tasks, i.e., at least one visual stimulus task (i) and at least one auditory stimulus task (ii), allows both reflexive and cognitive processes to be tested, meaning that multiple brain responses are assessed for the detection of an infection.

[0027] The pro-saccade task can measure time to first fixation and / or pupillary response. Subjects having a current or previous respiratory tract infection may demonstrate a time to first fixation that is faster than that of a healthy control subject, and / or a pupillary response that is greater than that of a healthy control subject. For example, a pupil of subjects having a current or previous respiratory tract infection may widen about 20% to about 40% more, or about 30% more, than that of subjects who do not have a current or previous infection. Pupillary response can be measured using the right and / or left pupil of the subject. The time to first fixation for subjects with a current or previous respiratory tract infection may be about 2% to about 15%, or about 4% to about 11%, or about 7% faster than that of subjects who do not have a current or previous infection.

[0028] The visual world paradigm test can comprise measuring the number of gazes on a target image compared to the number of gazes on a distractor image. Typically, subjects having a current or previous respiratory tract infection may achieve fewer gazes on the target image than a healthy control subject. For example, the number of gazes achieved by subjects having a current or previous respiratory tract infection may be reduced by about 5% to about 30%, or about 5% to about 15% or about 10%, as compared to healthy control subjects. The number of gazes can be measured in quadrants, which are typically identified as upper right, lower right, upper left and lower left. The number of gazes achieved by subjects having a current or previous respiratory tract infection may be reduced in all quadrants, with typical percentage reductions as outlined above.

[0029] In embodiments of the invention, the method may comprise measuring performance of time to first fixation, pupillary response and a visual world paradigm test. In some embodiments, the method may comprise detecting a current or previous respiratory infection from one or more oculography data comprising an increased time to fixation, increased pupil widening, and decreased visual search. Increased pupil widening may be measured by the change in pupil diameter. The data collected from the oculography may be analysed using a multimodal artificial intelligence (Al) model. Multimodal Al may be defined herein as "Al that can understand, interpret, process, and generate information from multiple types of data inputs, or "modes"". In other words, multimodal Al can be used to combine the data as derived from one or more saccadic paradigms (e.g., time to first fixation and / or pupillary response) and one or more visual world paradigms / visual search tests and detect specific patterns allowing for the differentiation between subjects having a current or previous respiratory tract infection and those who are uninfected.

[0030] Multimodal Al provides advantages including allowing for integration (i.e., combining data from different modalities and facilitating a process to synchronize and align data from different sources), processing (i.e., employing specialized Al models that are capable of handling specific types of data and integrating their outputs for comprehensive analysis) and contextual understanding (i.e., analysing the context in which data appears).

[0031] Types of algorithms that may be used in the multimodal Al include machine learning algorithms, such as supervised learning (i.e., algorithms that learn from labelled training data, making predictions based on that data) and unsupervised learning (i.e., algorithms that learn from unlabelled data, finding hidden patterns or intrinsic structures in input data); deep learning algorithms such as convolutional neural networks (i.e., a subset of machine learning that uses neural networks with many layers and which can be used for handling very large datasets and complex problems); bayesian algorithms (these algorithms involve statistical inference to make predictions); decision tree algorithms, such as random forests (these algorithms can be used for classification and regression tasks); and dimensionality reduction algorithms (which can be used to reduce the number of variables under consideration and to extract relevant information from large datasets).

[0032] The respiratory tract infection may be an upper respiratory tract infection. An upper respiratory tract infection (URTI) is an illness caused by an acute infection, which involves the upper respiratory tract, including the nose, sinuses, pharynx, larynx or trachea. This commonly includes nasal obstruction, sore throat, tonsillitis, pharyngitis, laryngitis, sinusitis, and / or otitis media. Most infections are viral in nature, and in other instances, the cause is bacterial. URTIs can also be fungal or helminthic in origin, but these are less common.

[0033] The respiratory tract infection may be a current infection or a previous infection within six months preceding the test, or an infection within three months preceding the test, or an infection within one month preceding the test.

[0034] The respiratory tract infection may be a viral infection or a bacterial infection. Certain viruses can affect the brain areas as well as the nerves and muscles responsible for controlling eye movements. This can lead to abnormalities in ocular motor function. Some bacterial infections can affect the nervous system, including nerves that control eye movements, resulting in ocular motor dysfunction. Optionally, the respiratory tract infection may be a parasitic infection. Parasites that affect the central nervous system can potentially impact ocular motor function due to their effect on the brain and nervous system. In some cases, the body's immune response to an infection can inadvertently target the nervous system, affecting areas that control eye movements.

[0035] The viral infection may be a coronavirus infection. Coronaviruses are RIMA viruses that cause respiratory tract infections including the common cold, SARS, MERS and COVID-19. Coronaviruses including HoV-NL63, SARS-CoV and SARS-CoV-2 are known to bind to angiotensin-converting enzyme 2 (ACE2) receptors. ACE2 is widely expressed in the central nervous system (CNS). Structural similarities found between SARS-CoV-2 and other coronaviruses have indicated that it might have neuroinvasive and neurotropic potential, with studies of other coronaviruses showing serious CNS complications. The coronavirus infection may be a Sars-CoV- 2 infection.

[0036] The bacterial infection may be a Pseudomonas infection (e.g., Pseudomonas aeruginosa), a Staphylococcus infection (e.g., Staphylococcus aureus), a Mycobacterium infection, Streptococcus infection, or a Burkholderia infection (e.g., Burkholderia cepacia complex).

[0037] Optionally, the methods described herein may comprise treating the respiratory tract infection. For example, the subject may be treated with an effective amount of one or more antiviral, antifungal, anthelmintic, antibacterial and antiparasitic medications. Optionally, the subject may be treated with an effective amount of one or more of antiviral, antibacterial and antiparasitic medications. In some embodiments, the subject may be treated with an effective amount of one or more of nirmatrelvir, ritonavir, sotrovimab, molnupiravir, remdisavir, oseltamivir, zanamivir, peramivir, and baloxavir marboxil.

[0038] The methods described herein for detecting a current or previous respiratory tract infection in a subject may comprise obtaining oculography data from the subject, the oculography data comprising one or more of a pro-saccade task, an antisaccade task, a visual search task and a visual world paradigm test. The oculography data may be collected in response to visual and / or auditory tests or stimulation as described herein. The method may further comprise detecting whether a current or previous respiratory tract infection is present from the oculography data. When a current or previous respiratory tract infection is detected, the methods may optionally further comprise treating the current or previous respiratory tract infection with an effective amount of an antibacterial, antiviral and / or anti-parasitic medication. The methods described herein may be computer-implemented methods.

[0039] The computer-implemented methods described herein may comprise: a means for acquiring the oculography data from the subject; and a means for calculating one or more values from the oculography data and comparing said one or more values to one or more predetermined values to detect a current or previous respiratory tract infection from the subject. The method may further comprise diagnosing a current or previous respiratory tract infection in the subject. The method may also comprise a means for reporting a current or previous respiratory tract infection in a subject. In some embodiments, the means for acquiring oculography data may comprise one or more of means for detecting an increased time to fixation, increased pupil widening, and decreased visual search from said oculography data. Pupil widening may be detected by the increase in pupil diameter.

[0040] The present invention also provides an apparatus for detecting a current or previous respiratory tract infection, comprising a processor configured to perform the methods described herein. The apparatus may further comprise a display, a non-transitory computer-readable storage medium and a video based eye tracker for acquiring oculography data. The present invention also provides a non-transitory computer-readable storage medium for storing instructions executable by a processor to enable an apparatus to perform the method of any of the methods described herein.

[0041] Brief Description of the Drawings

[0042] The invention will now be described in detail, by way of example only, with reference to the figures.

[0043] Figure 1 shows a side view and top view configuration of the VOG system, i.e., the remote eye-gaze testing equipment used in the examples. (1) demonstrates the vertical angle in which the camera can capture a participant's pupils and (2) the angle in which the participant views the screen on which stimuli are projected. The distance indicated (50cm) is the minimum distance between participant and screen specific to the equipment used in the examples. (3) indicates the space in which the participant's head can move without the camera losing connection with the participant's eyes. (4) demonstrates the horizontal angle in which the camera can capture a participant's pupils

[0044] Figure 2 shows the data output along the X-axis from the eye-gaze track eye from onset of visual stimulus until the moment the stimulus disappears from the screen where (1) depicts a failed test and (6), (11) and (16) depict successful tests. Each dot represents a gaze capture from the eye-tracking camera which in this example was operating at 60Hz. This data is used to calculate the time to first fixation. Furthermore, the rate of movement from onset of stimulus to the gaze entering the area of interest is calculated (from left to right in the image) as part of the algorithm to determine infectious state of the subject.

[0045] Figure 3 shows the data output along the Y-axis from the eye-gaze track eye from onset of visual stimulus until the moment the stimulus disappears from the screen where (1) depicts a failed test and (6), (11) and (16) depict successful tests. Each dot represents a gaze capture from the eye-tracking camera which in this example was operating at 60Hz. This data is used to calculate the time to first fixation. Furthermore, the rate of movement from onset of stimulus to the gaze entering the area of interest is calculated (from left to right in the image) as part of the algorithm to determine infectious state of the subject. Figure 4 shows the data output of the pupillary response of the right eye from onset of visual stimulus until the moment the stimulus disappears from the screen where (1) depicts a failed test and (6), (11) and (16) depict successful tests. Each dot represents a gaze capture from the eye-tracking camera which in this example was operating at 60Hz. This data is used to calculate the changes in pupillary response.

[0046] Figure 5 shows an example of a visual world paradigm stimulus where the subject is asked to listen to an instruction and predict the outcome in a set interval of time. In this example, the sentence is "The boy will bounce" and the anticipated response is "the ball" - for which the subject's response will be determined.

[0047] Figure 6 shows the time (in seconds) to first fixation in infected or previously infected subjects (Covid-19) and healthy (Non Covid-19) subjects. Differences in time to first fixation are significantly more pronounced on the right-hand side of the field of vision than the left.

[0048] Figure 7 shows differences in pupillary diameter in COVID-19 positive subjects as compared to uninfected (Non Covid-19) controls as measured 0.16s before onset of the stimulus and at time to first fixation on the stimulus.

[0049] Figure 8 shows the distribution of gazes onto the target and distractors by (A) Covid-19 positive (Covid-19), and (B) healthy (Non Covid-19) subjects.

[0050] Figure 9 shows a flow diagram of a computer implemented method that may be implemented by a software program for performing the diagnostic method for detecting the presence of a current or previous respiratory tract infection in a subject.

[0051] Detailed Description

[0052] Figure 9 is a flow diagram of a computer implemented method 100 that may be implemented by a software program for performing the diagnostic method for detecting the presence of a current or previous respiratory tract infection in a subject. The method, in an embodiment is executed on the apparatus shown in Figure 1. In further embodiments, the method can be executed on any device comprising a processor, a non-transitory storage medium, a display and a video based eye tracker.

[0053] The subject is typically seated at I positioned at a system suitable for acquiring oculography data, such as the VOG system as shown in Figure 1. The computing device comprises a number of software modules, each module configured to execute one or more of the visual and / or auditory or simulation tests as described with reference to Figures 1-8. Each of the tests may be implemented in the manner using the systems described with reference to Figures 1 to 8. In further embodiments, one or more of the individual tests may be performed using any known commercially available system for eye tracking, measuring pupil dilation, determining gaze as appropriate.

[0054] In use the software program loads a visual and / or auditory test or stimulation tests onto the system for a subject to perform. The visual and / or auditory test or stimulation tests can be a pro-saccade task, anti-saccade task, visual search task, visual world paradigm test or any other suitable visual and / or auditory test or stimulation tests. These tests are described with reference to Figures 1-8.

[0055] The system suitable for acquiring oculography data comprises a means for acquiring oculography data from the subject, such as a video based eye tracker. The video based eye tracker can be the VT3 mini eye tracker or any other suitable eye tracking device. Such eye tracking software is known in the art and any suitable commercially available software may be used. In addition, the means for acquiring oculography data may comprise one or more of means for detecting an increased time to fixation; increased pupil widening; and decreased visual search from said oculography data. The means for detecting an increased time to fixation; increased pupil widening; and decreased visual search from said oculography data may be any suitable known commercially available software.

[0056] The oculography data comprises performance of one or more of a pro-saccade task, an anti-saccade task, a visual search task and a visual world paradigm test. Oculography data of the subject may be collected from the means for acquiring oculography data in response to the subject performing a visual and / or auditory test or stimulation as described herein.

[0057] Therefore, the method is implemented by one or more software modules configured to perform the tests described in Figures 1-8.

[0058] At step 110, there is an optional calibration step, which allows for the video based eye tracker to be calibrated at the outset of the data collection. The calibration procedure can comprise a 5-point calibration procedure which requires the subject to track the movement of a red dot, with their eyes, as the dot moved across the screen of a display, taking in four corners and the centre of the screen. Successful calibration is defined as tracking of both pupils by the video based eye tracker for a minimum of 80% of the calibration period. The eye tracker records the pupil trajectory during the calibration process allowing for the calculation of pupil trajectory for each position on the screen during the data collection.

[0059] At step 120, after the subject has completed the assigned task, the software program will collect the oculography data, from the means for acquiring oculography data, of the user taking the test. This collected data can be stored in a memory for later analysis and the software program will load the next test on to the system, or the collected data will be analysed by the software program, as described in step 120. The data may be stored locally on a writeable memory, or stored remotely in a cloud server.

[0060] At step 130, the software program will analyse the collected oculography data. This analysis can comprise calculating one or more values from said oculography data and comparing said values with one or more values related to predetermined oculography data.

[0061] The predetermined oculography data comprises oculography data for one or more of the visual and / or auditory tests or stimulation tests performed by a matched control population. The matched control population can be defined based on, for example, health status, age, gender or any other characteristic that can affect the performance of the one or more tasks. Preferably the subjects within the control population do not have a current or previous respiratory tract infection. The predetermined oculography data may comprise values measured during a pro- saccade task, such as time to first fixation and pupillary response. Alternatively, or in addition to, the one or more values from said oculography data may comprise values measured during a visual world paradigm test, such as number of gazes on a target image compared to the number of gazes on a distractor image.

[0062] The one or more values from the collected oculography data may comprise values measured during a pro-saccade task, such as time to first fixation and pupillary response. Alternatively, or in addition to, the one or more values from said oculography data may comprise values measured during a visual world paradigm test, such as number of gazes on a target image compared to the number of gazes on a distractor image. Preferably, to allow for a simplified comparison the data collected is the same as the data collected for the control population thus allowing for an easy comparison to be made. Where the data is the same as the data collected in the control population, a simple comparison between the values is made.

[0063] These values are compared with corresponding values from the predetermined oculography data to determine if the subject has a current or previous respiratory tract infection. For example, subjects having a current or previous respiratory tract infection may demonstrate a time to first fixation that is faster than that of a healthy control subject, and / or a pupillary response that is greater than that of a healthy control subject. In another example, subjects having a current or previous respiratory tract infection may achieve fewer gazes on the target image than a healthy control subject.

[0064] The significance of any difference these values can be determined by the computer program, for example, by using standard statistical methods or any suitable computational method. For example, the percentage difference between the collected values, relating to the subject, and the predetermined values can be calculated. In further examples statistical analysis may be used, for example determining a standard deviation of a result of a user from the control population. The data stored locally on writeable memory may be transmitted to a device remote to the apparatus or a device remote to the apparatus may access the data stored remotely in a cloud server. The remote device can then carry out the methods according to steps 130 and 140 of Figure 9. At step 140, based on the comparison of the one or more values from the user's oculography data and the one or more values of the predetermined oculography, the computer program will detect the presence of a current or previous respiratory tract infection in the subject. For example, using the percentage difference calculated in step 130, if a certain threshold value is exceeded then it is determined that the subject has a current or previous respiratory tract infection. For example, an increase of more than 30% of pupil widening may be used as a threshold value. In further examples, other threshold values described previously may be used dependent on the type of test used.

[0065] If other statistical measures, for example the standard deviation of the result from the test population, other thresholds may be applied. Other known suitable statistical or computational methods can be applied to the data to identify the presence of a current or previous respiratory tract infection in the subject.

[0066] Optionally, the system may comprise means for reporting a current or previous respiratory infection in a subject and the computer program may instruct the system to report the presence of a current or previous respiratory tract infection in the subject.

[0067] In a further embodiment step 140 is performed using a trained machine learning model.

[0068] In such implementations the system uses prediction models or machine learning models. The skilled person will appreciate that, even if not expressly stated, a prediction model or machine learning model can be trained using standard training approaches as known in the art. For example, a standard approach for training a prediction model comprises obtaining relevant training data (e.g., using known data sources, databases, or data sets) and performing cross-validation to train the prediction model on the training data. Typically, cross-validation involves splitting the training data into K-folds (approximately equal partitions or sets of the training data) and withholding a single fold as a test set and, one by one, using one of the remaining folds as a validation set and the remaining K-2 folds as a training set. The model is then repeatedly trained on the training set using different model hyperparameters and the performance validated on the validation set. Once the best performing hyperparameters are obtained, the model trained according to the best hyperparameters are evaluated on the test set. For training classification models, the cross-validation strategy can be stratified such that the proportion of training instances within each category or class is approximately the same across each fold. Model hyperparameters can be selected using any suitable approach such as grid search or randomized search. Model performance can be estimated using any suitable performance measure and is dependent on the type of model being trained (e.g., mean square error for regression, binary cross entropy for classification, ranking loss for ranking, etc.).

[0069] In an embodiment the training data is a labelled data set, with the labels being indicative of whether or not the patient had a respiratory tract infection or not. In further embodiments, the label indicating the type of respiratory tract infection is used. The training data also preferably includes physiological and medical data such as age, sex, known medical conditions, weight etc. The data further includes the results one or more of the visual and / or auditory test or stimulation tests as well as an identification of the test.

[0070] The model in an embodiment is trained using any suitable supervised training methods. Examples of suitable techniques include classification techniques (to identify the absence or presence of the respiratory tract infection). In an embodiment logistic regression used to determine the presence, or absence of the respiratory tract infection. In a further embodiment a k-nearest neighbours technique is used to classify a result as being indicative of a respiratory tract infection.

[0071] The trained model therefore provides a model in which a prediction can be made as to whether a person has a respiratory tract infection based on their physiological data and the results of the test. In use the model is stored on the device, for example as the one shown in Figure 1, or remotely. When one or more tests are performed, the results are inputted to the model and a classification of the result(s) being indicative of a respiratory tract infection or not is outputted.

[0072] Examples

[0073] EXAMPLE 1:

[0074] In May 2023 the World Health Organisation declared that the disease (COVID-19) caused by the Severe Acute Respiratory Syndrome Coronavirus was no longer a Public Health Emergency of International Concern. There have been significant follow up studies on the effects of the disease, such as the Bowe et al (2023) treatise on the sequelae of COVID-19 and the Iwashyna et al (2023) paper on mortality amongst US Veterans.

[0075] Although it was initially detected through its impact on the respiratory system, COVID-19 is now recognized as a multisystemic disease, and the nervous system is one of its main targets. SARS-CoV-2 infects host cells through the binding of its spike protein to a surface receptor known as angiotensin-converting enzyme 2 (ACE2), which is widely expressed in the CNS. Structural similarities found between SARS-CoV-2 and other coronaviruses have indicated early on that it might have neuroinvasive and neurotropic potential, with studies of coronaviruses involved in previous outbreaks showing serious CNS complications.

[0076] We hypothesized that individuals who were infected with or had previously been infected with SARS-CoV-2 would perform significantly worse than healthy controls on all saccadic paradigms.

[0077] Methods

[0078] 28 healthy (Covid-19 negative) subjects and 25 subjects infected or previously infected with SARS-Cov-2 (Covid-19 Positive) were recruited, as shown in Table 1.

[0079] Table 1: Test Population: The physical arrangement of the data collection apparatus is illustrated in Figure 1.

[0080] Participants were seated facing a table with the remote eye-tracking equipment in front of them. Table and chair height were adjusted as necessary to ensure a consistent connection between the VT3 Mini eye-tracker and a participant's pupils. The venue was brightly lit both by natural light and internal overhead LED lighting. Ambient light conditions were monitored, and adjustments made to ensure consistency for optimal functioning of the VT3 Mini Eye-Tracker.

[0081] We used a pro-saccade task to invoke the prepotent characteristic of the oculomotor system by requiring participants to make a saccade, as quickly as possible, to the location of an abrupt onset peripheral target.

[0082] Video based eye tracking was conducted using a VT3 mini Eye Tracker at a 60 Hz recording rate at an Accuracy of 0.5°. All visual stimuli were shown on a 15-inch lap-top screen with a 1980 x 1280-pixel resolution The tracking distance was between 50 and 70 cm.

[0083] Eye tracking data was stored using the MangoldVision software package which allows for the setting up of the video-based stimuli, setting of Areas of Interest (Aol) and post data collection primary analysis. Raw data was further exported as .CSV files for further analysis.

[0084] Participants wore glasses or contact lenses for the duration of this eye-tracking task if these were usually worn to correct for reading purposes or when carrying out work on a computer screen.

[0085] The VT3 Mini Eye Tracker was calibrated at the outset of the data collection. A 5- point calibration procedure was used to do this, requiring the participant to track the movement of a red dot with their eyes as the dot moved across the screen of the laptop, taking in the four corners and the centre of the screen. Successful calibration was defined as tracking of both pupils by the VT3 Mini Eye Tracker for a minimum of 80% of the calibration period. The eye tracker recorded the pupil trajectory during the calibration process allowing for the calculation of pupil trajectory for each position on the screen during the experimental process. All participant records were anonymised in accordance with the GDPR regulations in force in the United Kingdom at the time of testing. Saccadic performance was compared between the two groups of subjects: infected or previously infected with SARS-Cov-2 (Covid-19 Positive); and healthy (Covid-19 Negative).

[0086] Test stimuli for Experiment 1 were conceptualised within a covert attention paradigm as proposed by Posner (1980) in the Cued Attention framework. Circles were 1.5 cm in diameter. Blue was preferred rather than green and red in order to create sufficient contrast with the background whilst avoiding any potential complications from colour blindness in participants. Circles were placed at 45 degrees to the central cross and at approximately 80% of the distance between the centrally placed cross and the corner of the page on the screen. This was done to achieve symmetry and allow adequate eye movement to be tracked and monitored without impeding eye movement.

[0087] Areas of interest (Aol) were set in the MangoldVision Analyzer module and marked as per Hessels, et al (2016). Use was made of Grid Aol marking to reduce subjectivity in data capture (Hessels et al 1026). Aols set for 0.5s from appearance of blue dot until its disappearance.

[0088] Data were analysed in the MangoldVision Analyser software package.

[0089] Results

[0090] NT COVID vs NO COVID - fatigue:

[0091] Task-related fatigue did not affect saccadic performance, as there was no difference between errors in the first and second half of the saccadic tasks in either group. First vs Second Task ANOVA Single Factor No COVID p=0.9410

[0092] COVID p=0.9471

[0093] NT COVID vs NO COVID - time to first fixation:

[0094] Two Tailed Two Sample T Test assuming equal variances p=0.0003

[0095] NT COVID vs NO COVID - Left vs Right Fixation:

[0096] NT CO VID Non-COVID

[0097] Left .25 .23 Two Tailed p=0.0007

[0098] Right .26 .24 Two Tailed p=0.08

[0099] Differences in time to first fixation are significantly pronounced on the right-hand side of the field of vision than the left (Figure 6).

[0100] The slower reaction time of the Non-Covid / Covid-Negative population is in line with research indicating that people who read left-to-right tend to have a spatial attention bias in the same direction. This means they might pay more attention to things presented on the left side of a visual field first, as their attention moves in the direction in which they usually read (Su et al., 2020).

[0101] This data suggests that for people reading right-to-left will have an inverse of the response - with a more significant response expected on the Left, or less dominant side (Su et al., 2020).

[0102] Conclusion

[0103] Using eye tracking we demonstrated that COVID-19 influences brain circuits that drive eye movements in reaction to visual stimuli, supporting reports that SARS- CoV-2 can disrupt neurotransmitter balance and amplify signals between neurons, thereby facilitating the use of eye tracking as a potential diagnostic for respiratory tract infections.

[0104] EXAMPLE 2:

[0105] Experimental set up as per Example 1

[0106] Methods We used a pro-saccade task to invoke the prepotent characteristic of the oculomotor system by requiring participants to make a saccade, as quickly as possible, to the location of an abrupt onset peripheral target.

[0107] Video based eye tracking was conducted using a VT3 mini Eye Tracker at a 60 Hz recording rate at an Accuracy of 0.5°. All visual stimuli were shown on a 15-inch lap-top screen with a 1980 x 1280 pixel resolution The tracking distance was between 50 and 70 cm.

[0108] Eye tracking data was stored using the MangoldVision software package which allows for the setting up of the video-based stimuli, setting of Areas of Interest (Aol) and post data collection primary analysis. Raw data was further exported as .CSV files for further analysis.

[0109] Prior to the experimental procedure, the VT3 mini Eye Tracker was calibrated by asking the participant to follow the motion of a red dot moving between the four corners and the centre of the screen. The eye tracker recorded the pupil trajectory during the calibration process allowing for the calculation of pupil trajectory for each position on the screen during the experimental process. Success of the calibration was determined by the VT3 mini Eye Tracker having been able to track both pupils for at least 80% of the calibration period.

[0110] Two groups of participants - infected or previously infected with SARS-Cov-2 and healthy were recruited. All participant records were anonymised in accordance with the GDPR regulations in force in the United Kingdom at the time of testing.

[0111] Changes in Pupil Diameter was compared between groups.

[0112] Results:

[0113] NT COVID vs NO COVID - fatigue (as per Example 1):

[0114] Task-related fatigue did not affect saccadic performance, as there was no difference between errors in the first and second half of the saccadic tasks in either group. First vs Second Task ANOVA Single Factor: No COVID p=0.9410 COVID p=0.9471

[0115] COVID vs NO COVID - pupil diameter: The pupil diameter in infected and previously infected COVID patients, widens 30% more than that of people who do not have a current or previous infection (range 20-40%) as shown in Figure 4 depicting the change in pupil diameter for a right eye.

[0116] As with time to first fixation in Example 1, a Left-Right bias was found in the change in pupil diameter (Figure 7).

[0117] Table 1. t-Test: Two-Sample Assuming Equal Variances

[0118] Variable 1 Variable 2

[0119] Mean -0.1299023 -0.0911012

[0120] Variance 0.03683904 0.03616206

[0121] Observations 200 640

[0122] Pooled Variance 0.03632282

[0123] Hypothesized Mean Difference 0 df 838 t Stat -2.5131573

[0124] P(T<=t) one-tail 0.0060761 t Critical one-tail 1.64667399

[0125] P(T<=t) two-tail 0.01215221 t Critical two-tail 1.96279888

[0126] Changes when focussing on the stimulus - the rate of change as measured in the X (Figure 2) and Y (Figure 3) axes, is 12% faster in infected and previously infected COVID patients with a range of 10-17% as compared to a saccade measured at 250 milliseconds.

[0127] Conclusion

[0128] Using eye tracking we demonstrated that COVID-19 infections affect pupillary response to rapid onset visual stimuli supporting reports that SARS-CoV-2 can disrupt neuromuscular reactions and amplify responses, thereby facilitating the use of eye tracking as a potential diagnostic for respiratory tract infections.

[0129] EXAMPLE 3:

[0130] The Visual World Paradigm approach is well-suited to being paired with remote eye-tracking technology for studying cognitive-linguistic processes. Multiple pictures are presented at a time, followed by an auditory phrase or sentence. The rate, timing and direction of eye movements have traditionally been recorded and analysed using an eye-tracking set-up in this VWP.

[0131] Methods

[0132] The task was administered both to adults either infected or previously infected with SARS-Cov-2 and healthy were recruited. All participant records were anonymised in accordance with the GDPR regulations in force in the United Kingdom at the time of testing. Saccadic performance was compared between the two groups.

[0133] Participants wore glasses or contact lenses for the duration of this eye-tracking task if these were usually worn to correct for reading purposes or when carrying out work on a computer screen.

[0134] The same experimental set up as per EXAMPLE 1 was used.

[0135] A block of practice stimuli was subsequently presented in order to familiarise participants with the type and pacing / tempo of stimuli. Opportunity was provided to participants for this practice block to be repeated as often as desired, should individual participants request this.

[0136] Experimental stimulus items were presented after this practice block. These stimuli were scene-based visual displays produced in a Visual World Paradigm (VWP) format and mirroring the arrangement of Altmann and Kamide (1999). Participants were instructed to listen to a spoken sentence and look at the corresponding images on a screen when these appeared on the screen. These audio-visual stimuli were preceded by a centrally-placed hairline cross that was visible on an off-white screen for one second and which served to encourage participants to fixate their eye gaze on the centre of the screen. An image of a visual scene together with its accompanying spoken English sentence were presented simultaneously to replace the hairline cross, and remained in situ for a further six seconds.

[0137] Groups of five or six individual audio-visual scene displays were combined into mp4 format video sequences that ranged in length from 39 to 42 seconds. These short video sequences were then presented sequentially, with the examiner controlling transitions between each sequence using a mouse click. The experimenter checked verbally with the participant that they were still comfortable at the end of each sequence and that they were ready to start the next block of stimuli. The task was completed in approximately 10 minutes. This included allowances for short breaks between video segments to allow participants to rest their eyes and recalibration of eye gaze midway through the block of experimental stimuli items. The same calibration procedure was used as at the outset.

[0138] Results:

[0139] The number of gazes on a Target image vs the Distractor images was reduced by 20% in infected and previously infected COVID patients as compared to healthy controls (Figures 8A and 8B, respectively). Range 12-27%. The relative gaze distribution to the distractors was similar between previously infected COVID patients as compared to healthy controls.

[0140] This data indicates that overall responses to auditory stimuli and subsequent cognitive processing is impaired as has previously been reported (Tyagi et al., (2023) but that this can be used to distinguish between infected and uninfected cohorts, provided that a culturally suitable set of stimuli and a culturally appropriate voice over is used for the verbal cues.

[0141] Conclusion

[0142] Using eye tracking we demonstrated that COVID-19 infection influences the response to auditory stimuli and subsequent cognitive processing. This results in the reduction of time that a person with current or historic infection can focus on a specific image identified by the auditory stimulus. Whilst from a neurological perspective this result supports the notion of "brain fog" attributed to SARS-CoV-2 infections, this significant change in response allows for the use of eye tracking as a potential diagnostic for respiratory tract infections.

[0143] References

[0144] All documents cited below are incorporated herein by reference. Bowe, B., Xie, Y. & Al-Aly, Z. Postacute sequelae of COVID-19 at 2 years. Nat Med 29, 2347-2357, 2023. https: / / doi.org / 10.1038 / s41591-023-Q2521-2

[0145] Cristina Delgado-Alonso, Maria Valles-Salgado, Alfonso Delgado-Alvarez, Miguel Yus, Natividad Gomez-Ruiz, Manuela Jorquera, Carmen Polidura, Mana Jose Gil, Alberto Marcos, Jorge Matias-Guiu, Jordi A. Matias-Guiu, Cognitive dysfunction associated with COVID-19: A comprehensive neuropsychological study, Journal of Psychiatric Research, 150, 40-46, 2022,

[0146] Hessels, R. S., Kemner, C., van den Boomen, C., & Hooge, I. T. The area-of- interest problem in eyetracking research: A noise-robust solution for face and sparse stimuli. Behavior research methods, 48, 1694-1712, 2016.

[0147] Iwashyna TJ, Seelye S, Berkowitz TS, et al. Late Mortality After COVID-19 Infection Among US Veterans vs Risk-Matched Comparators: A 2-Year Cohort Analysis. JAMA Intern Med. 2023;183(10): 1111-1119. doi: 10.1001 / jamaintern med.2023.3587

[0148] Posner MI. Orienting of attention. Quarterly journal of experimental psychology, 32,1, 3-25, 1980. httos: / / doi.onj / 10.1080 / 00335558008248231

[0149] Lisa Spitzer and Stefanie Mueller. 2022. Using a test battery to compare three remote, video-based eye-trackers. In 2022 Symposium on Eye Tracking Research and Applications (ETRA '22), June 08-11, 2022, Seattle, WA, USA. ACM, New York, NY, USA, 7 pages. https: / / doi.Q q / 10.1145 / 3517031.3529644

[0150] Kenneth Holmqvist, Marcus Nystrbm, Richard Andersson, Richard Dewhurst, Halszka Jarodzk and Joost van de Weijer. 2011. Eye tracking: A comprehensive guide to methods and measures. OUP, Oxford.

[0151] Kathleen M. Eberhard, Michael J. Spivey-Knowlton, Julie C. Sedivy and Michael K. Tanenhaus. 1995. Eye movements as a window into real-time spoken language comprehension in natural contexts. Journal of psycholinguistic research 24, 409- 436. Gary T. Altmann and Y Kamide. 1999. Incremental interpretation at verbs: Restricting the domain of subsequent reference. Cognition, 73, 3, 247-264.

[0152] Michael I. Posner. 1980. Orienting of attention. Quarterly journal of experimental psychology, 32,1, 3-25.

[0153] Niloofar N. Akhavan, Henrike K. Blumenfeld and Tracy Love. 2020. Auditory Sentence Processing in Bilinguals: The Role of Cognitive Control. Front. Psychol. 11 :898. doi: 10.3389 / fpsyg.2020.00898

[0154] Anuradha Kar and Peter Corcoran. 2017. A review and analysis of eye-gaze estimation systems, algorithms and performance evaluation methods in consumer platforms. IEEE Access, 5, 16495-16519.

[0155] Benjamin T. Carter and Steven G. Luke. 2020. Best practices in eye tracking research. International Journal of Psychophysiology, 155, 49-62.

[0156] Tanja Biascheck, Kuno Kurzhals, Michael Raschke, Michael Burch, Daniel Weiskopf and Thomas Ertl. 2017. Visualization of eye tracking data: A taxonomy and survey. In Computer Graphics Forum 36, 8, 260-284.

[0157] Roy S. Hessels, Chantal Kemner, Carlijn van den Boomen and Ignace T.C. Hooge. 2016. The area -of- interest problem in eyetracking research: A noise-robust solution for face and sparse stimuli. Behavior research methods, 48, 1694-1712. h 11 n s ’ / / d O i O rn / 1 O ’17 R / S 1 A 47 A - Q 1 R -■ 0676 ■ v

[0158] Su, J., Yin, G., Bai, X. et al. Flexibility in the perceptual span during reading: Evidence from Mongolia. Atten Percept Psychophys 82, 1566-1572 (2020).

[0159] Tyagi, K., Rai, P., Gautam, A. et al. Neurological manifestations of SARS-CoV-2: complexity, mechanism and associated disorders. Eur J Med Res 28, 307 (2023).

Claims

Claims1. A method for detecting a current or previous respiratory tract infection, the method comprising: analysing data collected from oculography of a subject performing one or more of a pro-saccade task, an anti-saccade task, a visual search task and a visual world paradigm test to determine values for said one or more prosaccade task, anti-saccade task, visual search task and visual world paradigm test for said subject; and based on the values determined for said subject, detecting the presence of a current or previous respiratory tract infection.

2. The method of claim 1, wherein the oculography is video-oculography.

3. The method of claim 2, wherein the video-oculography includes pupil center corneal reflection and / or infrared oculography.

4. The method of any of claims 1 to 3, wherein analysing the data comprises comparing the values determined for said subject with values determined for healthy control subjects.

5. The method of any of claims 1 to 4, wherein the data includes (i) at least one pro-saccade task and / or at least one anti-saccade task; and (ii) at least one visual search task and / or at least one visual world paradigm test.

6. The method of any of claims 1 to 5, wherein the pro-saccade task measures time to first fixation and / or pupillary response.

7. The method of claim 6, wherein the time to first fixation is faster than that of a healthy control subject.

8. The method of claim 6 or 7, wherein the pupillary response is greater than that of a healthy control subject.

9. The method of any of claims 1 to 8, wherein the visual world paradigm test measures the number of gazes on a target image compared to the number ofgazes on a distractor image and wherein the subject achieves fewer gazes on the target image than a healthy control subject.

10. The method of any of claims 1 to 9, wherein the data collected from the oculography are analysed using a multimodal artificial intelligence model.

11. The method of any of claims 1 to 10, wherein the respiratory tract infection is a current infection or a previous infection within six months preceding the test, or within three months preceding the test, or within one month preceding the test.

12. The method of claim 11, wherein the respiratory tract infection is a viral infection or a bacterial infection.

13. The method of claim 12, wherein the viral infection is a coronavirus infection, or wherein the bacterial infection is a Pseudomonas infection, a Staphylococcus infection, a Mycobacterium infection, Streptococcus infection, or a Burkholderia infection.

14. The method of claim 13, wherein the coronavirus infection is a Sars-CoV-2 infection.

15. A method for detecting a current or previous respiratory tract infection, the method comprising: measuring performance of one or more of a pro-saccade task, an antisaccade task, a visual search task and a visual world paradigm test by a subject using oculography; analysing data collected from the oculography to determine values for said one or more pro-saccade task, anti-saccade task, visual search task and visual world paradigm test for said subject; and based on the values determined for said subject, detecting the presence of a current or previous respiratory tract infection.

16. The method of claim 15, wherein the method additionally comprises the subject performing one or more calibration tasks prior to performing the one or more pro-saccade task, anti-saccade task, visual search task and visual world paradigm test.

17. The method of claim 16, wherein the calibration task comprises the subject tracking movement of a dot across a screen.

18. The method of any of claims 15 to 17, wherein the method comprises measuring performance of time to first fixation, pupillary response and a visual world paradigm test.

19. A method for measuring performance of one or more of a pro-saccade task, an anti-saccade task, a visual search task and a visual world paradigm test by a subject using oculography, wherein the method additionally comprises the subject performing one or more calibration tasks prior to performing the one or more pro-saccade task, anti-saccade task, visual search task and visual world paradigm test.

20. The method of claim 19, wherein the calibration task comprises the subject tracking movement of a dot across a screen.

21. The method of any of claims 19 or 20, wherein the method comprises measuring performance of time to first fixation, pupillary response and a visual world paradigm test.

22. A computer-implemented method for detecting a current or previous respiratory tract infection in a subject according which when executed on one or more processors cause the one or more processors to perform the method of any preceding claim.

23. A computing apparatus for detecting a current or previous respiratory tract infection, said apparatus comprising one or more processors configured to perform the method of claim 22.

24. The apparatus of claim 23 further comprising: a display; a computer-readable storage medium, and a video based eye tracker configured for acquiring oculography data.

25. A computer-readable medium for storing instructions executable by a processor to enable an apparatus to perform the method of any of claim 22.

Images