A digital motor score for sensitive detection of huntington's disease

Abstract

The present invention relates to the field of diagnostics. Specifically, it relates to a computer-implemented method for assessing Huntington's disease (HD) in a subject comprising the steps of determining a Huntington's disease digital motor score (HDDMS) based on a multitude of digital performance features derived from at least one or more of the following: accelerometer measurements, touch sensor measurements, and / or measurements of time, in a dataset of fine motoric and motoric activity measurements from said subject; comparing the determined HDDMS to a reference; and assessing HD in the subject based on said comparison. Further, the invention contemplates a device and a system for carrying out the afore-mentioned methods and the use of such device or system for assessing Huntington's disease in the subject.

Description

[0001] A digital motor score for sensitive detection of Huntington’s disease

[0002] The present invention relates to the field of diagnostics. Specifically, it relates to a computer- implemented method for assessing Huntington’s disease (HD) in a subject comprising the steps of determining a Huntington’s disease digital motor score (HDDMS) based on a multitude of digital performance features derived from at least one or more of the following: accelerometer measurements, touch sensor measurements, and / or measurements of time in a dataset of fine motoric and motoric activity measurements from said subject; comparing the determined HDDMS to a reference; and assessing HD in the subject based on said comparison. Further, the invention contemplates a device and a system for carrying out the aforementioned methods and the use of such device or system for assessing Huntington's disease in the subject.

[0003] Huntington’s disease (HD) is a rare neurodegenerative autosomal dominant monogenic disease. It is caused by a cytosine-adenine-guanine (CAG) trinucleotide repeat expansion in the HTT gene, resulting in the production of mutant mRNA and protein gene products, including mutant huntingtin protein [1-3], People with HD typically experience progressive cognitive and motor impairment as well as behavioural and metabolic alterations leading to progressive disability, loss of independence, and ultimately death. The median survival is 15 years after the onset of unequivocal motor symptoms [4],

[0004] Treatment options are needed that delay, stop, or ideally prevent disease progression. While novel therapeutic approaches are being investigated [5], quantifying disease progression in people with HD remains challenging, particularly in those experiencing minimal disease manifestation and functional decline, e.g., people in Huntington’s Disease Integrated Stating System (HD-ISS) Stage 2 [6], HD clinical anchors that have been accepted as clinically meaningful trial endpoints, like the composite Unified Huntington’s Disease Rating Scale (cUHDRS), typically show limited sensitivity or, like the Total Functional Capacity, suffer from ceiling effects in early-stage disease [6-8], Consequently, clinical trials often require large study cohorts and long observation periods, which may be difficult to attain.

[0005] Efforts to address these limitations include objective measures of motor function [9,10], including clinic-based platforms such as the Q-Motor Scale [11,12], A HD digital monitoring platform was previously described that consists of remote, self-administered, digital motor and cognitive tests that can be deployed on a large scale in terms of cohort size and study duration. It is shown that features derived from these tests, which by themselves allow quantification of test performance, capture HD-related disease signals and are robust in cross- sectional analyses

[0013] ,

[0006] Computer-implemented tests for assessing Huntington's disease have been described, inter alia, in WO 2019 / 081640 and WO 2021 / 089509.

[0007] However, diagnostic tools are needed that allow a reliable and sensitive diagnosis and identification of the Huntington's disease and in particular identification of progression in HD in order to allow for proper care and / or an accurate treatment. Further there is a need for simple and reliable monitoring of progression in HD.

[0008] Here, a new data-driven Huntington’s Disease Digital Motor Score (HDDMS) for sensitive measurement of disease progression (i.e., clinical decline) in HD is described. The HDDMS is derived from the motor tests included in the HD digital monitoring platform (Fig. 1). Using the best-performing features in terms of their ability to capture HD disease progression, their test-retest reliability, and their agreement with gold-standard clinical anchors, we developed the HDDMS for use in clinical research. The aim of the HDDMS is to complement existing clinical assessments by offering higher sensitivity to change, even in HD-ISS Stage 2, thus enabling a reduction in cohort size and / or study duration. To increase its generalisability, the HDDMS was developed and separately validated using data collected from 1,008 individuals enrolled across different studies.

[0009] The technical problem underlying the present invention may be seen in the provision of means and methods complying with the aforementioned needs. In particular, it is desirable to provide methods and means for accurate and sensitive HD assessment, in particular for assessment of progression in HD. The technical problem is solved by the embodiments characterized in the claims and described herein below.

[0010] This problem in particular is addressed by a method and a device for assessing Huntington's disease (HD), typically progression in HD with the features of the independent claims. Advantageous embodiments which might be realized in an isolated fashion or in any arbitrary combinations are listed in the dependent claims as well as throughout the specification.

[0011] The present invention relates to a computer-implemented method for assessing Huntington's disease (HD), typically progression in HD, in a subject comprising the steps of: a) determining a Huntington’s disease digital motor score (HDDMS) based on a multitude of digital performance features derived from at least one or more of the following: accelerometer measurements, touch sensor measurements, and / or measurements of time in a dataset of fine motoric and motoric activity measurements from said subject; b) comparing the determined HDDMS to a reference; and c) assessing HD, typically progression in HD, in the subject based on said comparison.

[0012] Advantageously, it has been found in the studies underlying the present invention and as detailed in the examples below that a Huntington’s disease digital motor score (HDDMS) allows quantitative assessment of HD and in particular progression in HD in patients with higher sensitivity than standard clinical assessments and / or pre-existing digital tests.

[0013] The method as referred to in accordance with the present invention includes a method which essentially consists of the aforementioned steps or a method which may include additional steps. Typically the method is an ex vivo method carried out on a preexisting dataset of activity measurements from a subject which does not require any physical interaction with the said subject.

[0014] As used in the following, the terms “have”, “comprise” or “include” or any arbitrary grammatical variations thereof are used in a non-exclusive way. Thus, these terms may both refer to a situation in which, besides the feature introduced by these terms, no further features are present in the entity described in this context and to a situation in which one or more further features are present. As an example, the expressions “A has B”, “A comprises B” and “A includes B” may both refer to a situation in which, besides B, no other element is present in A (i.e. a situation in which A solely and exclusively consists of B) and to a situation in which, besides B, one or more further elements are present in entity A, such as element C, elements C and D or even further elements. Further, it shall be noted that the terms “at least one”, “one or more” or similar expressions indicating that a feature or element may be present once or more than once typically will be used only once when introducing the respective feature or element. In the following, in most cases, when referring to the respective feature or element, the expressions “at least one” or “one or more” will not be repeated, non-withstanding the fact that the respective feature or element may be present once or more than once.

[0015] Further, as used in the following, the terms "preferably", "more preferably", "particularly", "more particularly", "specifically", "more specifically", “typically”, “more typically” or similar terms are used in conjunction with optional features, without restricting alternative possibilities. Thus, features introduced by these terms are optional features and are not intended to restrict the scope of the claims in any way. The invention may, as the skilled person will recognize, be performed by using alternative features. Similarly, features introduced by "in an embodiment of the invention" or similar expressions are intended to be optional features, without any restriction regarding alternative embodiments of the invention, without any restrictions regarding the scope of the invention and without any restriction regarding the possibility of combining the features introduced in such way with other optional or non-optional features of the invention.

[0016] The method is, typically, computer implemented. In other words, the method according to the present invention is a computer-implemented method, i.e. the steps a) to c) are typically carried out in an automated manner by use of at least one data processing device. Details are also found herein below and in the accompanying Examples. The term “computer-implemented” specifically may refer, without limitation, to a method involving at least one computer and / or at least one computer network. The computer and / or computer network may comprise at least one processor, which is configured for performing at least one of the method steps of the method according to the present invention. The term “computer” according to the present invention typically encompasses any type of mobile device or data- processing device specified elsewhere herein. Specifically, in a computer-implemented method, each of the method steps is performed by the computer and / or computer network. The method may be performed completely automatically, specifically without user interaction.

[0017] The term “assessing HD” as used herein includes diagnosing, staging, classifying, monitoring and / or predicting Huntington's disease, recommending therapies or monitoring new therapies against Huntington's disease of all stages. It typically relates to assessing progression in HD. As will be understood by those skilled in the art, such an assessment, although preferred to be, may usually not be correct for 100% of the investigated subjects. The term, however, requires that a statistically significant portion of subjects can be correctly assessed. Whether a portion is statistically significant can be determined without further ado by the person skilled in the art using various well known statistic evaluation tools, e.g., determination of confidence intervals, p-value determination, Student's t-test, Mann- Whitney test, etc.. Details may be found in Dowdy and Wearden, Statistics for Research, John Wiley & Sons, New York 1983. Typically envisaged confidence intervals are at least 50%, at least 60%, at least 70%, at least 80%, at least 90%, at least 95%. The p-values are, typically, 0.2, 0.1, 0.05. Thus, the method of the present invention, typically, aids the assessment of Huntington's disease by providing a means for evaluating a dataset of fine motoric activity measurements.

[0018] The term “Huntington's disease (HD)” as used herein relates to an inherited neurological disorder accompanied by neuronal cell death in the central nervous system. Most prominently, the basal ganglia are affected by cell death. There are also further areas of the brain involved such as substantia nigra, cerebral cortex, hippocampus and the purkinje cells. All regions, typically, play a role in movement and behavioral control. The disease is caused by genetic mutations in the gene encoding Huntingtin. Huntingtin is a protein involved in various cellular functions and interacts with over 100 other proteins. The mutated Huntingtin appears to be cytotoxic for certain neuronal cell types. Mutated Huntingtin is characterized by a poly glutamine region caused by a trinucleotide repeat in the Huntingtin gene. A repeat of more than 36 glutamine residues in the poly glutamine region of the protein results in the disease causing Huntingtin protein. Since Huntington's disease is inherited in a dominant autosomal manner, genome testing for CAG repeats in the huntingtin (HTT) alleles is recommended for individuals being genetically at risk, i.e. patients with a corresponding family history of the disease. Moreover, diagnosis of the disease involves DNA analysis but also imaging methods such as CT, MRI, PET or SPECT scanning, in order to determine cerebral atrophy as well as neurological assessment by a medical practitioner.

[0019] The symptoms of the disease most commonly become noticeable in the mid-age, but can begin at any age from infancy to the elderly. In early stages, symptoms involve subtle changes in personality, cognition, and physical skills. The physical symptoms are usually the first to be noticed, as cognitive and behavioral symptoms are generally not severe enough to be recognized on their own at said early stages. Almost everyone with Huntington's disease eventually exhibits similar physical symptoms, but the onset, progression and extent of cognitive and behavioral symptoms vary significantly between individuals. The most characteristic initial physical symptoms are jerky, random, and uncontrollable movements called chorea. Chorea may be initially exhibited as general restlessness, small unintentionally initiated or uncompleted motions, lack of coordination, or slowed saccadic eye movements. These minor motor abnormalities usually precede more obvious signs of motor dysfunction by at least three years. The clear appearance of symptoms such as rigidity, writhing motions or abnormal posturing appear as the disorder progresses. These are signs that the system in the brain that is responsible for movement has been affected. Psychomotor functions become increasingly impaired, such that any action that requires muscle control is affected. Common consequences are physical instability, abnormal facial expression, and difficulties chewing, swallowing, and speaking. Consequently, eating difficulties and sleep disturbances are also accompanying the disease. Cognitive abilities are also impaired in a progressive manner. Impaired are executive functions, cognitive flexibility, abstract thinking, rule acquisition, and proper action / reaction capabilities. In more pronounced stages, memory deficits tend to appear including short-term memory deficits to long-term memory difficulties. Cognitive problems worsen overtime and will ultimately turn into dementia. Psychiatric complications accompanying Huntington's disease are anxiety, depression, a reduced display of emotions (blunted affect), egocentrism, aggression, and compulsive behavior, the latter of which can cause or worsen addictions, including alcoholism, gambling, and hypersexuality.

[0020] At present, there is no cure for Huntington's disease. There are supportive measurements in disease management depending on the symptoms to be addressed. Moreover, a number of drugs are used to ameliorate the disease, its progression or the symptoms accompanying it. Tetrabenazine is approved for treatment of Huntington's disease, include neuroleptics and benzodiazepines are used as drugs that help to reduce chorea, amantadine or remacemide are still under investigation but have shown preliminary positive results. Hypokinesia and rigidity, especially in juvenile cases, can be treated with antiparkinsonian drugs, and myoclonic hyperkinesia can be treated with valproic acid. Ethyl-eicosapentoic acid was found to enhance the motor symptoms of patients, however, its long term effects need to be revealed.

[0021] Recent studies focus on evaluating the effects of Huntingtin lowering therapies, for example by tominersen treatment, on HD symptoms and progression seem promising. However, clinical studies require a certain number of participants. The numbers are directly related to the sensitivity with which diseases progression can be tracked. In rare diseases such as HD, suitable numbers of participant for clinical studies are often difficult to reach and / or the recruitment takes a long time. The disease can be diagnosed by genetic testing. Moreover, the severity of the disease can be staged according to Unified Huntington's Disease Rating Scale (UHDRS) or by using the Huntington's Disease Integrated Staging System (HD-ISS) described elsewhere herein. Using the Huntington's Disease Integrated Staging System addresses four components, i.e. the motor function, the cognition, behavior and functional abilities. The motor function assessment includes assessment of ocular pursuit, saccade initiation, saccade velocity, dysarthria, tongue protrusion, maximal dystonia, maximal chorea, retropulsion pull test, finger taps, pronate / supinate hands, luria, rigidity arms, bradykinesia body, gait, and tandem walking and can be summarized as total motor score (TMS). The motoric functions must be investigated and judged by a medical practitioner or trained HD rater.

[0022] The HD-ISS specifically characterizes individuals for research purposes from birth, starting at Stage 0 (i.e, individuals with the Huntington's disease genetic mutation without any detectable pathological change; may also be referred to as pre-manifest stage of HD) by using a genetic definition of Huntington's disease. Huntington's disease progression may then be marked by measurable indicators of underlying pathophysiology (Stage 1), a detectable clinical phenotype (Stage 2), and then decline in function (Stage 3). The HD-ISS is commonly used to precisely classify individuals into stages based on thresholds of stage-specific landmark assessments (Tabrizi et al, 2022, The Lancet Neurology, Vol. 21, P632-644).

[0023] Huntington's disease is a progressing disease typically associated with a worsening of the overall disease condition and symptoms thereof over time. The term “progression in HD” as referred to in accordance with the present invention specifically relates to any worsening of or increase of symptoms, more specifically motor symptoms. Progression in HD can be assessed by clinical tools known in the art, such as by the UHDRS or HD-ISS. Disease progression in HD of a subject typically relates to the proceeding of said subject formerly being assessed with HD from one stage to the following stage in disease, for example from HD- ISS stage 2 to HD-ISS stage 3. In line with the present invention, the term “progression in HD” relates to a significant change, typically an increase in the determined HDDMS over time. A change in HD over time, such as a reduction in Total Functional Capacity by 1 point, is commonly expected to occur within months time, i.e. over a period of time of 12 months, specifically for subjects with late HD-ISS stage 2 or early HD-ISS stage 3. Advantageously, the HDDMS is shown to be capable of assessing changes in HD much earlier, i.e. within 20 weeks. This is demonstrated in the examples provided herein below. Moreover, the present inventors found that the HDDMS could be used in stages 0 and 1 to collect reference data for individuals and may detect the onset of stage 2, where motor and cognitive symptoms are appearing for the first time, but function is not yet affected. The term “subject” as used herein relates to animals and, typically, to mammals. In particular, the subject is a primate and, most typically, a human. A subject in accordance with the present invention typically suffers from or is be suspected to suffer from Huntington's disease, for example prior to the onset of symptoms. A subject in accordance with the present invention is more typically a subject diagnosed with HD, for example by genetic testing known in the art. A subject diagnosed with HD may typically be understood as referring to a subject at any stage of HD, e.g. from HD-ISS stage 0 to stage 3. Genetic testing may specifically be used for diagnosing HD based on the presence of certain risk factors in the hun- tingtin gene (HIT). Genetic risk factors that relate to an increased probability for developing for Huntington's disease may be determined and subjects being affected by such risk factors may be deemed to be subjects being suspected to suffer from Huntington's disease. Even more typically, a subject diagnosed with HD according to the method of the present invention refers to a previously positively tested for HD in a genetic test as known in the art. The term “patient” or “HD patient” as used herein relates to a subject that suffers from HD and / or that known to suffer from HD Typically, a patient is a subject that is classified in any of the stages from 1 to 3 of the HD-ISS, more typically a patient is a subject that is classified in any of the stages from 2 to 3 of the HD-ISS.

[0024] The term “Huntington’s disease digital motor score (HDDMS)” relates to a digital biomarker for HD based on a multitude of digital performance features. Specifically, the HDDMS relates to a composite score derived from sensor-based fine motoric and motoric activity tests that may typically be included in a digital remote monitoring platform. Said sensors are typically sensors of a mobile device as known in the art.

[0025] The HHDMS preferably comprises at least three digital performance features, more preferably at least four, even more preferably, the HDDMS comprises four digital performance features derived from a dataset of fine motoric and motoric activity measurements from said subject. Still more preferably, the HDDMS comprises at least one digital performance feature derived from a dataset of fine motoric activity measurements and at least two, typically three, digital performance features derived from datasets of motoric activity measurements. Still more preferably, the HDDMS comprises one digital performance feature derived from a dataset of fine motoric activity measurements and three digital performance features derived from datasets of motoric activity measurements.

[0026] The digital performance features are typically combined with each other to give a sum score representing the HDDMS. The digital performance features may typically be standardized. More typically the standardization may be performed using means and methods as known in the art. Particularly suitable are standardization procedures based on large datasets such as subtracting the arithmetic mean of the dataset from each individual measurement of a performance feature, and dividing the result by the pooled standard deviation of the dataset.

[0027] More typically, digital performance features are weighted according to their ability to represent disease severity. Such weights may typically be determined by exploratory factor analysis.

[0028] Advantageously, the HDDMS as used in the method according to the present invention for assessing HD has a strong capacity to detect change in HD over time, a high test re-test reliability and a stable close to normal distribution. Notably, the HDDMS demonstrates superior sensitivity to change relative to commonly used HD clinical anchors such as composite UHDRS.

[0029] A “multitude of digital performance features” in accordance with the present invention refers to at least two, typically at least three, more typically at least four, even more typically four, i.e. exactly four, digital performance features derived from a dataset of fine motoric and motoric activity measurements from a subject. Typically, said features are derived from at least accelerometer measurements or touch sensor measurements in a dataset of fine motoric and motoric activity measurements, and may further include a time component such as a measurement of time, or iterating said measurements over a specified time period. Typically said period of time depends on the type of fine motoric and motoric test performed by the subject and defined elsewhere herein. A typical time period for said measurements is 30 s up to 120 s. Said measurements are typically obtained with, or as part of a dataset of fine motoric and motoric activity measurements from said subject. It is to be understood that besides said accelerometer measurements, touch sensor measurements, and / or measurements of time, the dataset of fine motoric and motoric activity measurements typically may comprise additional measurement data including voice recording and / or video recording that could be analyzed as an optional component for quality control purposes.

[0030] The term “digital performance feature” as used herein refers to a parameter, also referred to as “feature” herein, which is indicative for the ability of a subject to carry out a certain activity during the measurements performed in fine motoric activity or motor activity tests. Typically, the performance feature reflects the motor skills and / or fine motor skills of the subject. More typically, the performance feature is a measure for the variability of the motor skills and / or fine motor skills the activity test over time in. Even more typically, the digital performance feature is a measure of intra-performance fluctuations of fine motoric or motoric skills of the subject when carrying out the activity test. Still more typically, the digital performance feature is a measure of intra-performance fluctuations in at least one quantitative feature of fine motoric skills or motoric skills of the subject when carrying out the activity test.

[0031] In line with the present invention, a digital performance feature preferably comprises, more preferably is, a quantitative value derived from accelerometer measurements, touch sensor measurements, and / or measurements of time in a dataset of fine motoric and motoric activity measurements from a subject. Said quantitative value even more preferably comprises, typically consists of, statistical values derived from said accelerometer measurements, touch sensor measurements, and / or measurements of time. The quantitative value may typically be an aggregated value by taking the median or average across a specified period of time; more typically across a period of a few weeks, such as one to four weeks typically two weeks.

[0032] Said digital performance features may in particular comprise quantitative values; more particularly the quantitative values comprise statistical values derived from said accelerometer measurements, touch sensor measurements, and / or measurements of time. Typically, said statistical values relate to the standard deviation and / or the coefficient of variation derived from said accelerometer measurements, touch sensor measurements, and / or measurements of time. The accelerometer measurements, touch sensor measurements, and / or measurements of time are preferably obtained from fine motoric and motoric activity measurements carried out using a mobile device. Hence, the accelerometer, the touch sensor and the chronometer are more preferably sensors comprised in said mobile device.

[0033] More typically, said statistical values comprise two or more of the following: standard deviation of the accelerometer magnitude; absolute value of the standard deviation of the z-acceleration divided by the accelerometer magnitude; standard deviation of the time between the onset of one tap and the onset of the subsequent tap; coefficient of variation of step impulse.

[0034] Ways of calculation said statistical values are known in the art and further explained in the examples provided below. The statistical values are typically calculated from a series of data points from said accelerometer measurements, touch sensor measurements, or measurements of time. Specifically said data points are collected during the fine motoric and motoric activity measurements. More typically, said data points are part of the dataset of fine motoric and motoric activity measurements from said subject. Even more typically, said data points relate to measurement of iterations of tasks performed during the fine motoric and motoric activity tests; or said data points relate to iterated measurements obtained from said fine motoric and motoric activity tests. Details on said activity tests are described elsewhere herein.

[0035] The digital performance features according to the present invention are specifically advantageous as they are characterised by the following: high test-retest reliability, good convergent validity and high sensitivity to change. Further details can be found in the examples and figures given below (see: high test-retest reliability Table 6, good convergent validity Fig. 3B, and good sensitivity to change Fig. 4C and 4D).

[0036] The HDDMS may be built from the simple sum score of the digital performance features. Typically, the digital performance features in the HDDMS each are weighted according to their ability to represent disease severity. More typically, the digital performance features in the HDDMS may each comprise a factor weight or may alternatively each comprise a unit weight.

[0037] The fine motoric and motoric activity measurements in accordance to the present invention are typically carried out using a mobile device. More typically said dataset of fine motoric and motoric activity measurements comprises measurements of fine motor control of the fingers and measurements of motor control of lower limbs, trunk and upper limbs; preferably measurements of fine motor control of the fingers, measurements of motor control of lower limbs and measurements of chorea of trunk and upper limbs. More preferably, said measurements are obtained from active tests of motor and fine motor skills of the subject; even more preferably said tests are performed on a mobile device. Said active test are described elsewhere in more detail.

[0038] In line with the present invention, typically said dataset of activity measurements are collected from the subject while carrying out fine motoric or motoric activity tests. Typical fine motoric and motoric activity tests in accordance with the present invention assess balance, gait, finger movement accuracy, and / or movement speed; more typically these activity tests are one or more, up to all of the following fine motoric and motoric activity tests: Balance Test, Chorea Test, Speeded Tapping Test, Two-Minute Walk Test.

[0039] More typically, said data collection is performed by the mobile device, even more typically said data is stored by a data storage unit of said mobile device. The term “dataset of activity measurements” refers, in principle, to the entirety of data acquired, in particular by the mobile device, from a subject during fine motoric and motoric activity measurements or any subset of said data useful for deriving a digital performance feature. Details are also found elsewhere herein. In particular, the activity measurements in connection with the term “dataset of fine motoric and motoric activity measurements” as used in accordance with the present invention comprise measurements of data obtained from a subject performing an activity test in accordance with the present invention such as a balance test (BAL), chorea test (CHO), Two-minute walk test (TMW), speeded tapping test (STT). Further details on these test are described for example in WO 2021 / 089509 and WO 2020 / 157083 which is incorporated herewith by reference. Further test, which may be carried out together with the aforementioned ones on are those described in WO 2019 / 081640.

[0040] Typically, the fine motoric and motoric skills measured by said fine motoric and motoric activity tests is fine motor control of the fingers and / or motor control of lower limbs, trunk and / or upper limbs; preferably measurements of fine motor control of the fingers, motor control of lower limbs and / or chorea of trunk and upper limbs.

[0041] More typically, these fine motoric and motoric activity tests comprise iterations of specific fine motoric and / or motoric tasks over time or a certain number of iterations is performed by the subject while completing the activity test. The dataset of fine motoric and motoric activity tests hence particularly comprises a set of data points of fine motoric and motoric activity measurements relating to said iterated fine motoric and / or motoric tasks. These tasks more particularly depend on the specific fine motoric or motoric test.

[0042] In line with the present invention, the following four digital performance features are preferably combined for building the HDDMS: standard deviation of the accelerometer magnitude, preferably derived from the balance test; absolute value of the standard deviation of the z-acceleration divided by the accelerometer magnitude, preferably derived from the chorea test; standard deviation of the time between the onset of one tap and the onset of the subsequent tap, preferably derived from the speeded tapping test; coefficient of variation of step impulse, preferably derived from the two-minute walk test. The HDDMS may be built from the simple sum score of the digital performance features.

[0043] Typically, the digital performance features in the HDDMS each are weighted according to their ability to represent disease severity. More typically, the digital performance features in the HDDMS may each comprise a factor weight or may alternatively each comprise a unit weight. The highest weight may be accounted to the digital performance feature derived from the BAL test, the lowest weight may be accounted to the digital performance feature derived from the STT.

[0044] In the BAL test, the subject typically stands straight for a specified period of time, such as about 30 seconds, involuntary movements or shakings are measured in particular by the accelerometer of the mobile device. The BAL test is a preferred test for assessing chorea of the trunk. A digital feature particularly suitable for building the HHDMS derived therefrom is the standard deviation of the accelerometer magnitude. The standard deviation of the accelerometer magnitude may typically be calculated across measurements, in particular of meas- urements of accelerometer magnitude, taken during the duration of the BAL test, i.e. the specified period of time, in particular to characterize shakings and / or involuntary movements. This is advantageous as it specifically ensures that all involuntary movements and shakings occurring during the BAL test are captured.

[0045] In the CHO test, the subject typically holds the mobile device with one arm outstretched for a specified period of time, such as about 30 seconds, more typically while performing an additional cognitive task. The additional cognitive task specifically is intended to sufficiently distract the subjects so that they do not have the mental capacity to consciously suppress choreatic, i.e. involuntary, movements and are more specifically not being analyzed and / or are not included for calculating the HDMMS. Involuntary movements are measured in particular by the accelerometer of the mobile device, more particularly while the performance of the cognitive task may be recorded for the purposes of quality control via the microphone of the mobile device. Even more typically, the dominant and non-dominant arm are assessed separately. Still even more typically, the additional cognitive task consists of a simple math- ematical task. For example, the additional cognitive task consists of counting backwards from a number between 100 and 120 in steps of 7 A digital feature particularly suitable for building the HHDMS derived from the accelerometer data is the absolute value of the standard deviation of the z-acceleration divided by the accelerometer magnitude. The “z-acceler- ation” specifically refers to the acceleration in upward and / or downward direction of the tested arm. The standard deviation of the z-acceleration may typically be calculated across measurements, particularly z-acceleration measurements, during the duration of the CHO test, i.e. the specified period of time, more particularly to characterize involuntary movements. The accelerometer magnitude may typically be averaged across measurements during the duration of the CHO test. This is advantageous as it specifically ensures that all involuntary movements occurring during the CHO test are captured.

[0046] Still more typically, a digital feature derived from the CHO test comprises the average performance across both hands.

[0047] In the TMW test, the subject typically walks for two minutes without a break on a straight and even path. A digital feature particularly suitable for building the HHDMS is the coefficient of variation of step impulse derived from the TMW test. As used herein, the term “step impulse” may refer to the integral of the absolute residuals of the acceleration magnitude time series defined throughout a step. It is typically a model-free estimate of the movement intensity of a step, and more typically strongly correlated to step velocity. Even more typi- cally, the “absolute residuals” refer to the absolute value of the deviations of the acceleration magnitudes from the mean acceleration magnitudes calculated from the accelerometer measurements collected during the two minutes walk test.

[0048] The STT, specifically is a bimanual test. Typically, the subject performs as many taps as possible typically with the index finger of each the dominant and non-dominant hand separately on the touch screen of a mobile device, for a specified period of time, such as about 30 seconds. The data measured in the separate assessments may be averaged over both sides, dominant and non-dominant hand, for deriving the digital performance feature. A digital feature particularly suitable for building the HHDMS derived from the STT is the standard deviation of time between the onset of one tap and the onset of the subsequent tap, typically calculated across all taps performed during the STT, i.e. in the specified period of time.

[0049] The digital performance feature typically is derived from a statistical value representing the fluctuation between iterations in the dataset of fine motoric and motoric activity measure- ments, i.e. intra-test fluctuations also referred to as intra-performance fluctuations herein.

[0050] Typically, the dataset of fine motoric and motoric activity measurements from said subject is a preexisting dataset. “Preexisting dataset” in particular means that the method of the invention typically does not require data acquisition from the subject; more particularly, the dataset is already existing at the time the method according to the invention is performed.

[0051] More typically, said dataset of fine motoric and motoric activity measurements is collected or obtained by a mobile device, i.e. by a subject using a mobile device, at any time before the method of the invention is performed. Obtaining said dataset of fine motoric and motoric activity measurements typically generates a preexisting dataset of fine motoric and motoric activity measurements. More typically, the method according to the present invention such as the step of determining the HDDMS from said preexisting dataset of fine motoric and motoric activity measurements is performed on a further device, i.e. a second or further data processing device non- identical with said mobile device. Still more typically, the mobile device and the second data processing device are operatively linked to each other. Still more typically, the mobile device and the further data processing device are separated from each other, e.g. they are in separate locations.. Preferably, the further data processing device fur- ther comprises at least a database as well as software which is tangibly embedded to said device and, when running on said device, carries out the method of the invention.

[0052] The term “operatively linked to each other” according to the present invention is understood as the devices being connected in a way that allows data transfer from one device to the other device. Typically, it is envisaged the mobile device which acquires the dataset of fine motoric and motoric activity measurements from the subject is connected to the further device carrying out steps of the methods of the invention such that the acquired data can be transmitted for processing to the further device. However, the further device may also transmit data to the mobile device such as signals controlling or supervising its proper function. The connection between the mobile device and the further device may be achieved by a permanent or temporary physical connection, such as coaxial, fiber, fiber-optic or twisted-pair, 10 BASE-T cables. Alternatively, it may be achieved by a temporary or permanent wireless connection using, e.g., radio waves, such as Wi-Fi, LTE, LTE-advanced or Bluetooth. Further details may be found elsewhere in this specification. For data acquisition, the mobile device may comprise a user interface such as screen or other equipment for data acquisition.

[0053] Typically, the activity measurements can be performed on a touchscreen comprised in a mobile device, wherein it will be understood that the said screen may have different sizes including, e.g., a 5.1 inch screen. Still alternatively, the dataset of fine motoric and motoric activity measurements may be stored on a data carrier and may subsequently be transferred from the data carrier to the further device. A suitable data carriers are known in the art, a particularly suitable data carrier in line with the present invention is an SD memory card, a USB stick or a portable storage drive.

[0054] The term “determining the HDDMS” typically comprises determining at least two, more typically, at least three, digital performance features derived from accelerometer measurements, touch sensor measurements, and / or measurements of time in a dataset of fine motoric and motoric activity measurements from a subject. More typically said accelerometer measurements, touch sensor measurements, and / or measurements of time in a dataset of fine motoric and motoric activity measurements from said subject are carried out using a mobile device. Even more typically, the HDDMS is determined on a further device, in particular a further data processing device operatively linked to said mobile device.

[0055] Determining said digital performance features may in particular be achieved either by deriving a desired measurement value from the dataset of fine motoric and motoric activity measurements as the performance feature directly. Preferably, the digital performance feature may integrate one or more measurement values or data points of the dataset and, thus, may be a derived from the dataset by mathematical operations such as calculations. Specifically, the digital performance feature is derived from the dataset by an automated algorithm, e.g., by a computer program which automatically derives the digital performance feature from the dataset of measurements when tangibly embedded on a data processing device feed by the said dataset.

[0056] The “determining the HDDMS” particularly relates to determining quantitative values representing intra-test fluctuations also referred to as intra-performance fluctuations in a dataset of fine motoric and motoric activity measurements.

[0057] The term “accelerometer measurements” as used herein refers to any type of data collected by an accelerometer known in the art, typically an accelerometer of a mobile device determining a coordinate acceleration. According to the present invention, the accelerometer measurements relate to measurement data on the accelerometer magnitude and directional acceleration data. In line with the present invention, typically a series of data points of the accelerometer magnitude and / or of the a directional acceleration data is collected. The series of data points may typically be a time series, i.e. relate to measurements collected over a specified period of time, e.g. while the subject performs fine motoric and motoric activity tests; more typically the data points are collected periodically over a specified period of time, e.g. one data point is collected every 1 / 100thof a second, every 50th of a second, every 30th of second or the like. For accelerometer and touch sensor data measurements may specifically be collected at a frequency of between 100 Hz and 30 Hz, such as at 100 Hz, 50 Hz, 32 Hz or 30 Hz. Even more typically, the series of data points relates to measurements of iterations of fine motoric and / or motoric tasks performed by the subject during fine motoric and motoric activity tests. In line with the method according to the invention, the accelerometer preferably measures acceleration in the x-, y-, and z- direction. The z-acceleration is, typically, the acceleration in the direction perpendicular to the touch screen. Moreover, In line with the method according to the invention the acceleration magnitude may be determined as the square root of the sum of the squared acceleration in each direction: sqrt(xA2+yA2+zA2) . Typically, the data points are processed by suitable statistical measures known in the art to derive the digital performance features suitable in the present invention. Particularly suitable as a digital performance feature according to the invention is the standard deviation calculated for a series of data points collected by the accelerometer, such as the standard deviation of the accelerometer magnitude and / or the standard deviation of the z-acceleration di- vided by the accelerometer magnitude. Moreover, the coefficient of variation of step impulse is suitable as a digital performance feature according to the present invention.

[0058] The term “touch sensor measurements” as used herein refers to any type of data collected by or collectable by a touch sensor known in the art, typically a touch sensor or touch screen of a mobile device. According to the present invention, the touch screen measurements typically relate to data representing the speed at which a subject is able to touch the touch screen surface or touch sensor with a finger. This is more typically referred to as “tapping speed”; the tapping speed may be calculated from the number of touch interactions of the finger of the subject with the touch screen surface within a specified period of time. Hence, even more typically, the calculation of tapping speed requires parallel recordings of the time. The standard deviation of the time between the onset of one tap and the onset of the subsequent tap may be calculated and may be used as a digital performance feature according to the present invention. Typically, a “tap” may be understood as the contiguous period during which the touch sensor registers an interaction with the finger of the subject, typically with the index finger of the subject as referred to herein.

[0059] The term “measurements of time” as used herein refers to any type of time recordings collected by or collectable by a chronometer known in the art, typically a chronometer of a mobile device. Recordings of time are typically required in all of the digital performance features described herein. Time recordings are more typically required to determine the specified period of time in each of the fine motoric and motoric activity tests, e.g. the total duration of the test. In line with the method according to the invention, time recordings are taken for collecting data points of time series as described elsewhere herein. The step of “comparing the determined HDDMS to a reference” as used herein typically refers to any type of comparison between the value determined in the HDDMS with a reference value. The term “reference” as used in accordance with the present invention relates to a reference value which may in particular represent an HDDMS value determined for a healthy subject, or for a subject at stage 0 HD-ISS. More particularly, the reference value may be an HDDMS value previously determined for the same subject. For example for the same subject determined at the baseline visit of a clinical study, or in other words, the HDDSM determined at the beginning of an assessment period. The “same subject” particularly refers to a subject that HD or the progression of HD has been assessed in by the method according to the invention previously, i.e. several weeks or months before, and is now being assess again. Hence, according to the present invention, the reference typically is a HDDMS of a healthy subject or a cohort of healthy subjects, or a previously determined HDDMS of the same subject. In line with the method according to the invention the subject is typically a subject diagnosed with HD, preferably a HD patient. More typically, the subject is known to suffer from HD, i.e. has been diagnosed with HD.

[0060] Alternatively, the reference may relate to a HDDMS of the subject before a treatment, or while on the treatment with a certain dosage, or after the treatment. This type of reference may be particularly suitable for assessing HD in a subject in response to drug treatment and for therapy recommendation.

[0061] The possibility of choosing a previously determined HDDMS of the same subject as a reference is particularly advantageous as inter-individual fluctuations can be minimized. In particular, the progression in HD may be determined on an individualized scale which may enable determining disease progression much earlier.

[0062] The comparison of the determined HDDMS to a reference in particular involves determining a change or a difference between the determined HDDMS and the reference value, more particular a statistically significant change or difference. Whether a change or difference is statistically significant can be determined without further ado by the person skilled in the art using various well known statistic evaluation tools, e.g., determination of confidence intervals, p-value determination, Student's t-test, Mann-Whitney test, etc.. Details may be found in Dowdy and Wearden, Statistics for Research, John Wiley & Sons, New York 1983. Typically envisaged confidence intervals are at least 50%, at least 60%, at least 70%, at least 80%, at least 90%, at least 95%. The p-values are, typically, 0.2, 0.1, 0.05. The HHDMS is typically suitable for assessing all stages of HD, HD-ISS Stage 0 to 3, in particular assessing progression in all stages of HD. However, the HDDMS is particularly suitable for assessing HD, more particularly progression in HD, in subjects in HD-ISS Stage 2 and / or in subjects in mild to moderate HD-ISS Stage 3. Typically, a change or difference in the determined HDDMS compared to the reference indicates HD, typically progression in HD, more typically an increase in the determined HDDMS compared to the reference indicates progression in HD. The reference value may be considered as a threshold. Hence, in particular in case the determined HDDMS exceeds the reference, this is typically indicative for HD, more typically indicative for progression in HD. More particularly, in case the determined HDDMS is identical with the reference or falls below the reference, this is typically indicative for no progression in HD or for stagnation in HD. Preferably, the HDDMS is used for assessing a subject previously diagnosed with HD. It is more preferably used for assessing motor symptoms of HD such as progression, stagnation, or absence of motor symptoms in HD.

[0063] Preferably, the step of determining the HDDMS, the step of comparing the HDDMS to a reference and / or the step of assessing HD, is performed on a data processing device different from the mobile device carrying out the fine motoric and motoric activity measurements.

[0064] Further details are described elsewhere herein.

[0065] Comparing the determined HDDMS to a reference may in particular be achieved by an automated comparison algorithm implemented on a data processing device such as a computer. Compared to each other are the values of a determined HDDMS and a reference for said HDDMS as specified elsewhere herein in detail. As a result of the comparison, it can be assessed whether the determined HDDMS is identical or differs from or is in a certain relation to the reference (e.g., is higher or lower than the reference). Based on said assessment, the subject can be identified as suffering from or exhibiting symptoms of Huntington's dis- ease or as suffering from progression in HD (“rule-in”), or not (“rule-out”). For the assessment, the kind of reference will be taken into account as described elsewhere in connection with suitable references according to the invention.

[0066] Based on the comparison of the determined HDDMS and the reference HD, typically pro- gression in HD, is assessed. Hence, the term “assessing HD based on said comparison” as used herein typically refers to the result of said comparison.

[0067] The assessment made by the method of the present invention may be communicated to the subject or preferably to another person, such as a medical practitioner or a trained HD rater. Typically, this is achieved by displaying the diagnosis on a display of the mobile device or preferably, on a display or screen of the further device. Alternatively, a recommendation for a therapy, such as a drug treatment, or for a certain life style, e.g. rehabilitation measures, may be provided automatically to the subject or other person. To this end, the established assessment is compared to recommendations allocated to different assessments in a database. Once the established assessment matches one of the stored and allocated assessments, a suitable recommendation can be identified due to the allocation of the recommendation to the stored assessment matching the established assessment. Accordingly, it is, typically, envisaged that the recommendations and assessments are present in form of a relational database. However, other arrangements which allow for the identification of suitable recommendations are also possible and known to the skilled artisan. The determined HDDMS may also be stored on the mobile device or communicated to the subject, or preferably to another person, typically, in real time. The stored HDDMS may be assembled into a time course or similar evaluation measures. Such determined HDDMS may be provided to the subject as a feedback for the course of Huntington's disease investigated in accordance with the method of the invention. Typically, such a feedback can be provided in electronic format on a suitable display of the mobile device and can be linked to a recommendation for a therapy as specified above or rehabilitation measures.

[0068] Further, the determined HDDMS may also be provided to medical practitioners in doctor's offices or hospitals as well as to other health care providers, such as, developers of diagnostic tests or drug developers in the context of clinical trials, health insurance providers or other stakeholders of the public or private health care system.

[0069] The term “mobile device” as used herein is a broad term and is to be given its ordinary and customary meaning to a person of ordinary skill in the art and is not to be limited to a spe- cial or customized meaning. The term specifically may refer, without limitation, to a mobile electronics device, more specifically to a mobile communication device such as a cell phone or smart phone or tablet computer. As used herein the term typically refers to any portable device which comprises a sensor and data-recording equipment suitable for obtaining the dataset of fine motoric and motoric activity measurements. Typically, the mobile device comprises at least one sensor for measuring the activity; more typically, the mobile device comprises at least an accelerometer, a touch sensor and / or a chronometer. This may also require a data processor and storage unit as well as a display for electronically simulating an activity test on the mobile device. Moreover, in particular from the activity of the subject, data shall be recorded and compiled to a dataset, which is to be evaluated by the method of the present invention either on the mobile device itself or on a further device, different from said mobile device. A device different from said mobile device may typically relate to a second mobile device or any other suitable data processing device as described elsewhere herein. The mobile device may more particularly comprise data transmission equipment in order to transfer the acquired dataset from the mobile device to one or more further devices. Particular well suited as mobile devices according to the present invention are smartphones, smartwatches, wearable sensors, portable multimedia devices or tablet computers.

[0070] Hence, in line with the present invention, the mobile device specifically is a smartphone, smartwatch, wearable sensor, portable multimedia device or tablet computer; more specifically a smartphone. Alternatively, portable sensors with data recording and, optionally, processing equipment may be used. Further, depending on the kind of fine motoric or motoric activity test to be performed, the mobile device shall be adapted to display instructions for the subject regarding the activity to be carried out in the test. Particular envisaged activity tests to be carried out by the subject are described elsewhere herein and encompass the following tests: balance test (BAL), chorea test (CHO), two-minute walk test (TMW), speeded tapping test (STT).

[0071] Additionally or alternatively, as will be outlined in further detail below, the mobile device may also refer to a tablet computer or another type of portable computer having at least one or more sensors. According to the present invention, the mobile device typically comprises at least one, more typically all of the following sensors or functions: an accelerometer, a touch sensor, a chronometer.

[0072] The term "data processing device" generally refers to an arbitrary device adapted to perform the method step(s) as described above, in an embodiment by using at least one processor, and / or at least a database and / or at least one application-specific integrated circuit. Thus, as an example, the data processing device may comprise a software code stored thereon comprising a number of computer instructions. The data processing device may provide one or more hardware elements for performing one or more of the indicated operations and / or may provide one or more processors with software running thereon for performing one or more of the method steps.

[0073] Specifically, said fine motoric activity measurements of fine motor control of the fingers are carried out during tapping movements of a finger of each hand in a bimanual test. A test particularly suitable for fine motoric activity measurements of fine motor control of the fin- gers in line with the present invention is the speeded tapping test (STT). More specifically, said motoric activity measurements of motor control of lower limbs, trunk and upper limbs are carried out during walking, balancing or assessing of chorea of said subject. A test particularly suitable for motoric activity measurements of motor control of lower limbs is the two-minute walk test (TMW); a test particularly suitable for motoric ac- tivity measurements of motor control, in particular chorea, of the trunk is the balance test (BAL); a test particularly suitable for motoric activity measurements of motor control of the upper limbs is the chorea test (CHO).

[0074] Preferably, assessing progression in HD comprises detecting change in HD over time. Therefore, in line with the present invention, typically, steps a) to c) are repeated over time, more typically steps a) to c) are repeated once per day or once per week for at least 20 weeks up to two years. Alternatively, suitable schedules may be employed for repeating of steps a) to c) such as once per day or once every other day for a duration of 1 to 5 weeks which may then be repeated at an interval of for example 1.5, 6, 9, 12, 18, and 24 months. This may be timed in accordance with clinical visits or assessment by medical practioners.

[0075] Said repetition of steps a) to c) may be considered as a series of assessing HD or a series of assessing the progression of HD. This is advantageous as it allows for a detailed monitoring of HD over time

[0076] The method according to the invention may involve a series of assessing HD, typically the progression of HD, of the same subject, wherein the first determined HDDMS is used as a reference in the subsequent assessing of HD or progression of HD. A further aspect of the invention relates to assessing HD for monitoring the response of a subject to a treatment, such as to Huntingtin-lowering therapies. Thus, in that case the reference may typically be a HDDMS of the subject determined before the treatment, or while on the treatment with a certain dosage, or after the treatment. Typically, the result of such assessment can be provided in electronic format on a suitable display of the mobile device and can be linked to a recommendation for a therapy or dosage, or rehabilitation measures.

[0077] Typically, the method of the present invention for assessing Huntington's disease in a subject may be carried out as follows: First, at least one HDDMS is determined based on a multitude of digital performance features derived from accelerometer measurements, touch sensor measurements, and / or measurements of time in a preexisting dataset of fine motoric and motoric activity measurements from said subject.

[0078] Said preexisting dataset is a dataset of fine motoric and motoric activity measurements from said subject; said measurements are carried out using a mobile device. Said dataset may be transmitted from the mobile device to a further data processing device, such as a computer, or may be processed in the mobile device in order to determine the HDDMS.

[0079] Second, the determined HDDMS is compared to a reference by, e.g., using a computer-implemented comparison algorithm carried out by the data processor of the mobile device or by the data processing device, e.g., the computer. Third, the result of the comparison is assessed with respect to the reference used in the comparison and based on the said comparison Huntington's disease in the subject will be assessed.

[0080] The said assessment may be communicated to the subject or to another person, such as a medical practitioner, on a suitable display such as a screen connected or implemented in the mobile device or the data processing device.

[0081] Alternatively, a recommendation for a therapy, such as a drug treatment, or for a certain life style, is provided automatically to the subject or other person. To this end, the established assessment is compared to recommendations allocated to different assessments in a database.

[0082] Once the established assessment matches one of the stored and allocated assessments, a suitable recommendation can be identified due to the allocation of the recommendation to the stored assessment matching the established assessment. Yet as an alternative or in addition, the determined HDDMS may be stored on the mobile device. Typically, it may be evaluated together with other previously determined HDDMS by suitable evaluation tools over time, such as time course assembling algorithms, implemented on the mobile device which can assist electronically rehabilitation or therapy recommendation as specified elsewhere herein. The present invention further contemplates a device comprising a processor and a database as well as software which is tangibly embedded to said device and, when running on said device, carries out the method according to the present invention. Moreover, the present invention relates to a system comprising a mobile device comprising at least one sensor and a further device comprising a processor and a database as well as software which is tangibly embedded to said device and, when running on said device, carries out the method according to the invention, wherein said mobile device and said remote device are operatively linked to each other.

[0083] Still further disclosed as the present invention is the use of said device or said system for assessing HD, typically for assessing progression of HD in a subject using a Huntington’s disease digital motor score based on a multitude of digital performance features derived from a dataset of fine motoric and motoric activity measurements from said subject.

[0084] Accordingly, the method of the present invention may be used for: assessing HD, in particular progression HD; assessing progression of motor symptoms of HD; monitoring patients, in particular, in a real life, daily situation and on large scale; supporting patients with life style and / or therapy recommendations; investigating drug efficacy, e.g. also during clinical trials; facilitating and / or aiding therapeutic decision making; supporting hospital managements; supporting rehabilitation measure management; - improving the disease condition as a rehabilitation instrument stimulating higher density cognitive, motoric and walking activity supporting health insurances assessments and management; and / or supporting decisions in public health management. Furthermore, the present invention relates to a computer program adapted to perform the method of the invention and to a computer-readable storage medium comprising said computer program.

[0085] The invention, in light of the above, also specifically contemplates method for developing a digital score for assessing progression of HD, comprising:

[0086] - determining digital performance features from a dataset of fine motoric and motoric activity measurements; - ranking said digital performance features;

[0087] - selecting best performing digital performance features;

[0088] - building a digital score as a weighted composite score by exploratory factor analysis; and

[0089] - validating said digital score by confirmatory factor analysis.

[0090] Advantageously, it has been found in the studies underlying the present invention that the HDDMS robustly measures HD disease status and progression, and enables smaller and thus shorter clinical trials for early decision-making on clinical efficacy of novel HD therepies. Moreover, the HDDMS is advantageous as it has favourable characteristics, including reliability (intraclass correlation coefficient >0.95), correlation with the composite Unified Huntington’s Disease Rating Scale (cUHDRS) (r=-0.5), and better sensitivity to change (STC) than the cUHDRS. In a post-hoc analysis of GENERATION HD 1, the STC of HDDMS at Week 20 was comparable to that of the cUHDRS at Week 68. Thus, the HDDMS promises substantial reduction in sample size in clinical trials.

[0091] The definitions and explanations of the terms made above apply mutatis mutandis for the following embodiments except as specified otherwise.

[0092] Summarizing and without excluding further possible embodiments, the following embodiments may be envisaged:

[0093] 1. A computer-implemented method for assessing Huntington's disease (HD), typically progression in HD, in a subject comprising the steps of a) determining a Huntington’s disease digital motor score (HDDMS) based on a multitude of digital performance features derived from at least one or more of the following: accel- erometer measurements, touch sensor measurements, and / or measurements of time in a dataset of fine motoric and motoric activity measurements from said subject; b) comparing the determined HDDMS to a reference; and c) assessing HD, typically progression in HD, in the subject based on said comparison. 2. The method of embodiment 1, wherein the HDDMS comprises at least three digital performance features, typically at least four, more typically four digital performance features derived from a dataset of fine motoric and motoric activity measurements from said subject. 3. The method of any one of the preceding embodiments, wherein the HDDMS comprises at least one digital performance feature derived from a dataset of fine motoric activity measurements and at least two, typically three digital performance features derived from datasets of motoric activity measurements.

[0094] 4. The method of any one of the preceding embodiments, wherein said fine motoric and motoric activity measurements are carried out using a mobile device. 5. The method of any one of the preceding embodiments, wherein said dataset of fine motoric and motoric activity measurements comprises measurements of fine motor control of the fingers and measurements of motor control of lower limbs, trunk and upper limbs; preferably measurements of fine motor control of the fingers, measurements of motor control of lower limbs and measurements of chorea of trunk and upper limbs.

[0095] 6. The method of any one of the preceding embodiments, wherein said fine motoric activity measurements of fine motor control of the fingers are carried out during tapping movements of fingers in a bimanual test.

[0096] 7. The method of any one of the preceding embodiments, wherein said motoric activity measurements of motor control of lower limbs, trunk and upper limbs are carried out during walking, balancing or assessing of chorea of said subject.

[0097] 8. The method of any one of the preceding embodiments, wherein said dataset of fine mo- toric and motoric activity measurements has been collected from the subject while carrying out fine motoric or motoric activity tests; typically activity tests assessing balance, gait, finger movement accuracy, and / or movement speed; more typically these activity tests are one or more, up to all of the following fine motoric and motoric activity tests: Balance Test, Chorea Test, Speeded Tapping Test, Two-Minute Walk Test.

[0098] 9. The method of any one of the preceding embodiments, wherein said digital performance features comprise quantitative values; typically wherein the quantitative values comprise, more typically consist of, statistical values derived from said accelerometer measurements, touch sensor measurements, or measurements of time.

[0099] 10. The method of the preceding embodiment, wherein said statistical values comprise two or more of the following: standard deviation of the accelerometer magnitude, typically derived from the balance test; absolute value of the standard deviation of the z-acceleration divided by the accelerometer magnitude typically derived from the chorea test; - standard deviation of the time between the onset of one tap and the onset of the subsequent tap typically derived from the speeded tapping test; coefficient of variation of step impulse typically derived from the two-minute walk test. 11. The method of any one of the preceding embodiments, wherein the subject is a subject diagnosed with HD, preferably a patient.

[0100] 12. The method of any one of the preceding embodiments, wherein the reference is a HDDMS of a healthy subject or a previously determined HDDMS of the subject HD is being assessed in.

[0101] 13. The method of any one of the preceding embodiments, wherein the dataset of fine motoric and motoric activity measurements from said subject is a preexisting dataset.

[0102] 14. The method of any one of the preceding embodiments, wherein the step of determining the HDDMS, the step of comparing the HDDMS to a reference and / or the step of assessing HD, is performed on a data processing device different from the device carrying out the fine motoric and motoric activity measurements . 15. The method of any one of embodiments 4 to 13, wherein said mobile device is a smartphone, smartwatch, wearable sensor, portable multimedia device or tablet computer.

[0103] 16. The method of any one of the preceding embodiments, wherein a change or difference in the determined HDDMS compared to the reference indicates HD, typically progression in HD, more typically wherein an increase in the determined HDDMS compared to the reference indicates progression in HD.

[0104] 17. The method of any one of the preceding embodiments, wherein steps a) to c) are repeated over time, typically steps a) to c) are repeated once per day or once per week for at least 20 weeks. 18. The method of any one of the preceding embodiments, wherein said assessing progression in HD comprises detecting change in HD over time.

[0105] 19. The method of any one of the preceding embodiments, wherein the digital performance features in the HDDMS each are weighted according to their ability to represent disease severity.

[0106] 20. A device comprising a processor and a database as well as software which is tangibly embedded to said device and, when running on said device, carries out the method of any one of embodiments 1 to 19.

[0107] 21. A system comprising a mobile device comprising at least one sensor, the system further comprising a device comprising a processor and a database as well as software which is tangibly embedded to said device and, when running on said device, carries out the method of any one of embodiments 1 to 19, wherein said mobile device and said device are operatively linked to each other.

[0108] 22. The system according to embodiment 21, wherein the device further comprised in said system is a device according to embodiment 20.

[0109] 22. Use of the device according to embodiment 20 or the system of embodiment 21 or 22 for assessing HD, typically for assessing progression of HD in a subject using a Huntington’s disease digital motor score based on a multitude of digital performance features derived from a dataset of fine motoric and motoric activity measurements from said subject.

[0110] 23. A computer program adapted to perform the method of any one of embodiments 1 to 19.

[0111] 24. A computer-readable storage medium comprising said computer program of embodiment 23.

[0112] 25. A method for developing a digital score for assessing progression of HD, comprising:

[0113] - determining digital performance features from a dataset of fine motoric and motoric activity measurements ;

[0114] - ranking said digital performance features;

[0115] - selecting best performing digital performance features;

[0116] - building a digital score as a weighted composite score by exploratory factor analysis; and

[0117] - validating said digital score by confirmatory factor analysis. All references cited throughout this specification are herewith incorporated by reference with respect to their entire disclosure content and with respect to the specific disclosure contents mentioned in the specification.

[0118] Figures

[0119] Figure 1 visualizes the development of the HDDMS. From the six motor tasks (BAL, CHO, DAS, STT, TMW, UTU), an average of 60 performance features were extracted, which were ranked according to reliability, convergent validity, and sensitivity to change. For each task, one feature was selected that had a robustly high performance in both the development and validation dataset according to two ranking criteria. Subsequently, the HDDMS was constructed using factor analysis, which revealed that only four of the original six assessments were required to create a sensitive and valid score.

[0120] BAL, Balance Test; CHO, Chorea Test; DAS, Draw a Shape Test; HDDMS, Huntington’s Disease Digital Motor Score; STT, Speeded Tapping Test; TWM, Two-Minute Walk Test; UTT, U-Turn Test. Bold: tests included in the final HDDMS.

[0121] Figure 2 shows how the data were used in the different steps of the analysis plan summarized in (A) The HDDMS was derived in two steps: 1) feature ranking and selection, 2) score building and validation. (B) Data collected from the different studies were divided into a development dataset (i.e., data from HD Natural History and OLE of the tominersen Phase Ella study) and validation dataset (slice of GENERATION HD1).

[0122] Data collected in Digital-HD kept separately to assess the features’ validity (i.e., to ensure that only features that show worsening in response to disease progression are kept in the analysis). The timeline (arrows) indicates the order in which the development, validation and final analysis steps were conducted.

[0123] Figure 3 depicts cross-sectional Pearson’s correlations between HDDMS and clinical anchors using data from GENERATION HD1. (A) Cross-sectional correlations between HDDMS (x-axis) and cUHDRS (y-axis) across clinical visits. Pearson’s correlation coefficient is provided in the top right corner of each panel. The linear regression is indicated by the grey line. (B) Pearson’s correlation at baseline between the HDDMS or digital features and various HD clinical anchors. The legend on the right indicates the Pearson’s correlation coefficient. Correlations for different study periods are reported in Supplementary Fig. 9. *p<0.001. BAL, Balance Test; CHO, Chorea Test; cUHDRS, composite Unified Huntington’s Disease Rating Scale; EQ-5D-5L, EuroQol 5-dimension 5-level; HDDMS, Huntington’s Disease Digital Motor Score; r, Pearson correlation coefficient; SDMT, Symbol Digit Modalities Test; STT, Speeded Tapping Test; SWRT, Stroop Word Reading Test; TFC, Total Functional Capacity; TMS, Total Motor Score (31 items); TMS-4, short version of the Total Motor Score; TMW, Two-Minute Walk; UHDRS, Unified Huntington’s Disease Rating Scale.

[0124] Figure 4 shows the sensitivity to change of HDDMS, clinical anchors and top features in GENERATION HD1 estimated by a mixed model for repeated measures. (A) and (B) compare the sensitivity to change between HDDMS and HD clinical anchors in people with HD-ISS Stage 2 and HD-ISS Stage 3 disease, respectively (cUHDRS, SDMT, SWRT and TFC were multiplied by -1 so that an increase on the y-axis corresponds to a worsening of the disease status. (C) and (D) compare the sensitivity of change between HDDMS and top features selected by the feature ranking process instead.

[0125] BAL, Balance Test; CHO, Chorea Test; cUHDRS; composite Unified Huntington’s Disease Rating Scale; HD, Huntington’s disease; HDDMS, Huntington’s Disease Digital Motor Score; HD-ISS, Huntington’s Disease Integrated Staging System; SDMT, Symbol Digit Modalities Test; STT, Speeded Tapping Test; SWRT, Stroop Word Reading Test; TFC, Total Functional Capacity; TMS, Total Motor Score (31 items); TMS-4, short version of the Total Motor Score; TMW, Two-Minute Walk;.

[0126] Figure 5 depicts HDDMS properties. (A) The mean change from baseline (±SEM) in HDDMS is shown for the placebo arm of the GENERATION HD1 study (red) and the HD Natural History Study (blue). (B) In the full dataset, test- retest reliability (ICC [±95% CI]) remained high across all study weeks. (C) This Quantile - Quantile plot compares the HDDMS quantiles obtained for the placebo arm of the GENERATION HD 1 study (left panel) and for the HD Natural History Study (right panel) at baseline against a normal distribution. (D) Known-groups validity of the HDDMS are presented for gene-negative control volunteers, people with pre-manifest HD and people with manifest HD enrolled in the Digital-HD study. The notch represents the 95% CI of the median (middle horizontal line).

[0127] CI, confidence interval; GenHDl, GENERATION HD1; HD, Huntington’s Disease; HD NHS, Huntington’s Disease Natural History Study; HDDMS, Huntington’s Disease Digital Motor Score; ICC, intraclass correlation coefficient; SEM, standard error of mean.

[0128] Figure 6 shows the estimated number of subjects per arm (on a log 10 scale) required to detect a 40% effect size difference with a power of 0.8 and an alpha of 0.2, 01 or 0.05 when using the HDDMS or the cUHDRS for a study duration of 36, 52 or 68 in HD-ISS Stage 2 (left panel) or Stage 3 disease (right panel). cUHDRS, composite Unified Huntington’s Disease Rating Scale; HD-ISS, Huntington’s Disease Integrated Staging Scale; HDDMS, Huntington’s Disease Digital Motor Score.

[0129] Figure 7 The plots depict the number of digital motor test performed each week by the participants enrolled in OLE of the tominersen Phase I / IIa study (A), HD Natural History Study (B) and GENERATION HD1 (C), respectively. Adherence rules were defined as such: The participants performed at least number of tests per fortnight. For example, a participant performing at least seven test instances during the fortnight will be in the Rule Six, whereas a participant performing at least three test instances during the fortnight will be in Rule Three. BAL, Balance Test; CHO, Chorea Test; HD, Huntington’s Disease; OLE, open-label extension; STT, Speeded Tapping Test; TMW, Two-Minute Walk; UTU, U-Turn Test.

[0130] Figure 8 Quantile-Quantile plots of the HDDMS versus the normal distribution for the placebo arm of GENERATION HD1 (left panels) and HD Natural History Study (right panels) at Week 0 (A) and Week 68 (B) for the full dataset Gen HD1, GENERATION HD1; HDDMS, Huntington’s Disease Digital Motor Score; HD NHS, Huntington’s Disease Natural History Study.

[0131] Figure 9 Cross-sectional Pearson’s correlations between HDDMS and the digital features selected by the feature ranking process and HD clinical anchors by clinical visit. Data are presented for GENERATION HD1. Legend indicates the Pearson’s correlation coefficient.*p<0.001. BAL, Balance Test; CHO, Chorea Test; cUHDRS, composite Unified Huntington’s Disease Rating Scale; EQ-5D-5L, EuroQol 5-dimension 5-level; HDDMS, Huntington’s Disease Digital Motor Score; SDMT, Symbol Digit Modalities Test; STT, Speeded Tapping Test; SWRT, Stroop Word Reading Test; TFS, Total Functional Capacity; TMS, Total Motor Score (31-items); TMS-4, shortened version of the Total Motor Score consisting of 23 of the 31 TMS item;

[0040] TMW, Two- Minute Walk.

[0132] Figure 10 The scatter plots present the comparison between test-retest reliability (mICC) and sensitivity to change (mSTC) for each digital feature derived from the motor tests. The line corresponds to a simple linear regression fit. However, due the ceiling effect on the mICC, features with high mICC do not correlate with mSTC. Data are presented for GENERATION HD1.

[0133] BAL, Balance Test; CHO, Chorea Test; mICC, mean intraclass correlation coefficient; mSTC, mean sensitivity to change; STT, Speeded Tapping Test; TMW, Two-Minute Walk; UTU, U-Turn Test.

[0134] Figure 11 The digital features selected by the feature ranking process showed a stable linear between-feature relationships in GENERATION HD 1. (A) and (B) displays the between-feature relationship at Week 0 and Week 68, respectively. The line represents a linear model fitted on the data. (C) and D) present the standardised feature values distributions at Week 0 and Week 68, respectively.

[0135] BAL, Balance Test; CHO, Chorea test; STT, Speeded Tapping Test; TMW, Two-Minute Walk; UTU, U-Turn test.

[0136] Figure 12 Quality metric components computed for each feature in the feature ranking process. For each feature derived from each of the motor tests, the mean sensitivity to change (x-axis), mean cross-sectional correlation with the cUH- DRS (y-axis), and test-retest reliability (mICC; indicated by the colour bar) are presented for data from GENERATION HD 1.

[0137] BAL, Balance Test; CHO, Chorea Test; cUHDRS, composite Unified Huntington’s Disease Rating Scale; mICC, mean intraclass correlation coefficient; STT, Speeded Tapping Test; TMW, Two-Minute Walk; UTU, U-Turn Test. Figure 13 (A) and (B) show the comparison between the feature ranking calculated from the development (x-axis) and the validation (y-axis) datasets with the product method and the stepwise method, respectively. (C) shows the comparison between the stepwise (x-axis) and product methods (y-axis). The point size and colour refer to mSTC and mCSC (cross-sectional correlation with the composite Unified Huntington’s Disease Rating Scale), respectively. Only features with a mICC >0.08 are included in the plots.

[0138] BAL, Balance Test; CHO, Chorea Test; mCSC, mean cross-sectional correlation coefficient; mICC, mean intraclass correlation coefficient; mSTC, mean sensitivity to change; STT, Speeded Tapping Test; TMW, Two-Minute Walk; UTU, U-Turn Test.

[0139] Figure 14 Characteristics of the best features of each motor test selected by the feature ranking process for GENERATION HD1. (A) MMRM mean estimates and 95% CI are shown for the change from baseline. (B) Boxplots present the baseline feature values for gene-negative control volunteers, people with premanifest HD and people with manifest HD enrolled in in the Digital HD study. The notch represents the 95% CI of the median (middle horizontal line). (C) The scatter plots present the cross-sectional correlation between feature values (x-axis) and cUHDRS (y-axis) for each of the clinical visits (Weeks 0, 4, 20, 36, 52 and 68). The line represents a linear regression, and the Spearman’s rank correlation coefficient is provided in the top right comer of each panel. (D) ICC estimates and their 95% CI are plotted for consecutive 2-week periods. * p<0.001.

[0140] BAL, Balance Test; CHO, Chorea Test; CI, confidence internal; cUHDRS; composite Unified Huntington’s Disease Rating Scale; HD, Huntington’s disease; ICC, intraclass correlation coefficient; MMRM, mixed models for repeated measures; STT, Speeded Tapping Test; TMW, Two-Minute Walk; UTU, U-Tum Test.

[0141] Figure 15. The quality metric components values obtained at the beginning and end of data collection are depicted for (A) mean intraclass correlation coefficient (test-retest reliability), (B) mean sensitivity to change, and (C) mean cross- sectional correlations with the composite Unified Huntington’s Disease Rating Scale. For the mean sensitivity to change, the values at the beginning corresponded to the averaged sensitivity to change from Weeks 4, 6 and 8 and the end values corresponded to the averaged sensitivity to change from Weeks 64, 66 and 68. Data are shown for GENERATION HD 1. BAL, Balance Test; CHO, Chorea Test; STT, Speeded Tapping Test; TMW, Two-Minute Walk; UTU, U-Turn Test.

[0142] Figure 16. Longitudinal Spearman rank correlation are shown between a given feature (facet title) and all the other features (indicated by the coloured line curves) selected by the feature ranking process. Data are shown for GENERATION HD1. BAL, Balance Test; CHO, Chorea Test; STT, Speeded Tapping Test; TMW, Two-Minute Walk; UTU, U-Turn Test.

[0143] Figure 17 Four different measurement model hypotheses were tested in the confirmatory factor analysis. The index ‘ 1’ indicates a loading of 1. The double arrow in Model 2 indicates that covariance between the two LV was also modelled. BAL, Balance Test; CHO, Chorea Test; LV, latent variable; STT, Speeded Tapping Test; TMW, Two-Minute Walk; UTU, U-Turn Test.

[0144] Figure 18 The HDDMS was calculated by applying Thomson Regression on the latent variables obtained with Model 3 (see Figure 17) refined on the full dataset. (A) The individual scores (HDDMS) were computed by multiplying the feature values by their respective W. (B) The W were obtained by multiplying inverse of the between-feature correlations averaged over time with measurement model’s loadings. The standardized loadings, which correspond to the correlation between the loading and the latent variable, are also presented. These indicate that the HDDMS explains between 44% (STT) and 56% (CHO) of the feature variance (explained variance = standardised loadings2). (C) The correlation matrix shows the between- feature correlations averaged over all weeks.

[0145] BAL, Balance Test; CHO, Chorea Test; HDDMS, Huntington’s Disease Digital Motor Score; r, Pearson’s r; STT, Speeded Tapping Test; TMW, Two- Minute Walk Test; W, weights.

[0146] Figure 19 Initial measurement fit model tested in the confirmatory factor analysis at a time when not all data were available for analysis yet, and before certain improvements in the feature calculation process. While the specific factor weights have changed a bit, the best performing model for the HDDMS has remained the same, underlying the robustness of the score. (A) Measurement model with loading of 1 for BAL feature. (B) fit indices of measurement model with strong invariance over time (fixed intercept and fixed loadings). (C) loadings and individual score weights.

[0147] Examples

[0148] Example 1 - Methods

[0149] Studies and participants

[0150] Data from four studies that deployed our digital monitoring platform were used: the HD Natural History Study (NCT03664804); open-label extension (OLE) of the tominersen Phase I / IIa study (NCT03342053); GENERATION HD1 (NCT03761849),

[0014] a Phase 3 trial of tominersen; and Digital-HD, a dedicated observational study of the digital monitoring platform. All studies enrolled people with manifest HD. Digital-HD additionally enrolled gene-negative control volunteers and people with premanifest HD. Key inclusion and exclu- sion criteria are reported in Table 4. All people with either premanifest or manifest HD were clinically assessed by trained raters using the cUHDRS

[0015] ,

[0151] Table 4. Key inclusion and exclusion criteria.

[0152] HD Natural History Open-label e of the Digital-HD GENERATION HD1

[0153] Study tominersen Phase People with manifest People with premani- Gene-negative con- l / lla Study HD fest HD trol volunteers

[0154] Inclusion criteria • Diagnosis of early • Diagnosis of early • DCL = 4 • DCL <4 • No known family • Diagnosis of mani- manifest HD manifest HDa• Stage l-lll (UHDRS- • CAG expansion history of HD or fest HD

[0155] • Shoulson-Fahn • Shoulson-Fahn TFC 4-13) >40 CAG expansion • DCL = 4 stage l / ll disease stage 1 disease • CAG expansion <36 • Independence

[0156] • UHDRS-TFC score (UHDRS-TFC score >36 scale score >70

[0157] 7-13) 11— 13)a• Age 18-75 years • CAP score >400

[0158] • Completion of treatment period of the Phase l / lla study

[0159] Exclusion criteria • Any serious medi- • Any new condition • Any serious medical condition of laboratory finding that, in the in- • Any serious medi- cal condition of la- or worsening of vestigator's judgment, precludes the safe participation in and com- cal condition of la- boratory finding existing condition pletion of the study boratory finding that, in the investi- that could make • Inability or unwillingness to undertake any of the essential study that, in the investigator's judgment, the study partici- procedures gator's judgment, precludes the safe pant unsuitable for • Current use of investigational drug or participation in a clinical drug precludes the safe participation in participation or in- trial within 30 days prior to sampling visit participation in and completion of terfere with the • Current intoxication, or drug or alcohol abuse / dependence and completion of the study study participant • Use of inappropriate or unstable dose of any antidepressant, psy- the study

[0160] • Pregnancy, breast- participating in choactive, psychotropic or other medication or nutraceuticals used • Pregnancy, breastfeeding, or inten- and / or completing to treat HD feeding, or intention of becoming the study tion of becoming pregnant during pregnant during the study the study or within

[0161] • Current or previ- 5 months after the ous use of an anti- final dose of the sense oligonucleo- study drug tide (including

[0162] small interfering Current or previRNA) ous use of an anti¬

[0163] Current use of ansense oligonucleotipsychotics pretide (including scribed for psychosmall interfering sis, cholinesterase RNA) inhibitors, memanCurrent use of antine, amantadine, tipsychotics preor riluzole includscribed for psychoing use within 12 sis, cholinesterase weeks of enrollinhibitors, memanment tine, amantadine,

[0164] Treatment with an or riluzole includinvestigational ing use within 12 drug within 30 weeks of enrolldays prior to ment screening or 5 Treatment with an half-lives of the ininvestigational vestigational drug, drug within 30 whichever is days prior to longer screening or 5 half-lives of the investigational drug, whichever is longer aAt enrolment in the Phase IZIIa study.

[0165] CAG, cytosine-adenine-guanine; CAP, CAG-Age-Product; DCL, diagnostic confidence level; HD, Huntington’s Disease; UHDRS-TFC, Unified Huntington’s Disease Rating Scale Total Functional Capacity.

[0166] 5

[0167] Classification of disease severity according to HD-ISS was performed post-hoc where possible, since the initiation of these studies preceded its publication [6], Clinical assessments were taken at baseline and regularly thereafter. Digital tests were administered in clinic and collected regularly at home in the participants’ daily lives, using the digital monitoring platform, from baseline until end of study. The schedule of assessment and the duration of each of these studies are reported in Table 5.

[0168] Table 5. Study duration and schedule of assessment.

[0169] HD Natural History Open-label exten- Digital-HD GENERATION HD1

[0170] Study sion of the tominersen Phase l / lla Study

[0171] Study duration ~61 weeks 62 weeks 18 months 69 weeks3

[0172] Clinical assessments Baseline, and ap- Baseline, and ap- Baseline, Month 12, Baseline, and approximately every 3 proximately every 3 Month 18 proximately every 3 months months months

[0173] BAL Daily Daily Daily Once every other day

[0174] CHO Daily Daily Daily Once every other day

[0175] STT Daily Daily Daily Once every other day

[0176] TWM Daily Daily Daily Once per week

[0177] UTU Daily Daily Daily Once every other day

[0178] DASbDaily Daily Daily Once every other dayaThe study duration was initially 101 weeks. However, the dosing was stopped earlier in March 2021 based on the results of a pre-planned review of the data from the Phase III study conducted by an unblinded Independent Data Monitoring Committee (iDMC). The iDMC made its recommendation based on the investigational therapy's potential bene- fit / risk profile for study participants.bThe DAS was not included in the analysis.

[0179] BAL, Balance Test; CHO, Chorea Test; DAS; Draw a Shape Test; STT, Speeded Tapping Test; TWM, Two-Minute Walk Test; UTU, U-Turn Test.

[0180] Datasets

[0181] The development dataset consisted of data from people with manifest Huntington’s disease (HD) enrolled in either the HD Natural History Study and in the open-label extension (OLE) of the tominersen Phase I / IIa study. The validation dataset was a representative slice obtained by randomly sampling the GENERATION HD 1 dataset. The main advantage of using a representative slice of GENERATION HD1 for validation purposes is that sufficient ‘unseen’ data were available should a second validation attempt be necessary (e.g., if any step had to be adjusted in case the first validation had failed, though this turned out not to be necessary). These studies differed from each other in terms of the study cohorts they enrolled, differences in the treatment paradigm of the study participants, and the different contextual environments in which the studies took place (e.g., the COVID-19 pandemic) etc. Consequently, splitting the data into a development and validation dataset along study boundaries meant that these two datasets were relatively heterogeneous, thus further increasing the confidence of the validity of the Huntington’s Disease Digital Motor Score (HDDMS).

[0182] Digital monitoring platform

[0183] The digital monitoring platform consists of smartphone-based active tests (i.e., tests requiring active input by the user), smartphone- and smartwatch-based passive monitoring (i.e., sensor data collected passively without active input by the user on activities of daily living) and patient-reported outcomes

[0013] , The development of the HDDMS focused on active motor tests co-designed with people with HD and neurologists: the Balance Test, Chorea Test, Speeded Tapping Test, Two-Minute Walk Test, U-Turn Test, and Draw a Shape Test. Bimanual tests (i.e., Chorea Test, Speeded Tapping Test and Draw a Shape Test) were performed at each instance with both hands.

[0184] For each motor test, on average 60 different features quantifying the performance on the test were derived from the raw sensor data

[0013] , Bimanual tests included three sets of features (dominant hand, non-dominant hand, and average performance across both hands). Feature values were aggregated by taking the median across 2-week periods [16-18], Features that showed a skewed distribution were additionally log-transformed to facilitate subsequent analyses (see below Processing of digital data).

[0185] For each test, quality control checks were applied to ensure that they were performed according to protocol and were producing analysable data (see below Processing of digital data). For the Draw a Shape Test, it was found that there were an increasing number of failed quality checks as the shapes became more challenging, thus making the analysis of this data more complex. This added analysis complexity, combined with the observation that the Draw a Shape Test was the least favoured motor assessment by patients, led to the decision to exclude the Draw a Shape Test from the development of the HDDMS.

[0186] Processing of digital data

[0187] Checking number of data points The digital features underwent several processing steps before being included in the feature ranking process. Only continuous features or discrete features with sufficient granularity to measure a disease signal with a certain degree of precision were considered for the feature ranking process. Hence, any features with fewer than 10 unique values were disregarded.

[0188] Log transformation of non-normalised features

[0189] Features with skewed distributions were log-transformed to facilitate the downstream analyses. For this, a Shapiro-Wilks test for normalcy was performed on the baseline log-transformed and non-log-transformed features values. If the difference of the W test statistic between the log-transformed and non-log-transformed feature values was >0.005, i.e., the log- transformed distribution was clearly more normal, then the feature was log-transformed. The threshold of 0.005 was selected based on visual inspection of log-transformed and non-log- transformed distributions of randomly selected features. If a feature was log-transformed, it was ensured that the log transformation of ‘0’s resulted in ‘not a number’ values.

[0190] Quality control checks

[0191] For each test, quality control checks were applied to ensure that they were performed according to protocol and were producing analyzable data. For example, an execution of a Chorea Test where the phone remained motionless in horizontal orientation (“phone on the table”) was excluded from further analysis. For the Draw a Shape Test, it was found that there were an increasing number of failed quality checks as the shapes became more complex, resulting from attempts that were too different from the target shape to be analyzable. Consequently, this would have required creating composite features combining both the individual’s ability to complete a shape and their performance completing the shape, as either component alone would have suffered from either ceiling or floor effects. Furthermore, patient satisfaction with the Draw a Shape Test, the test with the longest duration, was below average. The combination of added complexity for analysis and patient burden led to the decision to exclude the Draw a Shape Test from the development of the HDDMS.

[0192] Deriving the Huntington’s Disease Digital Motor Score

[0193] Developing the HDDMS involved several steps (Fig. 1). First, digital features with non- congruent progression were filtered out (see below Filtering out non-congruent features). Non-congruent features were features that showed a longitudinal change in the opposite direction to their known-groups validity (i.e., the difference between gene-negative control volunteers, people with premanifest HD, and people with manifest HD). Subsequently, a feature ranking process was applied to all remaining features to rank them by their measurement properties (i.e., test-retest reliability, agreement with accepted clinical trial endpoints, and sensitivity to change) (see below Filtering out non-congruent features). This ranking was used to select the best performing feature from each motor test. Finally, a factor analysis approach was used to compute a weighted composite score, the HDDMS, based on the features selected by the feature ranking process ( see blow Feature ranking process). Factor analysis is a statistical method to study the effect of unobservable variables, or latent variables, (here: the underlying motor concept related to disease severity) on observed variables (here: the digital features selected by the feature ranking process).

[0194] Filtering out non-congruent features

[0195] Prior to the feature ranking step, digital features with a non-congruent progression were filtered out or excluded. Features were considered non-congruent if the longitudinal change from baseline in the feature value was in the opposite direction to the cross-sectional difference in the feature value across people with different levels of disease progression. To automate this filtering decision, the slope of the longitudinal change in the feature value was compared against the sign of the slope describing the difference in feature value between known groups with different levels of disease burden. If the signs of the two slopes were different, the feature was removed from consideration. The slope describing feature change from baseline was obtained by fitting a linear mixed model with random slope for time (weeks in the study) per subject (lme4 R package).

[0030] The dependent variable was feature change from baseline and the independent variables were weeks, age, cytosine-adenine-gua- nine (CAG) repeats, and feature value at baseline. The slope describing feature change across known groups was fitted with a simple linear regression over the averaged feature value per known groups in order of increasing disease burden (i.e., gene-negative control volunteers, people with pre-manifest HD, people with manifest HD).

[0196] Feature ranking process

[0197] Quality metrics

[0198] The quality metrics used for the subsequent feature ranking represented desired properties that a feature should have for measuring disease progression in HD. The three quality components used were mean sensitivity to change (mSTC), mean intraclass correlation coefficient (mICC) and absolute mean cross-sectional correlation coefficient (mCSC).

[0199] Mean intraclass correlation coefficient For each digital feature, intraclass correlation coefficients (2k) (ICC[2k]s) were computed between each pair of consecutive 2-week periods

[0031] , The mICC was defined as the mean of each ICC(2k). Importantly, although mSTC also includes information on reliability, mSTC and mICC are not redundant, especially at high mICCs where a ceiling effect occurs (Fig. 10).

[0200] Mean sensitivity to change

[0201] Sensitivity to change was obtained from mixed models with repeated measures. The dependent variable was the change from baseline of the digital features. Independent variables included study week (as factor), age, CAG repeats, and feature value at baseline. Separate mixed models were fitted for each digital feature. To facilitate convergence, data were averaged over 4 weeks per study participant, and a Toeplitz error structure was used to model residuals variance / covariance (mmrm R package with “BFGS” optimizer)

[0032] , Additionally, baseline data (Week 0) was removed from the dataset to assist with the model’s convergence. From the model, the estimated marginal mean of the change from baseline and their standard error per week were computed (emmeans R package)

[0033] , Using these models’ outputs, the sensitivity to change was then calculated for each week as: where STC is the sensitivity to change. To obtain the mSTC, the individual STCs were averaged across all study weeks by taking the mean.

[0202] Mean cross-sectional correlation coefficient

[0203] To compute the mCSC, the Spearman rank correlation coefficient was computed between the digital features and the composite Unified Huntington’s Disease Rating Scale (cUHDRS) for each clinical visit during which the cUHDRS was administered (for the digital features, the median feature value of a 2-week period centered at the clinical visit was used). Using the correlation with cUHDRS as a quality metric ensured that all selected features capture disease-relevant signals. Since the different studies scheduled the clinical visits at slightly different time points, clinical visit time points from the different studies were combined together to make the datasets more comparable. For example, clinical visits at ‘Day 85’, ‘Month 3’, ‘Week 17’ and ‘Week 21’ were grouped together. The mCSC corresponded to the absolute value of the mean correlation coefficients across all clinical visits.

[0204] Feature ranking methods To ensure robustness of the ranking, two separate ranking methods were used to rank the digital features. The first ranking method, the product method, consisted of calculating the product of the three quality metric components (i.e., mST C * mICC * mCSC). Features were then ranked from highest to lowest product. The rationale behind the use of a product was that features with a low-quality metric component would lower the overall ranking more than when using, for example, a sum, thus favouring features that perform well in all three components. The second ranking method, the stepwise method, consisted of three steps: 1) selecting features with mICC >0.8 (any feature with an mICC >0.8 was considered reliable enough and, therefore, the mICC above this threshold should not be a selecting criteria anymore), 2) ranking features by mSTC and 3) controlling for mCSC (i.e., selecting the top feature amongst the best ranked features according to their mCSC). The stepwise method ensured that the sensitivity to change (i.e., mSTC) remained the most important parameter to consider, while filtering out features with low test-retest reliability (i.e., mICC) and ensuring that the best features captured similar disease-related signals as the cUHDRS (i.e., mCSC). The final feature selection for each test was based on a compromise between the two ranking methods and the face validity of the feature (i.e., features that are simpler to understand and / or describe were favoured).

[0205] Normalising feature values

[0206] To prevent ‘data leaks’ for the validation process and allow the aggregation of the development dataset and GENERATION HD 1 into the full dataset for the final analysis, normalisation was performed using the mean and standard deviation of the development dataset. Visual inspection of the normalised feature values of those features selected by the feature ranking process revealed that they were linearly related to each other, their distributions were roughly multivariate normal, and the distribution and correlations between features were stable over time (Fig. 11). Thus, all selected features were normalised to have a mean of 0 and a standard deviation of 1. Furthermore, the U-Turn Test feature selected by the feature ranking process was multiplied with -1 so all features shared the same directionality, with higher values reflecting worse disease status.

[0207] Factor analysis

[0208] Given that all motor tests were designed to detect motor impairment related to disease severity, it was assumed that the common variance across all features was related to disease severity. Thus, factor analysis allowed us to quantify the ability of the selected features to represent disease severity and consequently modify the weights attributed to each feature according to this representation without relying on a clinical anchor. Being able to independently quantify disease severity is particularly important when having otherwise only imperfect (i.e., noisy) clinical anchors to validate a new score. Furthermore, factor analysis allows the assessment of measurement invariance. It ensures that the construct represented by the score remains comparable throughout time and datasets, thereby ensuring that the progression measured by the score remains interpretable and generalizable to people with Huntington’s Disease Integrating Staging System (HD-ISS) Stage 2-3 disease.

[0034] This is a crucial aspect when developing a new score that represents an abstract construct that is not physically measurable (such as disease severity, disease progression or clinical decline).

[0209] Digital features selected by the feature ranking process were first tested for suitability for the exploratory factor analysis with the Kaiser-Meyer-Olkin and Bartlett’s tests applied to each 2-week period. Additionally, comparison data and scree analysis methods were used on each 2-week period to define the optimal number of factors.

[0035] After verifying the features’ suitability for factor analysis, an exploratory factor analysis was conducted to help define several hypotheses to model, including which motor tests to include in the HDDMS, how many latent variables should be considered, and which features these latent variables explain. This exploratory factor analysis was conducted on the development dataset.

[0210] On the basis of the findings of this analysis, four hypothesised models, which differed from each other by the number of latent variables and the digital features that they included, were then formally tested on the development dataset with confirmatory factor analysis and measurement invariance analysis using maximum likelihood with robust standard deviation from the lavaan package in R (full information maximum likelihood estimation was used for handling missing values).

[0036] A confirmatory factor analysis was separately conducted for each 2-week period. Using these confirmatory factor analysis models, measurement invariance was assessed over the entire period (i.e., from baseline until end of study) by increasing the constraints on the model parameters to compute the invariance level.

[0211] The structure of the model with the best fit on the development dataset that additionally showed strong (or better) measurement invariance and at least adequate fit indices (i.e., scaled p-value >0.05, scaled root mean square error of approximation <0.08, standardised root mean square residual <0.08 and scaled comparative fit index >0.90)

[0037] was then validated on the validation dataset. The validation consisted of verifying that the same exact model used on the development dataset was displaying the same level of measurement invariance on the validation dataset. Using the same exact models implies using the same parameter constraints and, importantly, the same parameter values for the fixed parameters. Finally, the weights of the final model were refined using data from the full dataset (i.e., development dataset and GENERATION HD1 pooled together), which was followed by applying Thomson regression to compute the final HDDMS

[0038] ,

[0212] To develop and subsequently validate the HDDMS, the digital data were split into separate development and validation datasets (Fig. 2). The development dataset, which consisted of data collected in the HD Natural History Study and the OLE of the tominersen Phase I / IIa study, was used to initially develop the feature ranking process and factor analysis steps. Both steps were subsequently validated on the validation dataset, which consisted of a representative slice that was randomly sampled from GENERATION HD1. Using a representative slice would have permitted us to perform a second validation attempt should it have been necessary; however, this turned out not to be the case. Additionally, data from all four studies were used to filter out non-congruent features. Data from the development dataset and GENERATION HD1 were used to assess the longitudinal change, whereas data from Digital-HD, which enrolled gene-negative control volunteers and people with premanifest HD, were used to assess the known-groups validity ( see above Datasets). The final model parameters used to compute HDDMS were derived using data from the full dataset consisting of data from the HD Natural History Study, the OLE of the tominersen Phase Ella study, and GENERATION HD1. This allowed the HDDMS to be as representative as possible of a HD population typical for clinical studies.

[0213] Statistical analyses

[0214] Characterising the Huntington ’s Disease Digital Motor Score

[0215] Descriptive statistics described the distribution of the HDDMS and the change in HDDMS from baseline in selected studies. Test-retest reliability was assessed for all subsequent 2- week periods in all three studies included in the full dataset. Differences between gene-negative control volunteers, people with premanifest HD, and people with manifest HD were assessed in Digital-HD.

[0216] The following two analyses evaluated the HDDMS on data obtained in GENERATION HD 1 as the schedule of assessments, including clinical visits, differed in the other studies. The HDDMS and the digital features it is composed of were correlated against commonly used HD clinical anchors such as the cUHDRS; selected UHDRS items; EuroQol 5-dimension, 5-level (EQ-5D-5L); Symbol Digit Modalities Test (SDMT); and Stroop Wording Reading Test (SWRT) using Pearson’s correlation. These correlations were performed cross-section- ally for each clinical visit (Weeks 0, 4, 20, 36, 52, and 68). Sensitivity to change was assessed for: the HDDMS, the digital features the HDDMS is composed of, cUHDRS, selected UHDRS items, SDMT, and SWRT. A mixed model for repeated measures with an unstructured error structure was separately fitted for people with HD-ISS Stage 2 or Stage 3 disease to measure the change from baseline. The dependent variable was the change from baseline in either the features, HDDMS, or clinical anchor. Independent variables included weeks (as factor), age, CAG repeats, and the feature value at baseline. Using the estimate obtained from this model, sensitivity to change was then cal- culated as - . To ensure comparability with the cUHDRS, this analy- standard deviation sis included only people with HD who had both digital and cUHDRS data available.

[0217] Sample size calculation

[0218] To demonstrate the value of the HDDMS for upcoming clinical trials enrolling people with HD-ISS Stage 2-3 disease, post-hoc power analyses were performed on GENERATION HD1 data using the longpower R package19to estimate the sample size required for detecting a 40% effect size in the change from baseline on either the HDDMS or cUHDRS at Weeks 36, 52, and 68 with a power of 0.8 and an alpha ofeither 0.2, 0.1 or 0.05. Effect size estimates were calculated using the mixed model’s estimates of change from baseline and pooled standard deviation. For both HDDMS and cUHDRS, the change from baseline estimates were calculated using data from the placebo arm of GENERATION HD1

[0013] , Data from the placebo and one of the treatment arms (16-week dosing arm) of GENERATION HD1 were used to obtain the pooled standard deviation.

[0219] Yearly change on the Huntington’s Disease Digital Motor Score

[0220] To estimate the yearly change in people with HD-ISS Stage 2-3 disease, a linear mixed model was fit on the data from the placebo arm of GENERATION HD 1 using the R package lme4. Study week was divided by 52. Hence Week 52 corresponded to the value 1. The dependent variable was the change from baseline in the HDDMS. Independent variables included study week, age, CAG repeats, and the HDDMS value at baseline as well as an interaction between study week and HD-ISS group. The model included random intercepts and random slope per people across weeks. The R package ggeffects was used to compute the confidence interval (CI) of the slopes.

[0221] Example 2 - Results

[0222] Baseline demographic and disease characteristics for the development dataset (n=141), Digital-HD (n=120) and GENERATION HD1 (n=787) are presented in Table 1. Table 1. Baseline demographics and disease characteristics.

[0223] Development dataset Digital-HD GENERATION HDlaAll

[0224] - people HD- HD- Over- Gene- Pre- Mani- HD-ISS HD-ISS Over- .h

[0225] ISS ISS all negative mani- fest HD Stage Stage allHD

[0226] Stage Stage control fest HD 2b3b

[0227] 2b3bvolunteers n 23 118 141 40 40 40 118 668 787 1008

[0228] Fe- 8 46 54 18 (45.0) 22 19 42 313 355 450 male, n (34.8) (39.0) (38.3) (55.0) (47.5) (35.6) (46.9) (45.1) (44.6)

[0229] (%)

[0230] Age, 48.9 48.1 48.2 43.9 42.6 55.7 46.0 48.4 48.1 48.2 years, (10.0) (10.0) (9.9) (14.1) (9.1) (11.0) (10.2) (9.4) (9.6) (9.9) mean (SD)

[0231] CAG re- 43.3 44.4 44.2 N / A 41.9 42.6 45.0 44.9 44.9 44.6 peats, (2.4) (3.2) (3.1) (1.8) (3.3) (3.5) (3.2) (3.2) (3.3) mean (SD)

[0232] CAP 451.3 489.6 483.3 N / A 343.8 475.2 490.2 518.9 514.5 501.9 score, (56.4) (68.6) (68.0) (76.4) (83.2) (58.2) (67.0) (66.5) (76.0) mean (SD) cUH- 14.9 12.4 12.8 17.1 17.1 10.0 14.2 10.9 11.4 11.8

[0233] DRS, (1.4) (2.5) (2.5) (2.0) (2.1) (3.8) (1.8) (2.4) (2.6) (2.9) mean (SD)

[0234] SDMT 38.3 31.5 32.6 50.2 53.7 27.2 34.5 25.3±1 26.7 28.6 total (9.0) (11.9) (11.7) (16.6) (15.2) (12.4) (10.7) 0.3 (10.9) (12.5) score, mean (SD)

[0235] SWRT 83.3 70.5 72.6 103.4 102.6 68.4 78.5 63.8 66.0 68.5 score, (16.4) (20.8) (20.7) (16.1) (18.5) (20.3) (18.3) (18.0) (18.8) (20.4) mean (SD)

[0236] TFC 13.0 10.7 11.1 13 (0) 13 (0.2) 10.7 13.0 10.0 10.4 10.6 score, (0.0) (1.4) (1.5) (2.2) (0.1) (2.0) (2.2) (2.1) mean (SD) TMS 13.9 24.2 22.6 0.9 (1.9) 4.4 (3.3) 32.4 18.1 30.2 28.3 26.8 score, (5.8) (11.6) (11.4) (17.0) (7.8) (12.8) (12.9) (13.6) mean (SD)

[0237] TMS-4 10.9 17.2 16.2 0.6 (1.3) 3.6 (2.4) 22.1 13.3 21.8 20.6 19.3 score, (4.4) (8.8) (8.6) (11.1) (5.7) (9.5) (9.5) (9.9) mean (SD)CaH D-ISS Stage classification was not available for one participant.bHD-ISS classification was done retrospectively; key inclusion and exclusion criteria are summarised in Supplementary Table SI.

[0238] 'The TMS-4 is a condensed version of the TMS and includes 23 of the 31 TMS items (Siesling et al. Mov Disord. 1997; 12(2): 229-34).

[0239] CAG, cytosine-adenine-guanine trinucleotide; CAP, CAG-Age-Product; clIHDRS, composite Unified Huntington's Disease Rating Scale; HD, Huntington's disease; HD-ISS, Huntington's Disease Integrated Staging Scale; N / A, not applicable; SD, standard deviation; SDMT, Symbol Digit Modalities Test; SWRT, Single Word Reading Test; TFS, Total Functional Capacity; TMS, Total Motor Score (31-items); TMS-4, short version of the Total Motor Score.

[0240] For the validation dataset, a representative slice of 319 people with manifest HD was randomly sampled from GENERATION HD 1. The HDDMS model’s weights were computed on the full dataset, consisting of 928 people with manifest HD enrolled in studies included in the development dataset or in GENERATION HD1. Adherence to the digital motor tests remained at approximately 70% (Fig. 7).

[0241] Huntington’s Disease Digital Motor Score

[0242] The HDDMS model that best describes motor progression of HD uses one latent variable to explain the common variance among the four digital features described in Table 2 (Balance Test: standard deviation of the accelerometer magnitude; Chorea Test: absolute value of the standard deviation of the z-acceleration divided by the accelerometer magnitude; Speeded Tapping Test: standard deviation of the time between the onset of one tap and the onset of the subsequent tap; Two-Minute Walk Test: coefficient of variation of step impulse; see also appendix p. 10). These features are characterised by high test-retest reliability (Table 6), good convergent validity (Fig. 3B), and good sensitivity to change (Fig. 4C and 4D). Table 2. Description of features included in the HDDMS.

[0243] Digital motor test Digital feature Motor concept assessed by the

[0244] _ digital feature3_

[0245] BAL Standard deviation of the accel- Chorea of the trunk erometer magnitude

[0246] CHO Absolute value of the standard deChorea of the upper limbs viation of the z-acceleration divided by the accelerometer magnitude

[0247] Standard deviation of the time beFine motor control of the fingers tween the onset of one tap and the onset of the subsequent tap

[0248] TMW Variability of step impulse Motor control of the lower limbsaWhereas the digital features were developed with these motor concepts in mind, it is possible that they capture additional motor concepts.

[0249] BAL, Balance Test; CHO, Chorea Test; HDDMS, Huntington's Disease Digital Motor Score; STT, Speeded Tapping Test; TMW, Two-Minute Walk Test.

[0250] The HDDMS is not bounded by design, but physiological or measurement limits may occur at the extreme values of performance. Furthermore, as a score intended to measure disease progression, the scale anchor is of limited relevance. In the current work, the HDDMS value of 0 was assigned to the best-performing gene-negative control volunteer in Digital- HD. A linear mixed model showed that the HDDMS increases by 0.54 (95% confidence interval [CI]: 0.37-0.71) points per year in HD-ISS Stage 2 and by 0.43 (95% CI: 0.36- 0.50) points per year in HD-ISS Stage 3 (appendix p. 8), though the difference is not statistically significant (estimate of the difference between slopes = -0.11 [95% CI: -0.29 to 0.07, p-value = 0.24]).

[0251] Table 6. Top five digital features selected by the feature ranking process for each motor test using data from the development dataset.3

[0252] Quality metric components Ranking Ranking Digital mICC mSTC mCSC Quality using using feature11metric product stepwise method method

[0253] 2 1 BAL_1 0.96 0.45 0.38 0.16

[0254] 4 3 BAL_3 0.96 0.35 0.36 0.12

[0255] 9 4 BAL_4 0.88 0.31 0.30 0.08

[0256] 5 5 BAL_5 0.84 0.30 0.43 0.11

[0257] 1 1 CHO_1 0.95 0.63 0.42 0.25

[0258] 3 2 CHO_2 0.97 0.62 0.36 0.22

[0259] 4 6 CHO_3 0.93 0.53 0.40 0.20

[0260] 7 4 CHO_4 0.96 0.58 0.33 0.19

[0261] 1 1 STT_1 0.99 0.79 0.61 0.47

[0262] 3 3 STT_2 0.99 0.78 0.60 0.47

[0263] 4 4 STT_3 0.99 0.71 0.62 0.44

[0264] 6 6 STT_4 0.99 0.71 0.62 0.44

[0265] 6 1 TMW_1 0.96 0.61 0.34 0.20

[0266] 10 3 TMW_3 0.94 0.60 0.33 0.19

[0267] 2 4 TMW_4 0.95 0.60 0.40 0.23

[0268] 8 5 TMW_5 0.94 0.60 0.34 0.19

[0269] 10 1 UTU 1 0.93 0.62 0.33 0.19 3 2 UTU_2 0.91 0.61 0.37 0.21

[0270] 9 3 UTU_3 0.84 0.59 0.38 0.19

[0271] 13 4 UTU_4 0.92 0.59 0.34 0.18 aQuality check features are not included.bEntries shaded in grey are the features selected for the subsequent factor analysis.cMovement done on the screen with one or more fingers.dMovement done on the screen with only one finger.

[0272] BAL, Balance Test; CHO, Chorea Test; mCSC, mean cross-sectional correlation; mICC, mean intraclass correlation coefficient; mSTC, mean sensitivity to change; STT, Speeded Tapping Test; TMW, Two-Minute Walk; UTU, U-Turn Test.

[0273] This HDDMS model showed strong measurement invariance across time and datasets (appendix p. 11), suggesting that the HDDMS is generalizable across studies. Importantly, in selected studies the HDDMS showed a strong capacity to detect change over time (Fig.

[0274] 5 A), demonstrated high test-retest reliability (intraclass correlation coefficients >0.95) (Fig. 5B), and had a close-to-normal distribution that remained stable throughout the studies (Fig. 5C and Fig. 8). Higher HDDMS values reflect more severe disease, which is congruent to the differences observed across the different cohorts enrolled in Digital-HD (Fig. 5D).

[0275] In GENERATION HD1, cross-sectional Pearson’s correlations between HDDMS and HD clinical anchors at baseline (cUHDRS: -0.50; EQ-5D-5L: -0.11; SDMT: -0.44; SWRT: -0.41) were nearly as high as between the best digital feature (the feature derived from Speeded Tapping Test) and the same anchors (cUHDRS: -0.55; EQ-5D-5L: -0.15;

[0276] SDMT: -0.49; SWRT: -0.45) (Fig. 3). Thus, the associations observed with the individual features were also observed with the HDDMS.

[0277] Notably, the HDDMS demonstrated superior sensitivity to change relative HD clinical anchors that have been accepted as clinically meaningful trial endpoints (Fig. 4). In GENERATION HD1, HDDMS’s sensitivity to detect changes in HD-ISS Stage 2-3 disease at Week 20 was comparable to that of cUHDRS at Week 68. This suggests that by using the HDDMS, study duration might be reduced by up to 60%. The HDDMS’s sensitivity to change was also superior to that of its constituent digital features (Fig. 4), highlighting the benefit of combining multiple features in a single weighted composite score. Sample size calculation

[0278] Simulated sample size calculations demonstrated that using the HDDMS instead of the cUHDRS can substantially reduce the sample size (by 75-80%) required to detect a 40% effect size difference in people with HD-ISS Stage 2-3 while still maintaining a statistical power of 0.8 of the planned study (Fig. 6), which would also lead to reduced study duration due to faster recruitment. The estimated number of participants and study duration of upcoming trials using the HDDMS are reported in Table 3.

[0279] Table 3. Estimated number of participants required per study arm to detect a 40% effect size difference with a power of 0.8 in upcoming HD-ISS Stage 2-3 clinical trials.

[0280] Alpha Study duration Estimated number of participants per arm

[0281] (month) HD-ISS Stage 2 HD-ISS Stage 3 cUHDRS HDDMS cUHDRS HDDMS

[0282] 0.05 9 15223 511 1942 195

[0283] 0.1 9 11991 403 1530 154

[0284] 0.2 9 8743 294 1115 112

[0285] 0.05 12 2539 302 955 163

[0286] 0.1 12 2000 238 752 128

[0287] 0.2 12 1458 173 548 94

[0288] 0.05 16 593 105 526 121

[0289] 0.1 16 467 83 414 96

[0290] 0.2 16 341 60 302 70 cUHDRS, composite Unified Huntington's Disease Rating Scale; HDDMS, Huntington's Disease Digital Motor Score; HD-ISS, Huntington's Disease Integrated Staging Scale.

[0291] Feature ranking process Fig. 12 presents the quality metrics for each feature derived from the motor tests included in the remote digital motoring platform. Ideal features are located in the top right corner (i.e., showing high mCSC and high mSTC) and are additionally coloured in yellow (i.e., showing high mICC). Each motor test had at least one feature, if not a cluster of features, that showed the desired measurement properties. The ranking of the features with either the product or stepwise method was robust across the development and validation datasets, indicating possible generalisability of the best features derived from each motor test (Fig. 13). In other words, a feature able to adequately describe HD disease status in the present dataset should also be able to adequately describe HD disease status in future clinical trials. This observation is valid for all motor tests. The best five features from each of the remaining five motor tests and their ranking by either ranking method are described inTable 6.

[0292] The best feature from each motor test (highlighted in bold in Table 6) was selected for inclusion in the factor analysis. This selection was based according to the features’ ranking by both methods (product and stepwise method) and their face validity. The selected features displayed congruent progression directionality with the known-group difference (mixed model with repeated measures estimates of change from baseline in Fig. 14A show the same directionality as the known-group box plots in Fig. 14B), moderate cross-sectional correlation with cUHDRS (Fig. 14C) and other clinical anchors (Fig. 3), high test- retest reliability (Fig. 14D), and good sensitivity to change relative to clinical anchors (Fig. 4).

[0293] Feature ranking process is robust to decrease in adherence

[0294] In our analyses, adherence did worsen over time (Fig. 7). Such a decrease in adherence can increase the variability of feature values, which in turn can affect the data collected in different studies. In theory, this may impact the results of the feature ranking process. If this decrease in adherence did impact the feature ranking process results, the ranking of the quality metric components would differ between baseline (Weeks 4 through 8) and the end of data collection (Weeks 64 through 68). However, no such difference was observed (Fig. 15). The quality metrics component ranking remained similar between baseline and end of data collection despite a drop out of around 30% in adherence. This suggests that the feature ranking process is not impacted by the differences in adherence observed over time.

[0295] Huntington’s Disease Digital Motor Score

[0296] Suitability of digital features for factor analysis

[0297] The standardised feature values had stable between-feature linear relationships (Fig. 11 A- 1 IB), stable close-to-normal distribution (Fig. 11C-1 ID) and stable and moderate be- tween-features correlations across study weeks (Fig. 16). Bartlett’s tests were statistically significant (p<0.05) across datasets (development and validation datasets as well as the entire GENERATION HD1 dataset) and across all 2-week periods. Furthermore, the averaged Kaiser-Meyer-Olkin test result across weeks was 0.826 for the development dataset and 0.836 for GENERATION HD1. These observations indicated that the datasets were suitable for factor analysis, especially when using a robust estimator.

[0298] Exploratory factor analysis

[0299] Scree analysis indicated that one latent variable best explained the common variance among features in the development dataset, whereas the comparison data method indicated that one or two latent variables were optimal. Using the development dataset, subsequent exploratory factor analysis fitted on the five digital features (the top five features in Table 6) showed that the model fit was generally better when using two instead of one latent variable. However, the fit indices, with one or two latent variables, were not fully adequate across all 2-week periods. Nonetheless, the loadings suggested that digital features would be better grouped as follows: best feature from the Balance Test and the best feature from Chorea Test grouped together; and separately the best Two-Minute Walk Test feature, the best Speeded Tapping Test feature, and the best U-Turn Test feature grouped together.

[0300] Confirmatory factor analysis

[0301] Based on the exploratory factor analysis findings, four measurement models were developed and tested with confirmatory factor analysis on the development dataset (Fig. 17). Models 1 and 2 included the best features from all five motor tests that were considered (Balance Test, Chorea Test, Speeded Tapping Test, Two-Minute Walk Test, and U-Turn Test) and tested the use of one versus two latent variables. Models 3 and 4 tested one latent variable but differed in the digital features that were included (Model 3: Balance Test, Chorea Test, Speeded Tapping Test, and Two-Minute Walk Test; Model 4: Balance Test, Chorea Test, Speeded Tapping Test, and U-Turn Test). Of these four models, Model 3 showed the best-fit indices along with strong measurement invariance (Table 7).

[0302] Model 3 was subsequently successfully validated on the validation dataset (comparative fit index scaled = 0.972; root mean squared error of approximation scaled = 0.058 [95% CI: 0.048-0.063]; standardised root mean squared error = 0.053). The p value for the validation was <0.0001 likely because of the larger sample size of the validation dataset compared with the development dataset, slight deviations from a multivariate normal distribution, or the different studies used in the two datasets.

[0039] Following this validation, the model’s weights were refined using data from the full dataset (Fig. 18). The strong measurement invariance remained high after this refinement (scaled comparative fit index = 0.976, scaled root mean squared error or approximation = 0.052 [95% CI: 0.048-0.056], standardised root mean squared residual = 0.031, mean composite reliability omega over time = 0.824 [range: 0.784-0.849]). Table 7. Fit indexed of measurement hypothesis /

[0303] Tit indexes were always given for the strong model configuration since it is the model used for validation, even when a strict model was fitting data adequately. Red and blue font indicated a fit index below or above adequacy, respectively. The green background indicates that model 3 was selected. Data are presented for the development dataset.bMeasurement invariance of the best fitting model is reported.

[0304] AIC, Akaike information criterion; BIC, Bayesian information criterion; CFI, comparative fit index; RMSEA, root mean squared error or approximation; SRMR, standardised root mean squared residual.

[0305] Discussion

[0306] The HDDMS is a novel data-driven composite score for sensitive measurement of disease progression in HD. Its sensitivity to change is superior to that of HD clinical anchors that have been accepted as clinically meaningful trial endpoints. The superior sensitivity to change was even observed in people with HD-ISS Stage 2 disease, who typically show limited sensitivity to endpoints such as the cUHDRS [6-8], By using the HDDMS, changes in motor progression can be identified with a significantly reduced sample size in HD-ISS Stage 2-3 studies. This is particularly useful for a rare neurodegenerative disease such as HD, where clinical decline occurs slowly, clinical manifestations can be highly variable, and recruiting sufficiently large study cohorts for early-stage trials may be challenging. The improvement in sensitivity to change on the HDDMS is likely due to three factors: (1) Collecting data remotely enables more frequent measurements, which can be averaged to reduce noise

[0013] ,

[0307] (2) It also reduces the additional variability that may be introduced through the effect of clinic visits; this is exemplified by the short-term improvement on clinical measures at the first post-baseline visit in GENERATION HD1 that is absent in the digital data (the absence of such improvement in people in HD-ISS Stage 2 in Fig. 4B is likely due to ceiling effects).

[0308] (3) Measuring change on a continuous rather than an ordinal scale allows smaller signal fluctuations to be detected. It is worth noting that the sensitivity to change of the reference clinical anchors in this work was comparable to previously reported findings for the same anchors.20As sensitivity to change of clinical measures in HD is not very high to begin with, any reduction in noise will lead to substantial improvement.

[0309] Our digital motor tests were co-designed with people with HD and neurologists, based on assessments commonly used in clinical trials and practice, and are intended to measure both voluntary motor impairment and involuntary movements (e.g., chorea) in people with HD [13,21], Such motor changes occur early on and progress throughout the disease course of adult-onset HD

[0022] , While chorea tends to manifest first and progresses until it plateaus, or even decreases, at later stages of the disease, voluntary motor impairment progresses more steadily [23,24], Thus, assessing both voluntary motor impairment and chorea with the HDDMS makes it suitable for evaluating progression of HD.

[0310] The selection of features that best reflect performance was done via a data-driven approach. This guaranteed that features would have the measurement properties required for score construction

[0022] and avoided human bias (our dataset of 1,008 people with manifest HD was sufficiently large to ensured that any data-related bias would be limited). Indeed, our feature ranking process consistently selected the same digital features across different datasets and ranking methods and, additionally, was not impacted by the drop in adherence that can be expected with any digital health technology tool used for exploratory purposes (see Exploratory and confirmatory factor analysis above). Interestingly, several of the data- driven digital features included in the HDDMS measure variability of motor function. Previous studies reported increased variability of motor function in the upper and lower extremities in people with HD [25,26], Increased variability in tapping interval, a measure similar to our digital feature derived from the Speeded Tapping Test, correlated with atrophy in the caudate, putamen, and other brain structures associated with HD

[0025] , Together with our findings, this suggests that increased variability of motor function is a characteristic of HD that occurs early on and may, at least partially, explain the superior sensitivity to change of the HDDMS compared with accepted HD clinical trial endpoints. A composite score has the benefit of facilitating the use of multi-dimensional data and may have increased sensitivity to change due to averaging over multiple different measurements. It may be difficult to use, however, if it spans different concepts, and the weights of the individual features may not generalize well. Using factor analysis for constructing the score allowed addressing both challenges. Factor analysis indicated that that motor progression of HD can be appropriately captured with a scalar composite score. Furthermore, the HDDMS model showed strong measurement invariance over time and datasets, indicating that the relationship between the features remains valid across clinical studies. Importantly, the development dataset was not only drawn from entirely separate studies, it also had narrower patient characteristics than the validation dataset, further increasing the confidence that the performance of the HDDMS can be replicated in future HD clinical trials

[0027] . However, additional independent studies are required to evaluate the effectiveness of the HDDMS in different HD populations.

[0311] In conclusion, the HDDMS is a novel digital tool in HD clinical research for measuring motor severity and progression of HD in clinical research, designed for people with HD- ISS Stage 2-3 disease. It complements existing clinical assessments and offers improved sensitivity to change, thereby enabling smaller or shorter clinical trials for early decisionmaking on clinical efficacy of novel HD therapies.

[0312] References

[0313] [1], Bates GP, Dorsey R, Gusella JF, et al. Huntington disease. Nat Rev Dis Primers 2015; 1: 15005.

[0314] [2], Roos RA. Huntington's disease: a clinical review. Orphanet J Rare Dis 2010; 5: 40.

[0315] [3], Saudou F, Humbert S. The Biology of huntingtin. Neuron 2016; 89: 910-26.

[0316] [4], Keum Jae W, Shin A, Gillis T, et al. The HTT CAG-expansion mutation determines age at death but not Disease duration in Huntington disease. Am J Hum Genet 2016; 98: 287-98.

[0317] [5], Komatsu H. Innovative therapeutic approaches for Huntington’s disease: from nucleic acids to GPCR-targeting small molecules. Front Cell Neurosci 2021; 15: 785703.

[0318] [6], Tabrizi SJ, Schobel S, Gantman EC, et al. A biological classification of Huntington's disease: the Integrated Staging System. Lancet Neuro. 2022; 21: 632-44.

[0319] [7], Shoulson I, Fahn S. Huntington disease: clinical care and evaluation. Neurology 1979; 29: 1-3.

[0320] [8], Paulsen JS, Wang C, Duff K, et al. Challenges assessing clinical endpoints in early Huntington disease. Mov Disord 2010; 25: 2595-603.

[0321] [9], Reilmann R, Rouzade-Dominguez M-L, Saft C, et al. A randomized, placebo-controlled trial of AFQ056 for the treatment of chorea in Huntington's disease. Mov Disord 2015; 30: 427-31.

[0322]

[0010] , Waddell EM, Dinesh K, Spear KL, et al. GEORGE®: A pilot study of a smartphone application for Huntington’s disease. J Huntington's Dis 2021; 10: 293-301.

[0323]

[0011] , Dorsey ER, Papapetropoulos S, Xiong M, Kieburtz K. The first frontier: digital biomarkers for neurodegenerative disorders. Digit Biomark 2017 ; 1: 6-13.

[0324]

[0012] , Adams Jamie L, Dinesh K, Xiong M, et al. Multiple wearable sensors in Parkinson and Huntington disease individuals: a pilot study in clinic and at home. Digit Biomark 2017; 1: 52-63.

[0325]

[0013] .Lipsmeier F, Simillion C, Bamdadian A, et al. A remote digital monitoring platform to assess cognitive and motor symptoms in Huntington disease: cross-sectional validation study. J Med Internet Res 2022; 24: e32997.

[0326]

[0014] .McColgan P, Thobhani A, Boak L, et al. Tominersen in adults with manifest Huntington's disease. N Eng J Med 2023; 389: 2203-5.

[0015] , Huntington Study Group. Unified Huntington's Disease Rating Scale: reliability and consistency. Huntington Study Group. Mov Disord 1996; 11: 136-42.

[0327]

[0016] .Bourke AK, Scotland A, Lipsmeier F, Gossens C, Lindemann M. Gait characteristics harvested during a amartphone-based self-administered 2-Minute Walk Test in people with multiple sclerosis: Test-Retest Reliability and Minimum Detectable Change. Sensors 2020; 20: 5906.

[0328]

[0017] , Cheng W-Y, Bourke AK, Lipsmeier F, et al. U-turn speed is a valid and reliable smartphone-based measure of multiple sclerosis-related gait and balance impairment. Gait Posture 2021; 84: 120-6.

[0329]

[0018] , Graves JS, Ganzetti M, Dondelinger F, et al. Preliminary validity of the Draw a Shape Test for upper extremity assessment in multiple sclerosis. Ann Clin Transl Neurol 2023;

[0330] 10: 166-80.

[0331]

[0019] .Iddi S, Donohue MC. Power and sample size for longitudinal models in R - the long- power package and shiny app. R J 2022; 14: 264-82.

[0332]

[0020] , Schobel SA, Palermo G, Auinger P, et al. Motor, cognitive, and functional declines contribute to a single progressive factor in early HD. Neurology 2017; 89: 2495-502.

[0333]

[0021] , Berchtold D, Rennig J, Simillion C, et al. Evaluating tominersen effect on self-re- ported quality of life (QOL) with high-frequency longitudinal, digital Euroqol 5-Dimen- sions 5-Level (EQ-5D-5L) survey in GENERATION HD1, a phase III trial of tominersen in adult individuals with manifest Huntington's disease (HD). J Neurol Neurosurg Psychiatry 2022; 93: A94.

[0334]

[0022] .Ross CA, Aylward EH, Wild EJ, et al. Huntington disease: natural history, biomarkers and prospects for therapeutics. Nat Rev Neurol 2014; 10: 204-16.

[0335]

[0023] , Rosenblatt A, Liang KY, Zhou H, et al. The association of CAG repeat length with clinical progression in Huntington disease. Neurology 2006; 66: 1016.

[0336]

[0024] , Kirkwood SC, Su JL, Conneally P, Foroud T. Progression of symptoms in the early and middle stages of Huntington disease. Arch Neurol 2001; 58: 273-8.

[0337]

[0025] , Bechtel N, Scahill RI, Rosas HD, et al. Tapping linked to function and structure in premanifest and symptomatic Huntington disease. Neurology 2010; 75: 2150.

[0338]

[0026] , GaBner H, Jensen D, Marxreiter F, et al. Gait variability as digital biomarker of disease severity in Huntington’s disease. J Neurol 2020; 267: 1594-601.

[0027] . Hastie T, Tibshirani R, Friedman J. Model assessment and selection. In: Hastie T, Tibshirani R, Friedman J, eds. The elements of statistical learning: data mining, inference, and prediction. New York, NY: Springer New York; 2009: 219-59.

[0339]

[0028] , Our members. Vivli. Available at: https: / / vivli.org / ourmember / roche / . (accessed 7 June 2023).

[0340]

[0029] , Our commitment to data sharing. Roche. Available at: https: / / www.roche.com / innova- tion / process / clinical-trials / data-sharing / . (accessed 7 June 2023).

[0341]

[0030] .Bates D, Maehler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw 2015; 67: 1-48.

[0342]

[0031] .McGraw KO, Wong SP. Forming inferences about some intraclass correlation coefficients. Psychol Methods 1996; 1 : 30-46.

[0343]

[0032] Sabanes Bove D, Dedic J, Kelkhoff D, et al. mmrm: mixed models for repeated measures. R package version 0.2.2. 2022. https: / / openpharma.github.io / mmrm / (accessed 18 May 2023).

[0344]

[0033] , Searle SR, Speed FM, Milliken GA. Population marginal means in the linear model: An alternative to least squares means. Am Stat 1980; 34: 216-21.

[0345]

[0034] , Kline RB. Principles and practice of structural equation modeling. 4th ed: The Guilford Press; 2015.

[0346]

[0035] , Steiner MD, Grieder S. EFAtools: An R package with fast and flexible implementations of exploratory factor analysis tools. J Open Source Softw 2020; 5): 2521.

[0347]

[0036] .Rosseel Y. lavaan: An R Package for structural equation modeling. J Stat Softw 2012; 48: 1-36.

[0348]

[0037] .Hooper D, Coughlan J, Mullen M. Structural equation modelling: guidelines for determining model fit. Electron J Bus Res Methods 2008; 6: 53-60.

[0349]

[0038] , Grice JW. Computing and evaluating factor scores. Psychol Methods 2001; 6: 430- 50.

[0350]

[0039] .Bender PM, Dudgeon P. Covariance structure analysis: statistical practice, theory, and directions. Annu Rev Psychol 1996; 47: 563-92.

[0351]

[0040] .Siesling S, Zwinderman AH, van Vugt JPP, Kieburtz K, Roos RAC. A shortened version of the motor section of the unified Huntington's disease rating scale. Mov Disord 1997; 12: 229-34.

Claims

Claims1. A computer-implemented method for assessing Huntington's disease (HD), typically progression in HD, in a subject comprising the steps of: a) determining a Huntington’s disease digital motor score (HDDMS) based on a multitude of digital performance features derived from at least one or more of the following:- accelerometer measurements,- touch sensor measurements, and / or- measurements of time in a dataset of fine motoric and motoric activity measurements from said subject; b) comparing the determined HDDMS to a reference; and c) assessing HD, typically progression in HD, in the subject based on said comparison; wherein the HDDMS comprises at least three digital performance features derived from a dataset of fine motoric and motoric activity measurements from said subject.

2. The method of claim 1, wherein the HDDMS comprises at least four, more typically four, digital performance features derived from a dataset of fine motoric and motoric activity measurements from said subject.

3. The method of any one of the preceding claims, wherein said fine motoric and motoric activity measurements are carried out using a mobile device.

4. The method of any one of the preceding claims, wherein said dataset of fine motoric and motoric activity measurements comprises measurements of fine motor control of the fingers and measurements of motor control of lower limbs, trunk and upper limbs; preferably measurements of fine motor control of the fingers, measurements of motor control of lower limbs and measurements of chorea of trunk and upper limbs.

5. The method of any one of the preceding claims, wherein said dataset of fine motoric and motoric activity measurements has been collected from the subject while carrying out fine motoric or motoric activity tests; typically activity tests assessing balance, gait, finger movement accuracy, and / or movement speed; more typically these activity tests are one or more,up to all of the following fine motoric and motoric activity tests: Balance Test, Chorea Test, Speeded Tapping Test, Two-Minute Walk Test.

6. The method of any one of the preceding claims, wherein said digital performance features comprise quantitative values; typically wherein the quantitative values comprise, more typically consist of, statistical values derived from said accelerometer measurements, touch sensor measurements, or measurements of time.

7. The method of the preceding claim, wherein said statistical values comprise two or more of the following: standard deviation of the accelerometer magnitude; absolute value of the standard deviation of the z-acceleration divided by the accelerometer magnitude; standard deviation of the time between the onset of one tap and the onset of the subsequent tap; coefficient of variation of step impulse.

8. The method of any one of the preceding claims, wherein the subject is a subject diagnosed with HD, preferably a HD patient.

9. The method of any one of the preceding claims, wherein the reference is a HDDMS of a healthy subject or a previously determined HDDMS of the subject HD is being assessed in.

10. The method of any one of the preceding claims, wherein the dataset of fine motoric and motoric activity measurements from said subject is a preexisting dataset.

11. The method of any one of claims 3 to 10, wherein said mobile device is a smartphone, smartwatch, wearable sensor, portable multimedia device or tablet computer.

12. The method of any one of the preceding claims, wherein a change or difference in the determined HDDMS compared to the reference indicates HD, typically progression in HD, more typically wherein an increase in the determined HDDMS compared to the reference indicates progression in HD.

13. The method of any one of the preceding claims, wherein the digital performance features in the HDDMS each are weighted according to their ability to represent disease severity.

14. A device comprising a processor and a database as well as software which is tangibly embedded to said device and, when running on said device, carries out the method of any one of claims 1 to 13.

15. A system comprising a mobile device comprising at least one sensor, the system further comprising the device according to claim 14, wherein said mobile device and the device are operatively linked to each other.

16. A computer program adapted to perform the method of any one of claims 1 to 13.

17. A computer-readable storage medium comprising said computer program of claim 16.

Images