WO2021097533A1

WO2021097533A1 - Instrumented systems and methods for evaluating movement capacity of a person

Info

Publication number: WO2021097533A1
Application number: PCT/AU2020/051260
Authority: WO
Inventors: Malcolm Kenneth Horne; Hamid Khodakarami
Original assignee: Global Kinetics Pty Ltd
Priority date: 2019-11-20
Filing date: 2020-11-20
Publication date: 2021-05-27

Abstract

Systems and automated methods for evaluating movement capacity of a person ustilise time-marked motion data from a body-worn device. The time-marked motion data is processed by a processor to produce measures of kinetic state that are indicative of movements of the subject that are attributable to symptoms of the movement disorder. The processor calculates one or more evaluation scores that may be transmitted to a user interface for presentation to a user.

Description

INSTRUMENTED SYSTEMS AND METHODS FOR EVALUATING MOVEMENT CAPACITY OF A PERSON

Technical Field

[0001 ] The present invention relates to systems and methods for evaluating movement capacity of a person. It relates specifically but not exclusively to automated methods for evaluating movement capacity using time-marked motion data from a body-worn device.

Background of Invention

[0002] Parkinson’s Disease (PD) is a progressive neurodegenerative disorder that affects the frontal lobe, the brainstem and the autonomic nervous system. Impaired dopamine transmission is a defining feature of PD and can be treated by pharmaceutical therapies such as levodopa. People with PD have less dopamine, a neurotransmitter released by brain neurons in the part of the brain which helps regulate movement. People with PD experience movement related symptoms such as bradykinesia, rigidity, tremor and postural instability. Non-movement symptoms may include speech and swallowing difficulties, cognitive impairment or behavioural change, and sleep disturbance.

[0003] Dopamine is synthesised and stored in the terminals (nerve endings) of fibres that emanate from neurones (nerve cells) affected by PD. In the healthy brain, these nerve terminals release dopamine in response to nerve impulses from the neurone. The released dopamine is rapidly taken up again and stored so that in the normal brain, only a small proportion of available dopamine (~10%) is cycled.

Disease progression results in loss of neurones and their nerve terminals so that at the time of presentation, terminal numbers are already reduced and the dopamine available for release is also depleted - this leads to the bradykinesia.

[0004] Treatment with Levodopa or L-Dopa will increase the rate of synthesis of new dopamine and increase the amount stored in the remaining terminals. In these circumstances, the amount of dopamine that can be stored following a single dose will be used over a period of hours, at which time levels will return to the untreated state: this phenomenon is referred to as “wearing-off” as clinically the patient experiences “wearing-off” of the benefit of L-Dopa and re-emergence of bradykinesia.

[0005] Wearing off has been described as “generally predictable recurrence of motor and non-motor symptomatic that precedes scheduled doses of anti-PD medication, and usually improves after those doses”. These variations in bradykinesia and non-motor symptoms to the effect of medications can be referred to as fluctuations. Fluctuations cause distress and disability for people with PD and yet are eminently treatable. However, detecting fluctuations can be difficult because people with PD do not always recognise these symptoms.

[0006] Diaries of clinical state are used to overcome these difficulties. Diaries require people with PD to record each half hour (or hour) whether they are “OFF” (untreated), “ON” (responsive to therapy) and are used as a measure of the level of fluctuations in motor symptoms. However, patients frequently delay recording and do not always accurately self-assess. Partly this is because the level at which the “switch” between OFF and ON is perceived is different according to the subject. This may be due to an impaired self-awareness of motor states in PD.

[0007] It would be useful to provide an objective measure to evaluate a subject’s movement capacity for example, to determine if a subject with PD is “OFF” or “ON” regardless of the subject’s perception of their own movement capacity.

[0008] The discussion of the background to the invention included herein including reference to documents, acts, materials, devices, articles and the like is included to explain the context of the present invention. This is not to be taken as an admission or a suggestion that any of the material referred to was published, known or part of the common general knowledge in Australia or in any other country as at the priority date of any of the claims.

Summary of Invention

[0009] Viewed from one aspect, the present invention provides a machine automated method for evaluating a movement disorder disease state in subject, using passively collected motion data, the method comprising: receiving at a processor time-marked motion data from a device worn by the subject over an assessment period; the processor processing the received motion data to produce one or more measures of kinetic state of the subject that are indicative of movements of the subject that are attributable to symptoms of the movement disorder; the processor calculating one or more evaluation scores from the one or more calculated measures of kinetic state and transmitting an operational signal to a user interface on which the one or more evaluation scores are presented to a user; wherein the one or more evaluation scores represent the evaluated state of the movement disorder for the subject during the assessment period.

[0010] Preferably, the motion data is passively collected in that it comprises signals recorded from the subject while ordinarily ambulatory and in the absence of clinician-directed motor tasks performed by the subject. Thus, motion data obtained from the subject does not require the subject to receive attentional direction, unlike e.g. the UPDRS clinical tests used to evaluate movement disorder.

[0011 ] In some embodiments, the processor calculates one or more evaluation scores by applying a computational model to the received motion data. The computational model may be generated from sample motion data using machine learning.

[0012] In some embodiments, the movement disorder disease is PD and the computational model relates sample motion data obtained from a sample of people with PD undertaking a levodopa challenge test (LDCT) to one or more measures of kinetic state.

[0013] In some embodiments, the computational model is built by: a) mapping a clinical scoring range into a number of levels representing motor function severity; b) mapping the measures of kinetic state to the levels representing motor function severity; and c) determining a function that relates the measures of kinetic state to the clinical scoring range.

[0014] Mapping the measures of kinetic state may include extracting a set of candidate features from the sample motion data and applying the classifier function to the candidate features. The classifier function may be any suitable classifier function and in one embodiment, is a binary classifier. In some embodiments, the function is determined using a logistic regression model although other functions may be utilised. Features used in the computational model may be selected from a group of one more measures of kinetic state such as those listed in Table 4.

[0015] In some embodiments, the method includes the processor processing the one or more measures of kinetic state to identify automatically a plurality of distinctive occasions during the assessment period. A distinctive occasion may be selected from a group including: a) the subject receiving a therapy; b) the subject exhibiting a therapeutic motor response to a received therapy; c) after the subject has woken from night time sleep; d) after the subject has woken from night time sleep and before first therapy of the day; e) after the subject has woken from night time sleep and after first therapy of the day; f) after a period of time exceeding a duration for which therapy is known to be effective in treating disordered movement. In some embodiments a distinctive occasion occurs at least daily and in some cases, several times per day.

[0016] The evaluation score may indicate one or more of: a) whether the subject responds to therapy; b) time to maximum responsiveness to therapy; c) duration of response to received therapy; d) magnitude of response to received therapy; e) variability of response to received therapy over the assessment period; f) an extent to which the subject is treated/undertreated by received therapy; g) an extent to which the subject is untreatable by therapy; h) an extent to which the subject is a candidate for advanced therapy and i) severity of motor function symptoms that contribute to the movement disorder.

[0017] In some embodiments, the processor calculates automatically, changes in one or more of the evaluation scores since one or more prior assessment periods. An assessment period may be for a number of hours, for 24 hours, 2, 3, 4, 5, 6, 7, 10 ,14, 21 , 30 days or may correspond to a continuously assessed period. The evaluation scores may be qualitative or quantitative. In some embodiments, the processor automatically relates an evaluation score calculated by the processor to a score on a clinically accepted scale.

[0018] In some embodiments, the evaluation score is calculated at intervals during the assessment period such as during e.g. 2 minute intervals or “epochs”. The processor may calculate automatically a duration or proportion of time during the assessment period that the evaluation score exceeds a predetermined threshold. In some embodiments, the predetermined threshold refers to evaluation scores obtained from one or more healthy subjects without the movement disorder.

[0019] In some embodiments, the movement disorder disease is Parkinson’s disease (PD) and the processor calculates automatically a Unified Parkinson’s Disease Rating Score (UPDRS) equivalent score for a subject with PD in the absence of clinician-directed motor tasks required to be performed by the subject according to a clinical UPDRS protocol.

[0020] In some embodiments, the processor determines automatically if the subject is a fluctuator, wherein the evaluation score indicates magnitude of response to received therapy, and wherein the processor determines the subject to be a fluctuator when the magnitude of response meets a significance threshold. In some embodiments, the significance threshold is selected from the group including but not limited to: (a) a 30% improvement in a measure of the kinetic state; (b) an improvement of 14 points on the UPDRS scale; and (c) an improvement of at least 1 and preferably 1.15 Severity Level points.

[0021 ] In some methods, the processor determines automatically if the fluctuator is a controlled fluctuator, by comparing the measure of kinetic state or evaluation score following a received therapy with a target value, wherein the processor determines the fluctuator to be a controlled fluctuator when the measure of kinetic state is less than the target value. In some embodiments, the processor further determines automatically if a controlled fluctuator has wearing off, by calculating a reduction in response to received therapy over a time period, wherein the processor determines the controlled fluctuator to have wearing off when, over the set period, the reduction is one or both of: (a) 1 or more Severity Level points; and (b) 14 UPDRSIII points. Typically, the set period is 90 to 210 minutes, preferably 90 to 150 minutes and more preferably approximately 120 minutes following a received therapy.

[0022] In some embodiments, the machine automated method further comprises the processor determining automatically a PTB value calculated as proportion of time during the assessment period that the measures of kinetic state, or an evaluation score calculated from the measures of kinetic state, indicate that the subject exhibits movement disorder symptoms corresponding bradykinesia. In some embodiments, the processor automatically performs one or more of: (a) identifying a subject having a calculated PTB value less than 30% as having controlled motor symptoms; and (b) monitoring a change in PTB value over time for the subject and identifying a subject having PTB over 30% and increasing as requiring modified therapy.

[0023] In some embodiments, the machine automated method further comprises the processor determining automatically a PTD value calculated as proportion of time during the assessment period that the measures of kinetic state, or an evaluation score calculated from the measures of kinetic state, indicate that the subject exhibits movement disorder symptoms corresponding dyskinesia.

[0024] In some embodiments, the processor calculates an evaluation score which characterises the subject’s response to a first therapy after waking on one or more days during the assessment period, and wherein the processor automatically translates the evaluation score to a clinical score on a clinically accepted scale.

[0025] In some embodiments, the processor pre-processes the received motion data to remove automatically: a) data corresponding to periods of night time sleep by the subject; b) data corresponding to periods the device was not worn by the subject; and c) data corresponding to periods of inactivity by the subject.

[0026] In some embodiments, the processor automatically identifies in the data unreliable data segments including segments in which there is one or more of: a) motion data indicating the subject is already responsive to therapy at a distinctive occasion corresponding to the subject receiving a therapy; b) motion data indicating excess variability in responsiveness to therapy; and c) motion data indicating inactivity during periods of expected wakefulness. In such embodiments, the processor may transmit an operational signal to a report generating processor, and the report generating processor generates a report including a notification of unreliable reporting when the processor identifies in the motion data unreliable data segments.

[0027] In some embodiments, the processor transmits an operational signal to a report generating processor that generates automatically a report containing one or both of quantitative and qualitative measures of the subject’s movement disorder disease state. The report generating processor may include in the report a recommendation for modifying a prescribed therapy. [0028] In some embodiments, the device is a wrist worn device. In some embodiments, the motion data is accelerometer data from an accelerometer in the device.

[0029] In some embodiments, device is configured to receive a therapy input, indicative of the subject having received a dose of therapy. Alternatively/additionally, therapy inputs may be received by the processor from a therapy dispensing device, and are indicative of the subject having received a therapy. In other embodiments, the processor receives time-marked therapy data from the device which are indicative of therapy received by the subject during the assessment period.

[0030] Viewed from another aspect, the present invention provides a system for evaluating a movement disorder disease state in subject, using passively collected motion data, the system comprising: a processor, a user interface and a memory module containing code corresponding to instructions causing the processor to: (a) receive time-marked motion data from a device worn by the subject over an assessment period; (b) process the received motion data to produce one or more measures of kinetic state of the subject that are indicative of movements of the subject that are attributable to symptoms of the movement disorder; (c) and calculate one or more evaluation scores from the one or more calculated measures of kinetic state; and (d) transmit an operational signal to the user interface on which the one or more evaluation scores are presented to a user; wherein the one or more evaluation scores represent the evaluated state of the movement disorder for the subject during the assessment period.

[0031] In some embodiments, the processor calculates one or more evaluation scores by applying a computational model to the received motion data. The computational model may be generated from sample motion data using machine learning. In some embodiments, the movement disorder disease is PD and the computational model relates sample motion data obtained from a sample of people with PD undertaking a levodopa challenge test (LDCT) to one or more measures of kinetic state.

[0032] In some embodiments, the computational model is built by: mapping a clinical scoring range into a number of levels representing motor function severity; mapping the measures of kinetic state to levels representing motor function severity; and determining a function that relates the measures of kinetic state to the clinical scoring range. Mapping the measures of kinetic state may include extracting a set of candidate features from the sample motion data and applying the classifier function to the candidate features. The classifier function may be any suitable classifier function such as e.g. a binary classifier. The function may be determined using any suitable model such as e.g. a logistic regression model. In some embodiments, features used in the computational model are selected from a group of one more measures of kinetic state listed in Table 4.

[0033] The evaluation score may indicate one or more of: whether the subject responds to therapy; time to maximum responsiveness to therapy; duration of response to received therapy; magnitude of response to received therapy; variability of response to received therapy over the assessment period; an extent to which the subject is treated/undertreated by received therapy; an extent to which the subject is untreatable by therapy; an extent to which the subject is a candidate for advanced therapy; and severity of motor function symptoms that contribute to the movement disorder.

[0034] In some embodiments, the movement disorder disease is Parkinson’s disease (PD) and the processor calculates automatically a Unified Parkinson’s Disease Rating Score (UPDRS) equivalent score for a subject with PD in the absence of clinician-directed motor tasks required to be performed by the subject according to a clinical UPDRS protocol.

[0035] In some embodiments, the processor is configured to determine automatically if the subject is a fluctuator, wherein the evaluation score indicates magnitude of response to received therapy, and wherein the processor determines the subject to be a fluctuator when the magnitude of response meets a significance threshold. The significance threshold may be selected from the group including but not limited to (a) a 30% improvement in a measure of the kinetic state; (b) an improvement of 14 points on the UPDRS scale; and (c) an improvement of at least 1 and preferably 1.15 Severity Level points. [0036] In some embodiments, the processor is configured to determine automatically if the fluctuator is a controlled fluctuator, by comparing the measure of kinetic state or evaluation score following a received therapy with a target value, wherein the processor determines the fluctuator to be a controlled fluctuator when the measure of kinetic state is less than the target value. The processor may also be configured to determine automatically if a controlled fluctuator has wearing off, by calculating a reduction in response to received therapy over a set period, wherein the processor determines the controlled fluctuator to have wearing off when, over the set period, the reduction is one or both of: (a) 1 or more Severity Level points; and (b) 14 UPDRSIII points; and optionally, wherein the set period is 90 to 210 minutes and preferably 90 to 150 minutes and more preferably approximately 120 minutes following a received therapy.

[0037] In some embodiments, the processor is configured to determine automatically a PTB value calculated as proportion of time during the assessment period that the measures of kinetic state, or an evaluation score calculated from the measures of kinetic state, indicate that the subject exhibits movement disorder symptoms corresponding to bradykinesia.

[0038] In some embodiments the processor is configured to perform, automatically, one or more of: (a) identifying a subject having a calculated PTB value less than 30% as having controlled motor symptoms; and (b) monitoring a change in PTB value over time for the subject and identifying a subject having PTB over 30% and increasing as requiring modified therapy.

[0039] In some embodiments, the processor is configured to determine automatically a PTD value calculated as proportion of time during the assessment period that the measures of kinetic state, or an evaluation score calculated from the measures of kinetic state, indicate that the subject exhibits movement disorder symptoms corresponding dyskinesia.

[0040] In some embodiments, the processor calculates an evaluation score which characterises the subject’s response to a first therapy after waking on one or more days during the assessment period, and the processor automatically translates the evaluation score to a clinical score on a clinically accepted scale. [0041] In some embodiments, the processor processes the one or more measures of kinetic state to identify automatically a plurality of distinctive occasions during the assessment period, wherein a distinctive occasion is selected from a group including but not limited to: a) the subject receiving a therapy; b) the subject exhibiting a therapeutic motor response to a received therapy; c) after the subject has woken from night time sleep; d) after the subject has woken from night time sleep and before first therapy of the day; e) after he subject has woken from night time sleep and after first therapy of the day; and f) after a period of time exceeding a duration for which therapy is known to be effective for the subject in treating disordered movement.

[0042] In some embodiments, the processor automatically identifies in the data unreliable data segments, such as when there is one or more of a) motion data indicating the subject is already responsive to therapy at a distinctive occasion corresponding to the subject receiving a therapy; b) motion data indicating excess variability in responsiveness to therapy; and c) motion data indicating inactivity during periods of expected wakefulness. In some embodiments, the processor transmits an operational signal to a report generating processor, and the report generating processor generates a report including a notification of unreliable reporting when the processor identifies in the motion data unreliable data segments.

[0043] In some embodiments, the system includes a wrist-wearable device configured to collect the time-marked motion data. In some embodiments, the wearable device continuously records passive movement signals from the subject while ordinarily ambulatory and in the absence of clinician-directed motor tasks performed by the subject.

[0044] In some embodiments, the processor is configured to receive time-marked therapy data indicative of therapy received by the subject during the assessment period. In some embodiments, the system may include an input module configured to enable to processor to receive one or more of: a) a medication input, indicative of the subject having received a dose of therapy; b) time-marked therapy inputs from a dispensing device and indicative of the subject having received a therapy; and c) time-marked therapy data initiated by a user and indicative the subject having received a therapy. [0045] In some embodiments, the system includes a report generating processor that generates automatically a report containing one or more evaluation scores. In some embodiments, a report generated by the report generating processor contains a recommendation for modifying a prescribed therapy when the evaluation scores indicate that the subject is undertreated.

[0046] In some embodiments, the processor calculates a recommendation for modifying a prescribed therapy. In some embodiments, a recommendation for modifying a prescribed therapy calculated by the processor or generated by the report generating processor and may be used in a feedback control system controlling operation of a dispensing device that dispenses the prescribed therapy to the subject.

[0047] In some embodiments, the processor calculates automatically changes in one or more of the evaluation scores since one or more prior assessment periods.

[0048] In some embodiments, the processor automatically relates an evaluation score calculated by the processor to a score on a clinically accepted scale.

[0049] In some aspects, the present disclosure provides a machine automated method for evaluating a movement disorder disease state in subject, using passively collected motion data, the method comprising: receiving at a processor time-marked motion data from a device worn by the subject over an assessment period; the processor processing the received motion data to produce one or more measures of kinetic state of the subject that are indicative of movements of the subject that are attributable to symptoms of the movement disorder; the processor processing the one or more measures of kinetic state to produce one or more motor function severity levels related to a clinical scoring range; and the processor calculating one or more evaluation scores from the one or more calculated motor function severity levels and transmitting an operational signal to a user interface on which the one or more evaluation scores are presented to a user; wherein the one or more evaluation scores represent the evaluated state of the movement disorder for the subject during the assessment period.

[0050] In some aspects, the present disclosure provides a system for evaluating a movement disorder disease state in subject, using passively collected motion data, the system comprising a processor, a user interface and a memory module containing code corresponding to instructions causing the processor to: (a) receive time-marked motion data from a device worn by the subject over an assessment period; (b) process the received motion data to produce one or more measures of kinetic state of the subject that are indicative of movements of the subject that are attributable to symptoms of the movement disorder; (c) calculate one or more evaluation scores from the one or more calculated measures of kinetic state; and (d) transmit an operational signal to the user interface on which the one or more evaluation scores are presented to a user; wherein the one or more evaluation scores represent the evaluated state of the movement disorder for the subject during the assessment period.

[0051 ] It is to be noted that any one of the aspects mentioned above may include any of the features of any of the embodiments of other aspects mentioned above and may include any of the features of any of the embodiments described below, as appropriate.

[0052] It is to be understood each of the various aspects described herein may incorporate features, modifications and alternatives described in the context of one or more other aspects, such as but not limited to the various kinetic states, measures for evaluation score and other data characteristics used in the determination of an evaluation score. For efficiency, such features, modifications and alternatives have not been repetitiously disclosed for each and every aspect although one of skill in the art will appreciate that such combinations of features, modifications and alternatives disclosed for some aspects apply similarly for other aspects and are within the scope of and form part of the subject matter of this disclosure.

Brief Description of Drawings

[0053] The present invention will now be described in greater detail with reference to the accompanying drawings. It is to be understood that the embodiments shown are examples only and are not to be taken as limiting the scope of the invention as defined in the provisional claims appended hereto.

[0054] Fig. 1 is a schematic illustration representing a machine automated method for evaluating a movement disorder state in a patient using passively collected motion data. [0055] Fig. 2 is a schematic illustration representing a system for automated evaluation of a movement disorder in a subject using passively collected motion data.

[0056] Fig. 3 is a stylised representation of one day of motion data recorded from a subject with PD. [0057] Fig. 4 shows the relationship between %AUPDRS (Y axis) and absAUPDRS (X axis).

[0058] Figs 5A and 5B are box and whiskers plots of the distribution of Class 0 and class 1 in Tables 8 and 9.

[0059] Fig. 6A shows the range of total L-Dopa Equivalent dose (LED) and the dose of L-Dopa. Fig. 6B shows the LED from the first dose of levodopa (1 ^st dose) and from D2 agonists over the course of the day. Fig. 6C shows the percentage of the LED contributed to by D2 agonists.

[0060] Figs 7A and 7B show the change in LR according to duration of disease (in years). Fig. 7C shows absAPKG before and after deep brain stimulation (DBS). Fig. 7D shows the same data, with the difference in absAPKG before and after DBS (X axis) plotted against the absAPKG before DBS.

[0061] Fig. 8 outlines a process executed by the processor to categorise a subject as a non-fluctuator or fluctuator, and to sub-categorise the subject with in those two groups. [0062] Fig. 9A (not to scale) illustrates response to therapy and wearing off, Fig.

9B represents graphically 7 types of responsiveness which are described in Table 2

[0063] Fig. 10 plots, for PwP and controls, the percent of time spent in Motor Function Severity Levels 3-5 against various ranges of the percentage of time in bradykinesia (PTB). Detailed Description

[0064] A Levodopa (L-Dopa) challenge test may be given to subject if the subject’s responsiveness to a known dosage of L-Dopa is sought to be determined for the purpose of supporting a diagnosis of Parkinson’s disease (PD) or the decision to use certain therapies such as deep brain stimulation (DBS). The L-Dopa challenge test (LDCT) requires a clinician to perform a detailed question-based assessment of the subject using the Unified Parkinson’s Disease Rating Scale (UPDRS). While experienced neurologists are adept at performing such assessments, this is nevertheless a largely subjective exercise. Brain imaging can be used to provide detailed images of the dopamine system in the brain providing objective data to supplement an assessment of movement disorders such as PD. However, confirmation by a clinician after a thorough medical examination is still necessary. Brain scans do not support decisions to concerning advanced therapies including DBS.

[0065] Despite such tests and assessment tools being available for use in PD diagnosis, prior to the current invention there has been no instrumented, objective test for determining responsiveness to medications aimed at treating the symptoms of movement disorders. More specifically, there has been no instrumented, objective test for determining responsiveness to L-Dopa and other medications aimed at treating the symptoms of PD.

[0066] Embodiments of the present invention provide novel methods and systems for automated evaluation of a movement disorder disease state in a subject using passively collected motion data. One particular application is in the monitoring of PD state although other movement disorders (such as e.g. Multiple System Atrophy (MSA), Progressive Supranuclear Palsy (PSP) and Huntington’s disease) may also be evaluated using the current technology. Evaluation may include but is not limited to diagnosing, staging, measuring stage of progression, measuring rate of progression, or determining another specific criterion or classification having clinical relevance to the assessment of the subject’s movement disorder disease or PD state. One such classification, in the case of PD, is whether or not the subject is responsive to therapy and if responsive, the extent to which the response is sustained.

[0067] Fig. 1 is a schematic illustration representing a machine automated method 1000 for evaluating a movement disorder (such as PD) state in a patient using passively collected motion data. The method comprises, in a step 1001, receiving at a processor time-marked motion data obtained from a device worn by the subject over an assessment period. In a step 1002, the processor processes the received motion data to produce one or more measures of kinetic state of the subject that are indicative of movements of the subject that are attributable to symptoms of the movement disorder. For the special case of PD, these symptoms may include one or more of bradykinesia (BK), dyskinesia (DK) and tremor. These measures of kinetic state may be determined using any suitable method which involves the processing of passively collected motion data obtained from the device worn by the subject. The measures of kinetic state may include one or more of a measure of bradykinesia (BK), a measure of dyskinesia (DK), limb kinematics and the distribution features of limb kinematics. In a step 1003, the processor calculates one or more evaluation scores from the one or more measures of kinetic state and, in a step 1004, transmits an operational signal to a user interface on which the one or more evaluation scores are presented to a user. The one or more evaluation scores represent the evaluated state e.g. of PD for the subject during the assessment period.

[0068] In some embodiments, method 1000 further includes a step 1005 of receiving a therapy input, indicative of the subject having received a dose of therapy. The therapy input may be received via the device worn by the subject. In other embodiments, the therapy input is received from a dispensing device. In some embodiments, method 1000 further includes the step of automatically activating the device to provide medication reminders to the subject according to a prescribed medication regimen.

[0069] Fig. 2 is a schematic illustration representing a system 2000 for automated evaluation of a movement disorder (such as PD) in a subject, using passively collected motion data. The system 2000 includes a wearable device 2001 configured to generate motion data while the subject is ordinarily ambulatory and in the absence of clinician-directed motor tasks performed by the subject. A processor 2003 processes the motion data to produce one or more measures of kinetic state of the subject that are indicative of movements of the subject that are attributable to symptoms of the movement disorder. In the case of PD, the measures of kinetic state may include e.g. a measure of BK, DK and tremor. Processor 2003 calculates one or more evaluation scores from the one or more measures of kinetic state, and generates an operational signal that is provided to a user interface 2005. Ideally, the system 2000 includes an input module 2002 for receiving signals from the device 2001 and an output module 2004 for delivering control signals to one or more user interfaces 2005. This may be directly or indirectly, via a communication network 2006. Evaluation scores calculated by processor 2003 may be presented to a user, such as a clinician and/or a subject/patient on a user interface 2005 such as a screen of a computer, tablet, mobile device, medical device or printed report.

[0070] The user interface 2005 may be co-located with processor 2003. However, in some arrangements processor 2003 is located at a different physical site which may be in a different city, state or country from the subject (and/or clinician) and the user interface 2005 on which the evaluation score is presented. In some cases, there may be multiple user interfaces 2005 and multiple devices 2001 at multiple sites. Communication network 2006 enables implementation of a distributed system according to embodiments of the invention.

[0071] Advantageously, the motion data utilised is passively collected in that it comprises signals recorded from the subject while ordinarily ambulatory and in the absence of clinician-directed motor tasks performed by the subject. This means that the motion data can be collected while the subject is wearing the data collection device, during ordinary daily activities while at home, work, socialising or performing errands or the like. Admission to a hospital or clinic is not required, and none of the task-based assessments used by neurologists embodied in the UPDRS are necessary. Typically, the device is a wrist worn device, and the motion data is accelerometer data from an accelerometer in the device although it is to be understood that the motion data signal may be generated by other motion sensing components or other sensors such as gyroscopes, magneto sensors, optical or pressure sensors and the like.

[0072] According to some embodiments, the Parkinson's KinetiGraph® device and associated proprietary software and algorithms (PKG, Global Kinetics Pty Ltd) is used to provide continuous objective measurements of the subject's movement. Advantageously these objective measurements are able to be collected during activities of daily living and the PKG system delivers measures of kinetic state including, inter alia, BK, DK and tremor scores. The PKG system consists of a wrist- worn data logger (device), and a processor executing a series of algorithms that produce data points for 2-minute epochs. [0073] The device 2001 is typically worn on the most affected wrist and contains a rechargeable battery and a 3-axis iMEMS accelerometer (ADXL345 Analog Devices) set to record 11 -bit digital measurement of acceleration with a range of ± 4 g and a sampling rate of 50 samples per second using a digital micro-controller and data storage on flash memory. The device can be programmed to vibrate for 10 seconds at the time when therapy should be taken, as a reminder to the subject to take medications. Following a reminder, the device may receive a therapy acknowledgement by the subject pressing a button or swiping a screen of the device to indicate that medication has been consumed. Preferably device 2001 is water resistant. At the end of the assessment period, motion data recorded by the device 2001 are uploaded to the cloud via communication network 2006 for application of the algorithms executed by processor 2003. However, it is contemplated that in some embodiments, motion data could be uploaded to a remote processor during the assessment period, or the processor may reside within the device 2001 itself.

[0074] Preferably, the algorithms are for generating measures of kinetic state.

Such algorithms may be built using an expert system approach to model neurologists’ recognition of BK and DKfrom accelerometry data. Inputs to the expert system may include Mean Spectral Power (MSP) within bands of acceleration between 0.2 and 4 Hz, peak acceleration, and the amount of time within e.g. 2 minute epochs in the motion data that there was no movement. These inputs area ideally weighted to model neurologists’ rating of BK and DK and to produce a BK score and DK score for every 2 minute epoch of data.

[0075] Suitable methodologies for determining measures of kinetic state of the subject from the received motion data are disclosed in W02009/149520 entitled "Detection of Hypokinetic and/or Hyperkinetic States", the entire disclosure of which is hereby incorporated herein by reference. In that methodology, the algorithm for automated calculation of a BK score arises from knowledge that bradykinetic subjects (such as those suffering from PD) have longer intervals between movement and when they do move it is with lower acceleration. Thus, in that algorithm, a lower BK score suggests more severe bradykinesia whereas a high BK score indicates little or no bradykinesia. Notably, although not explicitly stated in that the disclosure, the BK scores generated according to that algorithm are negative values (attributable to the logarithmic function employed) such that a more negative score indicates more severe bradykinesia. For ease of clinical interpretation, the sign of the BK score calculated according to that algorithm is typically inverted and it is the inverted BK score which is employed in embodiments of the present disclosure. However, it is to be understood that alternative motion data processing methods and movement scoring regimens may be adopted.

[0076] Graphical representations plotting BK scores and DK scores against the time of day and showing the time that medications were due and consumed make it possible for a clinician inspecting the graphical representation to assess whether there were dose related variations in BK score or DK score. However, the graphical representation falls short in terms of providing an objective and quantitative analysis or evaluation of the subject’s PD state; clinician interpretation and curation of data presented in the graph is still necessary. Clinician interpretation is obviated with advances presented by the current invention.

[0077] Responsiveness to levodopa is often assessed with the levodopa challenge test (LDCT), following a protocol in which the test is performed in the morning, having ceased L-Dopa for 12 hrs and dopamine receptor 2 agonists (D2 agonists) for 24 hrs. In this protocol, the clinical state is assessed in the absence of medication (the OFF” state), 300 mg of levodopa is then administered, and the clinical state is reassessed between 45-60 minutes later, when the best clinical state will have been achieved (the “ON” state). The improvement from OFF to ON is known as the levodopa response (LR). BK is a central clinical feature of diminished dopamine transmission and this is assessed, along with tremor and limb rigidity, using a clinical scale known as the motor component (Part III) of the Unified Parkinson’s Disease Rating Score (UPDRS III).

[0078] In clinical assessment, LR is expressed either as the absolute difference in the clinical UPDRS III score (absA) or as a percentage of the “OFF” UPDRS III score (%D). A %D of 30% is widely accepted as demonstrating responsiveness to L-Dopa although changes including 20%, 25%, 33%, >40% and 25%-50% are cited. Some centres measure the “ON” state at a specific time, typically 45 mins after therapy, whereas others establish a peak UPDRSIII score. In clinical LDCT assessments there is no uniformity in the size of the dose: most use an absolute dose ranging from 150 mg -400 mg of levodopa but others use some multiple of the usual morning dose. There is also no clarity around handling of D2 agonists. Some protocols recommend ceasing D2 agonists for 24 hours before the LDCT despite their long half-life leaving concerns regarding residual effects even after 24 hours. Although the first dose in the morning is typically the most consistent, dose responses can vary even to the point of failure, and a single clinical LDCT study may not fully capture the variability of the LR in a given subject. Despite these problems, the LDCT has become accepted as a clinical benchmark measure of responsiveness to L-Dopa. The LDCT requires an early morning trip to the hospital in the OFF” state and is thus inconvenient, uncomfortable and not without complications. It is also time consuming for clinical centres and subjects. The present invention provides alternative means of assessing responsiveness to therapy such as L-Dopa that eliminates these problems.

[0079] According to embodiments of the present invention, the subject wears a device containing a motion sensor for an assessment period of 6 days while taking their usual medications and attending to their usual activities in the home. There is no requirement for the subject to attend the hospital or clinic, or perform any clinician directed motor tasks as is the norm for a LDCT done in the hospital or clinic. Instead, the inventive automated method estimates the LR from the change in motor function measured by the device following the first levodopa dose in the morning (LREST) using a novel technique.

[0080] In various embodiments, the one or more measures of kinetic state of the subject that are determined by the processor are features determined from the distribution of the kinetic state measure e.g. in time or in frequency. For example, for a measure of kinetic state being bradykinesia, the processor may determine for the purpose of calculating one or more evaluation scores, a mean BK score over a moving window of the continuously collected passive motion data (e.g. of 10, 20 or 30 minutes), or a mean BK score for such a window for the 10th, 25th, 50th, 75th or 90th percentile, or for a range such as the interquartile range of BK scores over the window.

[0081 ] Kinetic state measures determined from the distribution of the kinetic state feature (in time or frequency) may be utilized by a processor in a machine automated method for evaluating a movement disorder disease state, such as PD state. Alternatively/additionally, kinetic state measures determined from the distribution of the kinetic state feature (in time or frequency) may be utilized by a processor building a computational model that is used in an automated method/system for calculating an evaluation score and/or otherwise evaluating a movement disorder disease state.

[0082] In some embodiments, method 1000 includes in a step 1006, the processor 2003 processing the one or more measures of kinetic state to identify automatically a plurality of distinctive occasions during the assessment period. A distinctive occasion has relevance in determining clinical features in the motion data that impact an evaluation score calculated by the processor 2003. For example, a distinctive occasion may be a time corresponding to the subject exhibiting a therapeutic motor response to a received therapy (“ON”). This may be determined by the processor identifying in the calculated measures of kinetic state a statistically significant change that is consistent with a change in movement behaviour that is responsive to therapy. In the case of a subject with PD, the processor 2003 may automatically identify the subject having received therapy when there is a change in a BK score which is consistent with the subject’s movement changing from untreated, bradykinetic movement to movement that is within a target range. In one example, this may be determined by a change equivalent to 22 (±11 SD) UPDRS III points or a change in 4 points in a computer calculated BK score. For example, a change of 4 points from a BK score of 28 to 24 represents a transition from out of target (>26) to in target (<26).

[0083] In another example, a distinctive occasion may correspond to the subject having received a therapy, as determined automatically from time-marked therapy data received at processor 2003, e.g. at a step 1005. In some embodiments, such a distinctive occasion may be automatically verified by the processor 2003 by checking for a contemporaneous distinctive occasion from the measures of kinetic state corresponding to a therapeutic motor response to a received therapy (confirmed “ON”).

[0084] Another distinctive occasion identified by the processor 2003 may correspond to a time after the subject has woken from night time sleep. This may be determined by the processor 2003 automatically identifying in the measures of kinetic state a period of consecutive epochs of data in which there has been little or no movement, followed by a period of movement consistent with movement behaviour after waking. In one example, this may be determined by detection of a sequence of data that gives rise to a stable BK score >40 (indicating little or no movement) followed by a BK score <40 (indicating movement). In another example, the processor 2003 may determine the subject has woken from night time sleep using the received motion data and threshold detection, to determine when the subject’s movement behaviour has changed from little or no movement (corresponding to sleep) to movement (which is consistent with a state of wakefulness).

[0085] Another distinctive occasion identified by the processor 2003 may correspond to a time after the subject has woken from night time sleep and before first therapy of the day (first OFF”). This may be determined by the processor 2003 automatically identifying the distinctive occasion corresponding to a time after the subject has woken from night time sleep, and also automatically identifying the distinctive occasion corresponding to movement behaviour after waking from night time sleep which is bradykinetic. In one example, movement which is out of target or bradykinetic is determined automatically by a BK score >26, following a sequence of data having a stable BK score >40 (indicating little or no movement).

[0086] Another distinctive occasion identified by the processor 2003 may correspond to a time after the subject has woken from night time sleep and after first therapy of the day (LREST). This may be determined by the processor 2003 automatically identifying the distinctive occasions corresponding to a time after the subject has woken from night time sleep followed by bradykinetic movement, further followed by a change in BK score corresponding to a therapeutic motor response to a received therapy. In one example, a mean BK score <26 is consistent with controlled BK (i.e. BK in an acceptable target range) and is therefore indicative of a therapeutic motor response to a received therapy.

[0087] Another distinctive occasion identified by the processor 2003 may correspond to a time after a period of time exceeding a duration for which therapy is known to be effective in treating disordered movement symptoms of PD (time >LR). Such a duration may be determined automatically by the processor 2003 according to thresholds programmed into a memory component of or associated with the processor, or, as may be the case for other distinctive occasions discussed herein, it may be determined according to a computational model based on data obtained from a plurality of subjects used to the train the model.

[0088] Typically, a distinctive occasion occurs at least daily and, in some cases, several times per day. For example, the distinctive occasions corresponding to the subject exhibiting a therapeutic motor response to a received therapy, or the subject having received a therapy, will occur several times per day when the subject is compliant with medication regimens requiring regular therapy during a 24 hour period. In contrast, the distinctive occasions corresponding to the events following the subject waking from night time sleep will occur only once daily.

[0089] Evaluation scores calculated by the processor 2003 may be quantitative or qualitative but in any event they are objective and repeatable, unlike clinical scores determined by humans. For consistency and familiarity with clinically accepted benchmarks, in some embodiments the processor 2003 automatically relates an evaluation score calculated by the processor to a score on a clinically accepted scale. For instance, processor 2003 may calculate automatically a UPDRS equivalent score for a subject with PD in the absence of clinician-directed motor tasks required to be performed by the subject according to the UPDRS protocol. Thus the processor 2003 may calculate an evaluation score which characterises the subject’s movement behaviour on a number of distinctive occasions e.g. in response to a first therapy after waking on one or more days during the assessment period, and automatically translate the evaluation score to a clinical score on a clinically accepted scale such as the UPDRS. In other embodiments, the processor 2003 may calculate an evaluation score which characterises the subject’s movement behaviour on a number of repeated distinctive occasions, e.g. in response to a subsequent therapy (i.e. not the first morning dose) throughout one or more days during the assessment period. In other embodiments, the processor 2003 may calculate an evaluation score which characterises the subject’s movement behaviour at intervals during the assessment period. Such intervals may correspond to windows of time such as e.g. 2 minute epochs.

[0090] An evaluation score generated according to embodiments of the invention may indicate one or several factors of interest to a clinician seeking to determine the state of a subject’s movement disorder, such as PD. One factor is whether or not the subject responds to therapy. Evidence of responsiveness to PD therapy such as L- Dopa provides support for the clinical diagnosis of PD and for assessing the suitability of device assisted therapies such as Deep Brain Stimulation (DBS) and delivery of apomorphine or levodopa by pump.

[0091] Because the inventive method uses passively collected motion data that is absent of data components corresponding to clinician directed motor tasks, a number of technical challenges arise. One is that in a standard assessment period undertaken while the subject is ordinarily ambulatory, the first dose of L-Dopa is usually not supramaximal and D2 agonists may be taken. Furthermore, some subjects with PD choose to continue to rest for some time after their first dose to allow it to take effect and this immobility may obscure the presence of bradykinesia in the motion data collected by the device worn by the subject. Beneficially, however it is typical for data to be available for more than one day and in some embodiments, 6 days of data are available (although it is to be understood that a shorter, 1 , 2, 3, 4 or 5 day assessment period, or a longer assessment period such as 10, 21 or 30 days could be utilised). It is also to be understood that in some embodiments, the assessment period is ongoing, as long as the motion detecting device is worn. This may address some of the natural variation in response to levodopa that will be overlooked by studying a single dose administration, including in the scenario of a clinic-based LDCT.

[0092] Of interest to clinicians is not only that the subject is responsive to therapy, but the manner in which the subject responds. Features of a subject’s response to therapy can inform an assessment of disease state and progression, appropriateness of therapy, and can in some cases inform modifications to therapy. Advantageously, embodiments of the invention evaluate automatically from the passively collected motion data, the manner in which a subject responds to therapy and not merely the binary fact of there being a response or not. This may be achieved by the processor 2003 calculating one or more evaluation scores. An evaluation score may be presented in absolute terms (e.g. minutes or seconds or a score relating to a clinical or other scoring scale), as a range, or as a percentage of a reference value. In some embodiments, the processor further calculates a percentage of time, during the assessment period, that an evaluation score calculated for a subject, exceeds a target/threshold. [0093] In one example, an evaluation score calculated by processor 2003 represents the time to maximum responsiveness to therapy. In the clinical environment, it is assumed that for a subject with PD the time to maximum responsiveness to L-Dopa therapy is 45 to 60 minutes after the medication is taken. Thus in some embodiments, an evaluation score representing time to maximum responsiveness to therapy may be represented as a time in minutes, or a percentage of a reference value (e.g. % of median for subjects with PD), In some examples, time to maximum responsiveness to therapy is detectable by the processor 2003 identifying in the calculated measures of kinetic state a time corresponding to a clinically significant change that is consistent with a) the onset of a change in movement behaviour that is responsive to therapy and b) a change in movement behaviour that is consistent with maximal response to therapy. It is to be noted that responsiveness to L-Dopa is not instantaneous. Rather, there is a time to peak responsiveness which varies between subjects and with disease state progression. Thus, time to peak responsiveness is clinically relevant for determining effectiveness of therapy and disease state progression.

[0094] In another example, an evaluation score calculated by processor 2003 represents the duration of response to received therapy. In the clinical environment, duration of response to L-Dopa may range from 30 minutes to 24 hours after the medication is taken. Thus, in some embodiments, an evaluation score representing the duration of response to therapy may be represented as a duration in minutes, or a percentage of a reference value and is typically representative of the subject’s response to therapy over the assessment period. In some examples, duration of response is determined by processor 2003 identifying in the calculated measures of kinetic state the time duration of consecutive epochs of data in which the subject has a mean BK score which is in target, e.g. <26. Since duration of responsiveness diminishes with disease progression, an objective evaluation score representing duration of response can be used to determine effectiveness of therapy and can be used by processor 2003 to calculate automatically changes to medication regimens by recommending e.g. higher doses to increase the duration of response.

[0095] In another example, an evaluation score calculated by processor 2003 represents the magnitude of the subject’s response to received therapy. In the clinical environment, magnitude of response to L-Dopa diminishes with disease progression and may also be used to determine PD state. In one example, magnitude of response to therapy can be determined automatically by processor 2003 determining the change in BK values before and after therapy is administered, i.e. while the measures of kinetic state (e.g. BK scores) indicate the subject is OFF” (BKOFF) before therapy, and ON” (BKON) after therapy is effective. The evaluation score may express magnitude of response to therapy in absolute terms (absA) equivalent to the difference between BKOFF and BKON scores, or may be a presented as a percentage value (%D) equivalent to the % of BKOFF. In some embodiments, the magnitude of response is converted to an accepted clinical scale, such as e.g. the UPDRS.

[0096] Further, processor 2003 may provide a qualitative evaluation score (e.g. fluctuator, non-fluctuator) based on the calculated absolute or percentage score to indicate if the magnitude of the response is considered “effective” or “ineffective” for treating the subject’s motor symptoms, or using other qualitative terms such as “almost ineffective” or “20% less effective than this time last year”. While these terms are qualitative, they are nevertheless objective and repeatable to the extent that they are determined based on recorded motion data that has been passively collected from the subject while ordinarily ambulatory, and not based on a clinician’s visual observation of changes in the subject’s motor function when performing motor tasks directed by the clinician in the hospital or clinic.

[0097] In another example, an evaluation score calculated by processor 2003 represents the subject’s day-to-day variability of response to received therapy over the assessment period. In one example, the processor achieves this by detecting and/or reporting on variability in one or more of (i) measures of kinetic state at the time a therapy is received (i.e. Dose Time), (ii) measures of kinetic state at the time a therapy has maximum therapeutic effect (Effect Time) and (iii) latency to Effect Time.

[0098] In another example, an evaluation score calculated by processor 2003 represents an extent to which the subject is treated/undertreated by received therapy. In one example, the subject is considered to be “treated” when the LR correspond to a % change in BKOFF that exceeds 20%, 25%, 33%, >40% or25%-50%. In another example the subject is considered to be treated when BK scores are within a target range, e.g. the BK score is <26, otherwise the subject is considered to be “untreated” or OFF.

[0099] The inventors note, however, that some level of BKwill become the boundary between “in target” and “in BK”. Thus, a plurality of motor function severity levels may be mapped to BK scores or the UPDRS III scale to quantify the extent to which the subject is ON and indeed the extent to which the subject is OFF. This provides additional objective and clinically useful insight to the severity of the subject’s BK symptoms that contribute to the movement disorder while OFF, which is not available from patient diaries. Table 1 shows the relationship between UPDRSIII scores and 6 intervals of Motor Function Severity Level (MFSL) and is utilised as a support set in the computational model described in the Example.

_ LEVEL _ 0 1 _ 2 _ 3 _ 4 5

Number or Class Labels 305 54 80 76 46 29

UPDRS Interval 0-10 10-22.5 22.5-35 35-47.5 47.5-60 > 60

Table 1

[0100] In some cases, the subject may be considered to be undertreated based on the amount of “OFF” time during the assessment period. For instance, in the case of subjects for whom the processor 2003 calculates the level of BK to be above target for more than 8 hours in the period between 07:00 and 22:00, the processor may determine the subject to be undertreated.

[0101 ] In other examples, an evaluation score calculated by processor 2003 represents an extent to which the subject is untreatable by therapy. This may be achieved by processor 2003 determining that there has been no response to therapy, in which case the evaluation score may be a qualitative indicator such as “untreatable” or a qualitative indicator such as “0” designating that the subject’s movement disorder has zero capacity to be treated using the therapy, or “misdiagnosed” designating that the subject has been misdiagnosed as having a movement disorder disease which is capable of being treated by administration of the therapy. Such use of the system may be deployed in testing assumptions that a subject diagnosed with PD does not actually have PD because they have not experienced or perceived reduced PD symptoms with administration of the therapy (e.g. L-Dopa)

[0102] In some cases, the processor 2003 may calculate an evaluation score designating the subject as an “alternator” or “fluctuator” when kinetic state measures alternate between being in target or in BK such that their time in BK is less than e.g. 8 hours or 50% of the day (for assessment periods monitored between 07:00 and 22:00) during one or more days of the assessment period, or on average over the assessment period. In some cases, the processor identifies automatically a subject as “undertreated” when kinetic state measures indicate that the subject is above a target BK score for more than 8 hours or 50% of the day (for assessment periods monitored between 07:00 and 22:00) during one or more days of the assessment period, or on average over the assessment period. In some cases, the processor determines automatically for the assessment period or part thereof, the proportion of time spent in BK (PTB) and is configurable to report on changes in PTB over time such as by comparison to prior assessment periods.

[0103] In other examples an evaluation score calculated by processor 2003 represents an extent to which the subject is a candidate for advanced therapy. This may be achieved by the processor 2003 determining from an assessment of a plurality of evaluation scores conducted over an extended period that (i) the subject is an “alternator”; and (i) the prescribed therapy regimen has been altered over the extended period with the aim of reducing alternations between being in target and being out of target; and (ii) the subject continues to be an alternator. In one example, alterations to the prescribed therapy regimen are calculated by the processor 2003 and may be implemented by transmitting a control signal to a dispensing device 2008 in the system.

[0104] Where the evaluation score is indicative of LR magnitude, duration or time to maximum response to therapy, a change since a prior assessment period may indicate that the subject’s disease has progressed to a point where they are undertreated by their existing therapy regimen.

[0105] The evaluation scores may be qualitative or quantitative. In an embodiment where the inventive method is used to replace or predict a LDCT, the evaluation score may be a score equivalent to and familiar with a score utilised by clinicians performing the LDCT in clinics with their patients. Thus, the evaluation score produced by the processor may be on a scale of e.g. 0 to 199 points, or another maximum score for the format of the prevailing UDPRS, or a % change, following administration of therapy.

[0106] In some embodiments the method includes processor 2003 pre-processing the received motion data to remove automatically one or more of: a) data corresponding to periods of night time sleep by the subject (e.g. by excluding data recorded between 22:00 and 7:00); b) data corresponding to periods the device was not worn by the subject (e.g. as may be determined by an accelerometer, capacitor or other sensor in device 2001); and c) data corresponding to periods of inactivity by the subject (e.g. as may be determined by an accelerometer in device 2001).

[0107] In some embodiments, where the movement disorder disease is PD, the processor 2003 determines automatically that the subject is a fluctuator when the processor determines to the subject’s response to therapy to be significant. Typically this is achieved by the processor 2003 calculating the magnitude of response to received therapy, and, in the case that the magnitude of response meets a significance threshold, identifying the subject as a fluctuator. In preferred embodiments, the significance threshold is e.g. a 30% improvement in the kinetic state measure (e.g. BK score (BKS) or mean BKS (mBKS)), an improvement of 14 points on the UPDRSIII scale or an improvement of at least 1 and preferably 1.15 on the MFSL.

[0108] For subjects who do not exhibit a significant response to therapy (i.e. the magnitude of response does not meet the significance threshold), the processor 2003 determines the subject to be a non-fluctuator. In some embodiments, the processor 2003 may further determine that the subject is a controlled non-fluctuator (low BK scores before the therapy (e.g. BK < 26) and therefore at an early stage of PD or not requiring treatment) or an uncontrolled non-fluctuator (high BK scores before the therapy, insignificant response to therapy and undertreated or unresponsive to therapy). [0109] For a subject identified by the processor as a fluctuator, the processor may further categorise the subject in one of four categories according to whether the magnitude of the response to therapy is sufficient to reduce the subject’s BK symptoms to be in target. Firstly, the processor determines automatically that the fluctuator is a controlled fluctuator when a BKS or mBKS or MFSL following a received therapy is less than a target value being the value at which the subject’s BK symptoms are considered to be treated. The processor determines automatically that the fluctuator is an uncontrolled fluctuator when a BK score (BKS) or mean BKS (mBKS) following a received therapy meets or exceeds the target value. In one embodiment, the target value is a MFSL of 2.5 or a BKS of 26. In some embodiments, the processor calculates an evaluation score comprising Active mBKS (AmBKS) being the median of BKS<42 from the assessment period (e.g. 6 days). AmBKS removes artefact from most sitting and other forms of inactivity from the assessment.

In one embodiment the target value for AmBKS is 23 and the subject is out of target when that value is exceeded.

[0110] To determine which of the four categories of fluctuator the subject belongs, the processor determines automatically if the subject’s response to therapy persists or if there is wearing off. The subject is determined to be in a first fluctuator category being fluctuator - controlled with persistent response (F-CP) when the processor 2003 determines that there is no wearing off . The subject is determined to be in a second fluctuator category being fluctuator - controlled with wearing off (F-CWO) when the processor 2003 determines wearing off to have occurred. The subject is determined to be in a third fluctuator category being fluctuator - uncontrolled with persistent response (F-UP) when the processor 2003 determines that there is no significant response to therapy but also no wearing off since the response, albeit non-significant, is sustained. The subject is determined to be in a fourth fluctuator category being fluctuator - uncontrolled with wearing off (F-UWO) when the processor 2003 determines that there is no significant response to therapy and the non-significant response wears off. Fig. 8 outlines a process executed by the processor to categorise a subject as a non-fluctuator or fluctuator, and to sub-categorise the subject with in those two groups.

[0111] Fig. 9A (not to scale) illustrates response to therapy and wearing off, where vertical lines represent therapy doses. The x axis shows time and the y axis shows BK increasing in severity toward the bottom of the graph. The shaded area represents the target range for BK. The magnitude of the response (LR, levodopa response) is determined by the difference in MFSL score (severity of bradykinesia) at effect time ET compared to dose time DT. An improvement in MFSL of 1.15 estimates an improvement of 14 UPDRSIII points which also approximates, in some patient samples, a 30% improvement in motor symptoms. Fig. 9A depicts where the severity of early morning bradykinesia at DT (first Dose Time) is outside the target range. The LR is the difference in bradykinesia severity (D1) at ET (46-90 minutes after the first dose) and at DT. Two examples of subsequent clinical response to the first dose are shown; dotted line B shows a persisting response without significant decline from the best LR response. Solid line (A) shows a case with wearing off which occurred when the MFSL increased by Ί” (D2) between 165 and 210 minutes after DT (depending on ET latency). Horizontal lines immediately under the target range indicate time above target and conceptually, the proportion of time in BK (PTB) which may be represented as a percentage of total time (t). Fig. 9B represents 7 types of responsiveness (see letters at the right of each curve) which are described and discussed in detail with reference to Fig. 8 and Table 2 which provides descriptions of categories of fluctuators and categories of responsiveness to therapy as determined according to embodiments of the invention.

Table 2

[0112] Thus, in one example, the processor 2003 determines wearing off to have occurred when the calculated measure of kinetic state or evaluation score evidences a reduction in the response to therapy over a time period. The time period may be e.g. 90 to 210 minutes and preferably 90 to 150 minutes and more preferably, 120 minutes. In one example, the processor determines there to have been wearing off when the response to therapy decreases by e.g. 1 or more MFSL points or 14 UPDRSIII points over a time period of 2 hours. [0113] In some embodiments, processor 2003 determines a PTB value for the subject, calculated as the proportion of time during the assessment period that the measures of kinetic state, or an evaluation score calculated from the measures of kinetic state, indicate that the subject exhibits movement disorder symptoms corresponding bradykinesia. Monitoring PTB can be a useful tool to monitor progression of disease state over time, since disease progression corresponds to higher PTB and objectively monitoring this parameter enables objective identification of subjects in need of modified therapy. In some embodiments, once therapy is modified, monitoring subsequent PTB enables objective evaluation of the effectiveness of the new therapy, since one should see a reduction in PTB. [0114] Thus, in one example, processor 2003 identifies in the received data, intervals of time when the subject exhibits bradykinetic motor symptoms. In one embodiment, processor 2003 determines the motor symptoms to be bradykinetic when the MFSL is above 2.5 MFSL points (that is, when estimated MFSL is 3, 4 or 5). Thus, PTB is estimated as the number of intervals or epochs of time when the subject is estimated to be in MFSL 3, 4 or 5 or when a UPDRSIII score estimated by the processor exceeds 35. Preferably, the processor excludes from the calculation of PTB segments of time corresponding to periods of night time sleep by the subject, periods of time that the device was not worn by the subject, and periods of inactivity by the subject.

[0115] In one embodiment, processor 2003 determines automatically that for a subject having a calculated PTB value less than 30% the motor symptoms are “controlled”. In another example, processor 2003 monitors changes in PTB over time for the subject and identifies a subject having PTB of at least 30% and increasing as requiring modified therapy.

[0116] It is to be noted that the 30% threshold for determining if a subject’s motor symptoms are “controlled” is not limiting and has been determined based on an investigation and analysis of data (using motion data and algorithms disclosed herein) collected from 228 PwP and 157 control subjects, where the 90th percentile of PTB in controls was 32.7%, and PTB of PwP with mBKS<23 were similar to controls whereas 90% of those PwP whose mBKS>23 had higher PTB. Thus, a threshold of around 30% such as for example 25%, 26%, 27%, 28%, 29%, 31%, 32%, 32.7%, 33%, 34% or 35% or thereabouts may be appropriate. Larger and more diverse data sets and may refine the threshold value.

[0117] The calculated PTB provides an estimate of the time spent above the target range, and percent time in Level 5 (UPDRS lll>60) provides an estimate of the time spent at the subject’s most severe level of bradykinesia. Fig. 10 plots, for the PwP and control subjects investigated, the percent of time spent in MFSL 3 and 4 and MFSL 5 against various ranges of the PTB (x axis). This figure shows that as PTB increases it is as a result of increase time with high levels of bradykinesia (i.e. in Severity Level 5: UPRS lll>60). This figure indicates that when PTB is around 45%- 60%, ~2/3 of the PTB is at Severity Level 5 and implies that OFF UPDRS III scores have become greater than 60. The objective analysis provided by processor 2003 in this regard provides richer clinical analysis of the subject’s motor function symptoms than the simple assessment of ON/OFF which is self-reported by PD subjects using patient diaries. [0118] In some embodiments, processor 2003 determines a PTD value for the subject, calculated as the proportion of time during the assessment period that the measures of kinetic state, or an evaluation score calculated from the measures of kinetic state, indicate that the subject exhibits movement disorder symptoms corresponding dyskinesia. In one embodiment, processor 2003 determines the motor symptoms to be dyskinetic when a DK score (DKS) calculated by the processor is above 10, and excluding time during which the processor identifies walking or tremor in the received motion data. In one embodiment, walking may be detected in the motion data using supervised gradient boosting decision tree model to identify walking activity with sufficient energy to influence the DK scores. This approach may use features obtained from conventional gait detection algorithm and from the resonance of the acceleration signal during the time interval under examination. Tremor may be identified using known tremor detection algorithms. Time intervals for which the processor 2003 has identified “walking” are excluded by the processor since it is possible for dyskinesia and walking to occur in the same interval and thus, these epochs are uninterpretable. If the processor detects tremor in an interval and the DKS > 10, then the DKS for that interval is set to zero for the purpose of estimating PTD. Thus, the PTD is estimated by processor 2003 as the number of intervals or epochs of time when the subject is estimated to have a DKS>10, preferably excluding those intervals of time corresponding to periods of night time sleep by the subject, periods of time that the device was not worn by the subject, and periods of inactivity by the subject.

[0119] The number of intervals removed due to sleep, inactivity, non-worn monitoring device, and in the case of PTD, walking and tremor, will vary from person to person. Therefore, time in BK and time in DK are ideally expressed as a percentage or proportion, allowing objective comparison between subjects. Measuring PTB and PTD using an objective ambulatory measurement system for PD show promise as tools for aiding in the routine care of PD and as outcome measures in clinical trials of PD therapies and management.

[0120] In some embodiments, the processor calculates one or more evaluation scores by applying a computational model to the received motion data. The following example is provided to demonstrate an example of one such computational model. However, it is to be understood that other models could be devised which perform to an acceptable level of accuracy in calculating an evaluation score (e.g. in the automated prediction of a subject’s performance of a LDCT) using continuously passively collected motion data while the subject is ordinarily ambulatory, typically at home or in their usual environment, and performing tasks of their typical everyday living, in the absence of clinician-directed motor tasks.

Example

[0121] Sample motion data was collected from 199 people with PD who attended one of three clinics for a LDCT as part of routine clinical care. Then, subjects were excluded if; a) the motion sensing device had not been worn at or near the time of the LDCT (median 6 days and 75th percentile 22 days); b) it was documented that PD medications were consumed prior to a device-generated medication reminder; c) PD medications were used at the time of the first dose whose pharmacological profile would mask or confound the usual response to L-Dopa (apomorphine, levodopa in intestinal gel formulation, levodopa in extended release capsules (Rytari) but not D2 agonists or entacapone). Of the original 199, 48 were excluded for one or more of the above reasons. The clinical features of the remaining 151 subjects with PD are shown in Table 3.

Clinic (Mean ± SD (number of case where data is available)

Clinical parameter 1 2 3

UPDRS III OFF 54 ± 15 (53) 51 ± 11 (49) 41 ± 12 (49) UPDRS III ON 27 ± 10 (53) 34 ± 9 (49) 18 ± 8 (49) absAuPDRS ¹ 27 ± 11 17 ± 10 24 ± 8

% AUPDRS* 51 ± 14 32 ± 15 58 ± 13 Age 61.6 ± 6.7 (53) 65.6 ± 9.1 (49) 58.8 ± 12.0 (2)

Disease duration 10.0 ± 4.1 (47) 9.9 ± 5.1 (49) - (0) D2 agonist LEDD§ 20 ± 16 (53) 10 ± 12 (49) 16 ± 32 (49)

LEDD 869 ± 386 (53) 1226 ± 745 (49) 979 ± 544 (49)

LEDD 1st dose 138 ± 55 (50) 194 ± 114 (49) 162 ± 56 (49)

§ expressed as a percentage of total LEDD c absolute difference between “OFF” to “ON” UPDRS III * difference between “OFF” to “ON” UPDRS III expressed as percent of “OFF” UPDRS III

Table 3

[0122] Additionally, sample motion data was collected from 191 people without PD or neurological disorder and aged over 60 years (Controls) for an assessment period of 6 days using a motion sensing device worn on the wrist. BK scores were calculated for the assessment period and Control subjects whose BK scores were >90th percentile were removed leaving 174 Control subjects. In this example, the purpose of including Control subjects was to provide a large cohort of people who would not be bradykinetic in the early morning and whose BK score would not change after a nominal “dose time” (DT). For building the computational model, DT for the Control subjects was chosen to be 07:00 to ensure that levels of sleep and inactivity would be similar to the first dose in the morning for a subject with PD.

[0123] The UPDRS is a formalised version of the clinical examination with a series of levels (usually 0-4) corresponding to increasing severity of motor function. The score of each of the questions is summed to give a total score. It is important to note that the UPDRS scores tremor, rigidity and bradykinesia, with tremor and rigidity encompassing approximately 40% of the total score whereas the current example requires measures of kinetic state for only bradykinesia and tremor.

[0124] The UPDRS score was deemed to be “0” at DT. The “effect time” was obtained from a smoothed time series of the BK score of all the epochs in the period from 46 mins to 90 mins after the nominated DT from all days in the assessment period (6 days). The lowest BK score in this time series became the peak (i.e. the “effect time”). Fig. 3 is a stylised representation of one day of motion data recorded from a subject with PD. The Y axis shows the BK score in BKS units and the X axis is time in minutes, before the first reminder of the morning (vertical dotted line marked A at “0” time). The subject’s acknowledgement that the dose was consumed is shown as a diamond at 301. The dots represent individual BK scores for each two minute epoch: dots appearing below the x axis represent epochs that lie in the “inactive” range and with linearly weighted moving median of BKS > 40 (indicating sleep or inactivity). Dots appearing above the x axis represent epochs that lie within the active” range”, where active epochs are those epochs whose linearly weighted moving median of BK score <0. The line C represents the smoothed time series from all 6 days of recording, with the heavy line being from 46 minutes to 90 minutes after the first acknowledgement of the dosage reminder (diamond 301 at approx. 6 mins). The shaded area marked S and E shows the ten minute period (5 epochs, circled) used to establish the BK scores at DT and around the time of the peak response to therapy (Effect time, ET), respectively. [0125] To design the computational model that estimates the LR, the UPDRS was divided into a plurality of motor function severity levels (MFSLUPDRS). This was to allow severity of motor function to be categorically classified instead of requiring regressions. This allowed more robust employment of the sample motion data set with the non-identically distributed noise and variations in UPDRS scoring, which introduces complexity for regression models. MFSLUPDRS “0” was set as the level for a UPDRS score <10 and levels above that were separated by increments of 12.5 UPDRS units, with all UPDRS scores > 60 in level 5 (see Table 1). It is to be understood that larger or smaller increments in UPDRS score could be adopted. Table 1 shows the number of UPDRSOFF and UPDRSON in each MFSLUPDRS.

[0126] Measures of kinetic state used to objectively measure motor function from the motion data obtained from the device worn by the subjects were then assigned to one of the 6 MFSL class labels using the corresponding UPDRS score. Table 4 shows candidate predictor features and their marginal mutual information (Ml) of candidate features with six target classes. Rows in italics were features selected in the refinement process.

Marginal

Feature Name Description

MI

BK BK Score 0.04

BKSJA10P 30 min window moving 10th percentile ofBKS 0.3

BKSJA25P 30 min window moving 25th percentile ofBKS 0.24

BKSJA50P 30 min window moving 50th percentile ofBKS 0.17

BKS_M75P 30 min window moving 75th percentile of BKS 0.16

BKSJA90P 30 min window moving 90th percentile ofBKS 0.22

BKS_WM10P 30 min window weighted movinglOth percentile of BKS 0.22

BKS_WM25P 30 min window weighted moving25th percentile of BKS 0.20

BKS_WM50P 30 min window weighted moving50th percentile of BKS 0.17

BKS VM75P 30 min window weighted moving75th percentile ofBKS 0.18

BKS_WM90P 30 min window weighted moving90th percentile of BKS 0.2

TA Tremor Amplitude 0.04

30 min window moving 50th percentile of Tremor

TA_M50P 0.07 Amplitude

30 min window weighted moving 50th percentile of

TA WM50P 0.08 Tremor Amplitude

Logl0(l + 30 min window weighted moving 50th percentile of

TA_WM50P_Log 0.1 Tremor Amplitude)

Table 4 [0127] The multi-class classification problem was decomposed into 5 binary classification problems, where samples were separated according to whether they fell below or above the threshold for each class. For instance, Classifier 3 was a binary classifier separating UPDRS < 35 and UPDRS > 35.

[0128] A set of candidate features (enumerated in Table 4) for performing the classification task were extracted from the sample motion data and statistically matched to the class labels in Table 1. When dealing with a relatively small dataset, having many structurally dependent features can degrade the performance of the statistical model and risks overfitting due to noise induced variance. What follows is a reduction in the size of the feature space by elimination and selection to enhance the model generalisability and interpretability in a supervised manner. The first phase of feature selection was to identify features of same nature and select the most relevant. For example, the two features BKS_M50P and BKS_WM50P in Table 4 are of the same nature in that they are both moving medians of BK Score and would thus contain similar information.

[0129] The mutual information (Ml) test was performed to assess the relevance of these features to the target classes. This approach is neutral with respect to models and identifies any statistical relationship, linear or nonlinear, between the features and the class labels. Table 4 shows the marginal Ml of each feature, estimated using 6 nearest neighbours for these continuously valued features, with the 6 class labels.

[0130] The feature set was refined to BKS_M10P, BKS_M25P, BKS_M50P, BKS_WM75P, BKS_M90P and TA_WM50P_Log. Although these scores each have considerable relevance, some carry redundant information with respect to one or a combination of the others. Joint Mutual Information Maximisation (JMIM) was used to maximise relevancy while minimising redundancy. This method first picks the feature with maximal Ml with target classes and adds it to the set of selected features. It then iteratively adds the feature for which the minimum joint mutual information together with any of the already selected features is maximum among other candidates for selection. This heuristically ensures that a newly selected feature has greater relevancy and less redundancy. Table 5 shows the Joint Mutual Information Ranking of a refined set of candidate features and the Ml of their Weekly Aggregate Mean with the six target classes. Row Feature Joint MI Ranking Marginal MI (Weekly Aggregate)

1 BKS_M10P 1 (0.3) 0.17

2 BKS_M25P 2 (0.3) 0.19

3 BKS_M90P 3 (0.28) 0.04

4 BKS_M50P 4 (0.24) 0.12

5 BKS_WM75P 5 (0.19) 0.06

Table 5

[0131 ] Also shown are the marginal mutual information of Weekly Aggregate Mean of each feature, i.e. the average of values of the features over a 10 minutes interval of entire week, with target classes. This shows the relevance of a feature when aggregated at DT. Evidently, BKS_M90P and BKS_WM75P were relatively less relevant. Furthermore, BKS_M50P was relatively redundant and was also less relevant after weekly aggregation. Although there was the high level of collinearity expected between BKS_M10P and BKS_M25P, both features were deployed as they represent different nonparametric measures of the temporal distribution of BKS. In one case, the following set of features found in rows 1 , 2 and 6 of Table 5 were selected following assessment of relevancy and redundancy:

BKS_M10P

BKS_M25P

TA_WM50P

[0132] Several discriminative statistical models were examined for the purpose of designing the five binary classification algorithms. Table 6 shows cross validation (CV) and training performance metrics on Weekly Aggregate Mean predictions for candidate classifier models. Although the selected features are monotonically and linearly correlated with the UPDRS (see Table 6), linear and nonlinear models were considered: namely Logistic Regression, Support Vector Classifier with Radial Basis Function (RBF) kernel and Gradient Boosting Decision Trees. These discriminative models represent categories of linear, kernel nonlinear and arbitrarily nonlinear approaches which have proved to be effective in a variety of problem types. Classifier

1 2 3 4 5

CV 0.77 0.79 0.87 0.87 0.83

ROC AUC

Logistic Train 0.78 0.8 0.87 0.88 0.85

Regression CV 0.75 0.78 0.81 0.72 0.58

PR AUC

Train 0.76 0.79 0.83 0.74 0.59

CV 0.77 0.79 0.87 0.87 0.83

ROC AUC

PCA + Logistic Train 0.77 0.79 0.87 0.88 0.85 Regression CV 0.75 0.77 0.81 0.72 0.58

PR AUC

Train 0.75 0.78 0.83 0.74 0.6

CV 0.69 0.77 0.85 0.84 0.66

ROC AUC

Train 0.76 0.78 0.87 0.87 0.91

SVC- RBF Kernel -

CV 0.67 0.75 0.78 0.69 0.54

PR AUC

Train 0.74 0.75 0.81 0.76 0.75

CV 0.77 0.77 0.86 0.86 0.8

Gradient ROC AUC

Train 0.8 0.81 0.89 0.89 0.88

Boosting -

CV 0.75 0.75 0.79 0.69 0.57

Decision Trees PR AUC

Train 0.79 0.8 0.85 0.78 0.65

Table 6

[0133] When comparing the performance of the learning models described above, it was noted that Logistic Regression assumes that the observations in the dataset are independent and that multicollinearity is not present. However, the structural collinearity between BKS_M10P and BKS_M25P has been noted above therefore there is a violation of this assumption on Logistic Regression. Therefore, a fourth model was considered, where unsupervised reduction of the dimension of these two features into one was achieved using the first component of Principal Component Analyses (PCA) of BKS_M1 OP and BKS_M25P together with TA_WM50P_Log, followed by a Logistic Regression classifier. As the employed features are moving statistics, neighbouring epochs (observations of the same subject) are not independent when 5 epochs in a row are sampled. Two strategies were used to address assumptions regarding independence of observations: (i) using other candidate models whose requirements around this assumption are relaxed (SVM and Gradient Boosting Trees), (ii) assessing the performance of the models both in terms of model selection and out of sample predictions on Weekly Aggregate predictions, where two observations from one subject (weekly aggregate of dose time and effect time) are spaced more than 30 minutes apart and are thus reasonably independent. After deploying these strategies, the use of Logistic Regression as one of the candidate classifiers was surprisingly found to be possible with meaningful performance comparisons.

[0134] For each of the classification problems (i.e. into one of the levels defined in Table 1 ), the Sample data was divided into a training set and a testing set. Two performance measures were monitored: (i) area under curve of Receiver Operating Characteristics (ROC AUC) which provides an average Sensitivity at different Specificities; (ii) area under curve of Precision (ratio of true positives to all positive declarations including false positives) vs. Recall (=Sensitivity) (PR AUC) which provides an average Precision at different Recalls averaged over the two classes.

The performance metric (ii) is reported to account for the class imbalance that varies over the five Classifiers.

[0135] Surprisingly, the Logistic Regression model performs as well as if not better than the other models and is simpler in that it enhances the likelihood of generalisability and interpretability of the algorithm. The generalisability of the classifier models was examined by applying them to an unseen test set. Table 7 shows the performance of each classifier model on train and test sets.

Classifier

_ 1 _ 2 _ 3 _ 4 _ 5_

ROC Train Set 0.78 0.8 0.87 0.88 0.85

AUC Test Set 0.79 0.88 0.85 0.83 0.82

PR Train Set 0.76 0.79 0.83 0.74 0.59

AUC Test Set 0.79 0.88 0.83 0.7 0.65

Table 7 Results

[0136] Advantageously, the estimated change in motor function following first morning dose of therapy, as determined by processor 2003 using passively collected motion data over an assessment period of 6 days, could predict a subject’s response to therapy, in this case L-Dopa, as measured during the in-clinic LDCT. [0137] A LR in the LDCT (LRUPDRS) is calculated as the change in the UPDRS scores (absAupDRs) from the OFF state to the ON state, which was 22 (±11 SD) UPDRS points for the whole cohort (see Table 3 for each clinic). The LRUPDRS is also commonly expressed clinically as %AUPDRS (absAuPDRs/UPDRSoFF)x100), or the percent improvement, which was 47 (±18 SD) for the whole cohort (see Table 3 for each clinic). Fig. 4 shows the relationship between absAupDRs and %AUPDRS. In Fig. 4, the black dots represent all subjects with PD. Black circles with a light grey centre are subjects with disease duration of 5 years or less and black circles with a white centre are subjects with PD having disease duration of 11 or more years. The vertical shaded region shows an “uncertain” zone. To the right of this shaded area are cases where the absAupDRs was considered to be a clinically meaningful increase whereas to the left, absAupDRs was not clinically meaningful. The 3 horizontal lines indicate the three commonly used %AUPDRS, showing that a region of clinical uncertainty also exists.

[0138] The Logistic Regression model designed above was used to obtain from the passively collected motion data, a binary prediction for each epoch in the DT and ET as to whether it was below or above each MFSL at DT. The estimated MFSL for all available epochs were averaged to produce an objective estimate of the motor function score at DT (MFSLDT), and similarly at ET(MFSLET). These were used to produce the model’s estimate of the LR: absAEST (MFSLDT-MFSLET) and %AEST (absAEST/M F S LDT)X100).

[0139] Controls were used to augment the number of cases whose LR could be considered insignificant (Support Class 0 for ROC) and ensure class balance. For controls, DT was set at 10:00 am and ET was at the peak of the BK score smoothed weekly summary from 45 minutes to 90 minutes. UPDRS was set at “0”. Clinical features of Subjects with PD and Controls included in the assessment of the response to the first morning dose of therapy (L-Dopa) in the generation of the model for the LDCT are shown in Table 8 which shows clinical features of sample subjects with PD undertaking the LDCT meeting the criteria for model building. [0140] Criteria Clinic Clinic Clinic Total

Controls

1 2 3 PwP basic criteria 53 49 49 151 174

Plus® Dose time motor function

47 47 47 (DT) 141 132

Plus Effect Time motor function

51 48 49

(ET ) 148 170

Plus DT & ET (for LDCT) 46 47 47 140 46

§ “Plus” indicates that this line includes subjects who met basic criteria PLUS the added criteria in the column

Table 8 [0141 ] Performance of the evaluation score absAEST as a prediction of absAupDRs was estimated in terms of the two metrics of ROC AUC and Precision-Recall Curve as represented in Table 9. When all subjects (column headed “AN” in Table 9 and Fig. 4) in each class were included, the ROC AUC was 0.8 and PR AUC was 0.79.

Excluded Cases

Metric All Already Uncertain^ Already Already "ON"^§ "ON"^§ & "ON"^§ &

Uncertain^ Uncertain^ &Variable^¾

ROC AUC 0.8 0.87 0.83 0.89 0.92

PR AUC 0.79 0.85 0.81 0.87 0.87

Support Class 0 198 193 185 182

(total n) 181

PwP (n) 24 19 11 8 7

Controls (n) 174 174 174 174 174

Support Class 1 (n) 116 95 107 88 50

§ Subjects whose motor function at Dose Time did not represent their worst motor function level f Subjects where there was uncertainty as to whether AUPDRSIII was meaningful c Day to day variability in the amplitude and latency to peak response to levodopa

Table 9

[0142] The performance of the model was potentially degraded by three possibilities:

- that an uncertain region lay between absAUPDRS that was a clinically meaningful (UPDRS Class 1 ) and one which was not clinically meaningful (UPDRS Class 0). - that the MFSLDT did not reflect the subjects’ worse motor function (i.e. they were already “ON” in the sense that their usual level of motor function was not being seen at this time);

- day to day variability in the amplitude MFSLDT and MFSLET, and in latency to peak response, especially from clinical variation in enteric delivery of the drug to absorption sites.

[0143] These issues have the potential to mask or degrade the quality or accuracy of objective clinical evaluation scores calculated according to the invention.

[0144] Means for identifying one or more of the above degrading factors were developed and then excluded from the comparison to assess their effect on the ROC statistics (Table 9 and Fig. 5A and Fig. 5B). Figs 5A and 5B are box and whiskers plots of the distribution of Class 0 and Class 1 in Table 9 (Fig. 5A) and Table 10 (Fig. 5B) plotted according the four corresponding column groups in that Table. “U” indicates “uncertain”, “O” indicates “Already ON” and “V” indicates “Variable”.

_ Excluded Cases _

Metric ^A11 Already"ON"^§ Uncertain^⁵ Already Already

"ON"^§ & "ON"^§ &

Uncertain^ Uncertain^ &Variable^x

ROC AUC 0.74 0.78 0.76 0.8 0.83 PR AUC 0.71 0.74 0.72 0.75 0.74

Support Class 0 (total) 205 198 191 185 181 Support Class 0 (PwP) 31 24 17 11 7 Support Class 0 174 174 174 174 174 (controls)

Support Class 1 (PwP) 109 90 101 85 48

Table 10

[0145] The small “box and whiskers” plot, between Class 0 and Class 1 in Fig. 5A in the Exclude U group shows the distribution of absAEST of the uncertain cases. The boxes are the median and quartiles with the “whiskers” showing the 90th and 10th percentile.

[0146] In some embodiments, the processor automatically identifies in the data unreliable data segments including segments in which there is one or more of: a) motion data indicating the subject is already responsive to therapy; b) motion data indicating excess variability in responsiveness to therapy; and c) motion data indicating inactivity during periods of expected wakefulness.

[0147] In some embodiments, the processor transmits an operational signal to a report generating processor 2007, which generates a report including a notification of unreliable reporting when the processor identifies in the motion data unreliable data segments as identified above and discussed below.

[0148] Motion data indicating the subject is already responsive to therapy may be identified automatically in cases where the subject’s motor function at DT did not represent their worst motor function level. If MFSLDT was not the most severe level of motor function present in a subject’s motion data, it suggests that they may be “already ON”: that is, they may have already received treatment. This is relevant because the aim clinically, when performing an LDCT, is for motor function at the time of the dose during the LDCT to represent the subject’s worst motor function (highest score) and to this end, anti-PD medications are not taken after the prior evening in preparation for the LDCT. Motor function is usually assessed several hours after awakening so any “so-called” sleep benefit has dispersed.

[0149] When performing an in-clinic LDCT, the “OFF” state is considered to represent the subject’s worst motor function. Similarly, the motor function at DT during assessment using passively collected motion data according to embodiments of the invention is assumed as being the time of maximum motor dysfunction. In training the computational model, subjects were excluded if they were known to routinely take medications prior to the first reminder, however on inspection of the received motion data and calculated measures of kinetic state of some cases, it was apparent that higher levels of BK occurred later in the day and so it was possible that medications were consumed prior to the first reminder without this being recorded. This possibility was examined further. First, subjects whose MFSLDT was in the treated range were found: That is MFSLDT was less than MFSL 3 (Table 1 ). Next was to establish whether motor function deteriorated significantly later in the day by estimating the subjects MFSL (as per the model: average plus one standard deviation) from 46 minutes after the first dose until 18:00 hours. If this was more than 1 MFSL greater than the motor function level at DT, they were flagged as “already ON”: 19% of subjects with PD met this criterion and its effect was analysed using the ROC statistic (Table 9). Exclusion of these “already ON” subjects improved the ROC AUC to 0.87 and PR AUC was 0.85.) When cases that were “already ON” and those that were in the uncertain absAupDRs were both excluded, the ROC AUC and PR AUC further improved to 0.89 and 0.87 respectively.

[0150] Sixty-eight percent of all subjects with PD in this study were on D2 agonists and the percent of subjects with PD taking D2 agonists who were flagged as “already ON” was also 32%. Furthermore, the average L-Dopa equivalent dose (LED) was 177 in those who were already ON, compared to 170 in the rest of the cohort. D2 agonist contributed less than 20% of the total LED dose in most subjects with PD. Fig. 6A shows the range of total LED and the dose of levodopa (L-Dopa). Fig. 6B shows the LED from the first dose of L-Dopa (1st dose) and from D2 agonists over the course of the day. Fig. 6C shows the percentage of the LED contributed to by D2 agonists. Neither of absAEST and absAUPDRS were smaller when the size or the relative contribution of the D2 agonist dose was large. This suggested that D2 agonists did not contribute to whether a LR could be assessed using the passively collected motion data according to the present invention.

[0151 ] In terms of applying the model, estimates of MFSLDT or MFSLET could be affected in some subjects by epochs excluded because of inactivity or exercise. This affected the sample size and increased the variability in estimates of motor function at DT and ET. As well the size and time to peak response to a dose of L-Dopa (such as in an LDCT), variability in gut motility and gastric emptying can cause erratic delivery of L-Dopa to transport sites in the gut, which would result in day to day variability even if measured by UPDRS. There is also variation in the UPDRS itself with variation in assessment by an individual clinician and between clinician. In the present case, excess variability in amplitude of LR was assessed by measuring the standard deviation in the MFSLDT and MFSLET of all epochs in DT and ET of all available days. If both were greater than 1 MFSL (Table 1 ) then it was deemed that there is significant variability in the motor function at both times: 23% of subjects with PD had excess variability in amplitude. Variability in latency from dose to peak was also assessed: cases with excess variability were cases where the standard deviation of the estimation of MFSL was greater than 1 for both DT and ET. Thirty five percent of subjects with PD were flagged as having increased variability in latency to peak. Exclusion of cases with variability measured in this way, in addition to cases that were “already ON” and were in the uncertain absAUPDRS resulted in ROC AUC of 0.92 and PR AUC of 0.87 (Table 9).

[0152] As discussed above, it is common practice clinically to use the %AUPDRS as an outcome of the LDCT. While the performance of the model in predicting the percent improvement in motor function was not as effective as predicting the absolute change in motor function (Table 10 and Fig. 5), it nevertheless resulted in ROC_AUC and PR AUC of 0.83 and 0.74 (respectively) when cases that were “already ON”, in the uncertain zone and with excess variability were excluded.

[0153] For 95/142 subjects with PD, the duration of disease was available, and this was plotted against the LR measured by absAEST and absAupDRs (Figs. 7A to 7D). Figs 7A and 7B show the change in LR according to duration of disease (in years).

Fig. 7A shows the absAEST and Fig. 7B shows the absAupDRs. (note that 1 unit on the Y axis of Fig. 7A approximates 12 UPDRS units shown on the Y axis of Fig. 7B). Fig. 7C shows absAEST before and after deep brain stimulation (DBS). Fig. 7D shows the same data, with the difference in absAEST before and after DBS (X axis) plotted against the absAEST before DBS. In Figs 7A, 7B and 7C, the boxes are the median and quartiles with the “whiskers” showing the 90th and 10th percentile.

[0154] In this study the absAEST increases with disease duration, which is similar to reports of others using absAupDRs. In this study the absAupDRs showed only a weak trend to increase with disease and presumably the variation between different clinical reporters of the UPDRS may have obscured this trend. The average time to peak response was 45 minutes (±19 SD) and this compares to 51 minutes (±25 SD) reported elsewhere. This also provides support that the inventive method and system processing passively collected motion data is indeed measuring an LR similar to that measured by the LDCT. [0155] The absAupDRs improves following deep brain stimulation (DBS) due to improved motor function in the OFF” state. Thus, we should expect that absAEST will improve following DBS. Evaluations performed using the inventive system and method before and 6 months after DBS were available for calculation of an evaluation score, absAPKG (Fig. 7C and 7D). As expected, there was marked reduction in absAPKG and the size of the reduction was predicted by the pre-surgery LR as measured by the absAPKG (Fig. 7 D). The data regarding the size of the LR in relationship to age and the response to DBS are presented here to show that the findings of the assessment of the LR determined using passively collected motion data according to the inventive method and system, reproduce the findings found by UPDRS.

[0156] The Example provided above demonstrates that an objective measure of L-Dopa responsiveness can be provided according to present embodiments, with a similar classification performance to the LDCT. The machine learning model exemplified herein defines 6 levels of motor function severity before taking L-Dopa and after the L-Dopa has taken effect in relation to the UPDRS scoring protocol. The models were designed and validated against unseen data to ensure their generalisability. Employing these models, the passively collected motion data could be used to predict the absAupDRs with ROC AUC of 0.92 indicating that the inventive method and system can be used to accurately replicate the LDCT in an ambulatory fashion, negating requirement for a hospital visit and the costs and complications that go with it.

[0157] A substantial improvement (ROC AUC =0.92) could be obtained by excluding those in the uncertain absAupDRs range (uncertain zone), those who were “already ON” with deterioration observed later in the day and those with excess variability in response.

[0158] It is to be understood, however, that the computational model exemplified herein, and the process deployed to determine it, is only one example of a computational model that can estimate motor function behaviours, such as response to therapy, using continuously passively collected motion data. Logistic regression is only one example of how features in the measures of kinetic state may be combined to obtain Motor Function Severity Level based on the training data set. Already ON

[0159] When an LDCT is performed, there is an expectation that “OFF” motor function represents the subject’s worst motor function. To achieve this, anti-PD medications are not taken after the prior evening and the test is usually performed several hours after awakening so any “so-called” sleep benefit has dispersed. Similarly, the measures of kinetic state produced by processor 2003 representing the level of motor function at the DT should represent the worst motor function experienced over the course of the day. In many cases, medications were taken immediately on awakening so, if “sleep benefit” was a real entity, it would interfere with the model’s estimations. Subjects were excluded from the sample data if the clinical notes indicated routine consumption of medications (including apomorphine) prior to the first reminder because this would improve the motor function at DT. Flowever, it is possible that some subjects consumed medications prior to their first reminder without this being recorded.

[0160] Whatever the mechanism, subjects whose BK had been partially alleviated at the time of the first dosage reminder could be revealed as having adequately treated BK at the time of the first dose but having worse levels of motor function later in the day. The statistical method used in this study identified subjects whose level of motor function was already in the treated range and also experienced higher levels of BK (as identified by the processor 2003) later in the day. These subjects were flagged as “already ON” at DT (19% of subjects with PD met this criterion and excluding these subjects improved the ROC AUC and PR AUC (to 0.87 and 0.85 respectively)).

[0161] Thus, in a preferred embodiment, the inventive system and method identifies automatically unreliable data segments by, identifying in the measures of kinetic state, epochs where the subject’s motor function is at a level indicating it is already in a treated range followed by epochs later in the day which are indicative of worse motor function (as may be represented by higher BK scores e.g. >26). In a preferred embodiment, the inventive system and method flags the existence of such data as an indication of early morning alleviation of symptoms (e.g. BK) though whatever mechanism (e.g. medication or sleep benefit) which may in turn reduce the performance of the processor 2003 calculating an evaluation score, such as a LR. This may also provide a marker to question the compliance of the subject in relation to medication consumption in the morning.

Increased variability

[0162] Estimates of MFSLDT or MFSLET could be affected in some subjects by epochs excluded because of inactivity or exercise. In the above example, this affected the sample size and increased the variability in estimates of motor function at DT and ET. Also, the size and time to peak response to a dose of L-Dopa (such as in an LDCT) will be affected by variability in gut motility and gastric emptying causing erratic delivery of levodopa to transport sites in the gut, which would result in day to day variability even if measured by UPDRS III. There is also variation in the UPDRS III itself with variation in assessment by an individual performer and between performers. Excess variability in amplitude of response was assessed by measuring the standard deviation in the MFSLDT and MFSLET of all epochs in DT and ET of all available days in the assessment period. If both were greater than 1 MFSL (Table 1 ) then it was deemed that there is significant variability in the motor function at both times: 23% of subjects with PD had excess variability in amplitude. Variability in latency from dose to peak was also assessed: cases with excess variability were cases where the standard deviation of the estimation of MFSL was greater than 1 for both DT and ET. Thirty five percent of subjects with PD were flagged as having increased variability in latency to peak. Exclusion of cases with variability measured in this way, in addition to cases that were “already ON” and were in the uncertain absAupDRs resulted in ROC AUC of 0.92 and PR AUC of 0.87 (Table 9).

[0163] Including an “uncertain zone” improved the ROC statistics. The inventors are of the view that this is due to clinical uncertainty about what constitutes a positive response to L-Dopa. Clinical convention usually recognises a significant LR as %AUPDRS of 30% but range from e.g. 33% to 20%. The literature is not forthcoming as to why a percentage improvement is preferred. In clinical practice, LDCT is most commonly performed in subjects with significant levels of BK in the untreated state because the questions for performing the LDCT relate to suitability for advanced therapy or the diagnosis of responsive PD. While both %AUPDRS and absAupDRs were compared with the LR measured by the UPDRS, we also found the absAupDRs estimated from the passively collected motion data to the best predictor. [0164] The justification for the size of the uncertain zone (4 UPDRS points) and its location on the UPDRS scale (11-14) was empirical. The current model suits the clinical question as to whether the LR is suitable for deploying advanced therapies such as deep brain stimulation. In these subjects, fluctuations are well developed and LR is usually large.

[0165] In some embodiments, processor 2003 transmits an operational signal to a report generating processor (report generator 2007 in Fig. 2) that generates automatically a report containing evaluation scores including one or more quantitative or qualitative measures of the subject’s PD state. In some embodiments, the report generating processor 2007 includes in the report a recommendation for modifying a prescribed therapy, where the evaluation scores indicate that the subject is undertreated. Modifications to prescribed therapy recommended by the processor 2003 may include one or more of a) increasing size of dose and b) increasing frequency of dose. In some embodiments, modifications to prescribed therapy include a prescription for advanced therapy such as DBS. In other embodiments, modifications to prescribed therapies maybe used in a feedback control or closed loop system to automatically select or titrate dosages delivered by a dispensing device 2008 such as a pill dispenser or levodopa pump.

[0166] In some embodiments, system 2000 receives therapy inputs (e.g. via input module 2002) from a dispensing device 2008 that is indicative of the subject having received a therapy. The dispensing device may be an electronic pill dispenser, levodopa pump or DBS system that is communicatively coupled with processor 2003, typically via communication network 2006 to enable it to receive automatically time- marked therapy data indicative of therapy having been received by the subject during the assessment period.

[0167] Where the terms “comprise”, “comprises”, “comprised” or “comprising” are used in this specification (including the provisional claims) they are to be interpreted as specifying the presence of the stated features, integers, steps or components, but not precluding the presence of one or more other features, integers, steps or components or group thereof. [0168] It is to be understood that various modifications, additions and/or alterations may be made to the parts previously described without departing from the ambit of the present invention as defined in the provisional claims appended hereto.

[0169] Future patent applications may be filed on the basis of the present application. It is to be understood that the following claims are provided by way of example only, and are not intended to limit the scope of what may be claimed in any such future application. Features may be added to or omitted from the claims at a later date so as to further define or re-define the invention or inventions.

Claims

The claims defining the invention are as follows

1. A machine automated method for evaluating a movement disorder disease state in subject, using passively collected motion data, the method comprising: a) receiving at a processor time-marked motion data from a device worn by the subject over an assessment period; b) the processor processing the received motion data to produce one or more measures of kinetic state of the subject that are indicative of movements of the subject that are attributable to symptoms of the movement disorder; c) the processor calculating one or more evaluation scores from the one or more calculated measures of kinetic state and transmitting an operational signal to a user interface on which the one or more evaluation scores are presented to a user; wherein the one or more evaluation scores represent the evaluated state of the movement disorder for the subject during the assessment period.

2. The machine automated method according to claim 1 , wherein the motion data is passively collected in that it comprises signals recorded from the subject while ordinarily ambulatory and in the absence of clinician-directed motor tasks performed by the subject.

3. The automated method according to claim 1 or claim 2, wherein the processor calculates one or more evaluation scores by applying a computational model to the received motion data.

4. The automated method according to claim 3, wherein the movement disorder disease is PD and the computational model relates sample motion data obtained from a sample of people with PD undertaking a levodopa challenge test (LDCT) to one or more measures of kinetic state.

5. The automated method according to claim 4, wherein the computational model is built by: a) mapping a clinical scoring range into a number of levels representing motor function severity; b) mapping the measures of kinetic state to levels representing motor function severity; and c) determining a function that relates the measures of kinetic state to the clinical scoring range.

6. The automated method according to claim 5, wherein mapping the measures of kinetic state include extracting a set of candidate features from the sample motion data and applying the classifier function to the candidate features.

7. The automated method according to claim 6, wherein the classifier function is a binary classifier and/or the function is determined using a logistic regression model.

8. The automated method according to any one of claims 3 to 7 wherein features used in the computational model are selected from a group of one more measures of kinetic state listed in Table 4.

9. The machine automated method according to any one of the preceding claims, further comprising: the processor processing the one or more measures of kinetic state to identify automatically a plurality of distinctive occasions during the assessment period.

10. The machine automated method according to claim 9, wherein a distinctive occasion is selected from a group including: a) the subject receiving a therapy; b) the subject exhibiting a therapeutic motor response to a received therapy; c) after the subject has woken from night time sleep; d) after the subject has woken from night time sleep and before first therapy of the day; e) after the subject has woken from night time sleep and after first therapy of the day; f) after a period of time exceeding a duration for which therapy is known to be effective in treating disordered movement.

11. The machine automated method according to claim 9 or claim 10, wherein a distinctive occasion occurs at least daily and in some cases, several times per day.

12. The machine automated method according to any one of the preceding claims, wherein the evaluation score indicates one or more of: a) whether the subject responds to therapy; b) time to maximum responsiveness to therapy; c) duration of response to received therapy; d) magnitude of response to received therapy; e) variability of response to received therapy over the assessment period; f) an extent to which the subject is treated/undertreated by received therapy; g) an extent to which the subject is untreatable by therapy; h) an extent to which the subject is a candidate for advanced therapy; and i) severity of motor function symptoms that contribute to the movement disorder.

13. The machine automated method according to any one of the preceding claims, wherein the processor calculates automatically changes in one or more of the evaluation scores since one or more prior assessment periods.

14. The machine automated method according to any one of the preceding claims, wherein the evaluation scores are quantitative.

15. The machine automated method according to any one of the preceding claims, wherein the processor automatically relates an evaluation score calculated by the processor to a score on a clinically accepted scale.

16. The machine automated method according to any one of the preceding claims, wherein the evaluation score is calculated at intervals during the assessment period.

17. The machine automated method according to any one of the preceding claims, wherein the processor calculates automatically a duration or proportion of time during the assessment period that the evaluation score exceeds a predetermined threshold.

18. The machine automated method according to claim 17, wherein the predetermined threshold refers to evaluation scores obtained from one or more healthy subjects without the movement disorder.

19. The machine automated method according to any one of the preceding claims, wherein the movement disorder disease is Parkinson’s disease (PD) and the processor calculates automatically a Unified Parkinson’s Disease Rating Score (UPDRS) equivalent score for a subject with PD in the absence of clinician- directed motor tasks required to be performed by the subject according to a clinical UPDRS protocol.

20. The machine automated method according to claim 19, further comprising the processor determining automatically if the subject is a fluctuator, wherein the evaluation score indicates magnitude of response to received therapy, and wherein the processor determines the subject to be a fluctuator when the magnitude of response meets a significance threshold.

21. The machine automated method according to claim 20, wherein the significance threshold is selected from the group including: a) a 30% improvement in a measure of the kinetic state; b) an improvement of 14 points on the UPDRS scale; and c) an improvement of at least 1 and preferably 1.15 Severity Level points.

22. The machine automated method according to claim 20 or claim 21, further comprising the processor determining automatically if the fluctuator is a controlled fluctuator, by comparing the measure of kinetic state or evaluation score following a received therapy with a target value, wherein the processor determines the fluctuator to be a controlled fluctuator when the measure of kinetic state is less than the target value.

23. The machine automated method according to claim 22, further comprising the processor determining automatically if a controlled fluctuator has wearing off, by calculating a reduction in response to received therapy over a set period, wherein the processor determines the controlled fluctuator to have wearing off when, over the set period, the reduction is one or both of: a) 1 or more Severity Level points; and b) 14 UPDRSIII points; and optionally, wherein the set period is 90 to 210 minutes and preferably 90 to

150 minutes and more preferably approximately 120 minutes following a received therapy.

24. The machine automated method according to any one of the preceding claims, further comprising the processor determining automatically a PTB value calculated as proportion of time during the assessment period that the measures of kinetic state, or an evaluation score calculated from the measures of kinetic state, indicate that the subject exhibits movement disorder symptoms corresponding to bradykinesia.

25. The machine automated method according to claim 24, further comprising the processor automatically performing one or more of: a) identifying a subject having a calculated PTB value less than 30% as having controlled motor symptoms; and b) monitoring a change in PTB value over time for the subject and identifying a subject having PTB over 30% and increasing as requiring modified therapy.

26. The machine automated method according to any one of the preceding claims, further comprising the processor determining automatically a PTD value calculated as proportion of time during the assessment period that the measures of kinetic state, or an evaluation score calculated from the measures of kinetic state, indicate that the subject exhibits movement disorder symptoms corresponding dyskinesia.

27. The machine automated method according to any one of the preceding claims, wherein the processor calculates an evaluation score which characterises the subject’s response to a first therapy after waking on one or more days during the assessment period, and wherein the processor automatically translates the evaluation score to a clinical score on a clinically accepted scale.

28. The machine automated method according to any one of the preceding claims, wherein the processor pre-processes the received motion data to remove automatically: a) data corresponding to periods of night time sleep by the subject; b) data corresponding to periods the device was not worn by the subject c) data corresponding to periods of inactivity by the subject.

29. The machine automated method according to any one of the preceding claims, wherein the processor automatically identifies in the data unreliable data segments including segments in which there is one or more of: a) motion data indicating the subject is already responsive to therapy at a distinctive occasion corresponding to the subject receiving a therapy; b) motion data indicating excess variability in responsiveness to therapy; and c) motion data indicating inactivity during periods of expected wakefulness.

30. The automated method according to claim 29, wherein the processor transmits an operational signal to a report generating processor, and the report generating processor generates a report including a notification of unreliable reporting when the processor identifies in the motion data unreliable data segments.

31 .The machine automated method according to any one of the preceding claims, wherein the processor transmits an operational signal to a report generating processor that generates automatically a report containing one or both of quantitative and qualitative measures of the subject’s movement disorder disease state.

32. The machine automated method according to claim 31, wherein the report generating processor includes in the report a recommendation for modifying a prescribed therapy.

33. The automated method according to any one of the preceding claims, wherein the device is a wrist worn device.

34. The automated method according to any one of the preceding claims, wherein the motion data is accelerometer data from an accelerometer in the device.

35. The automated method according to any one of the preceding claims, wherein the device is configured to receive a therapy input, indicative of the subject having received a dose of therapy.

36. The automated method according to claim 35, wherein the processor receives therapy inputs from a therapy dispensing device that is indicative of the subject having received a therapy.

37. The automated method according to any one of the preceding claims, further comprising receiving at the processor time-marked therapy data indicative of therapy received by the subject during the assessment period.

38. A system for evaluating a movement disorder disease state in subject, using passively collected motion data, the system comprising: a processor, a user interface and a memory module containing code corresponding to instructions causing the processor to: a) receive time-marked motion data from a device worn by the subject over an assessment period; b) process the received motion data to produce one or more measures of kinetic state of the subject that are indicative of movements of the subject that are attributable to symptoms of the movement disorder; c) and calculate one or more evaluation scores from the one or more calculated measures of kinetic state; and d) transmit an operational signal to the user interface on which the one or more evaluation scores are presented to a user; wherein the one or more evaluation scores represent the evaluated state of the movement disorder for the subject during the assessment period.

39. The system according to claim 38, wherein the processor calculates one or more evaluation scores by applying a computational model to the received motion data.

40. The system according to claim 39, wherein the movement disorder disease is PD and the computational model relates sample motion data obtained from a sample of people with PD undertaking a levodopa challenge test (LDCT) to one or more measures of kinetic state.

41. The system according to claim 39 or claim 40, wherein the computational model is built by: a) mapping a clinical scoring range into a number of levels representing motor function severity; b) mapping the measures of kinetic state to levels representing motor function severity; and c) determining a function that relates the measures of kinetic state to the clinical scoring range.

42. The system according to claim 41 , wherein mapping the measures of kinetic state include extracting a set of candidate features from the sample motion data and applying the classifier function to the candidate features.

43. The system according to claim 43, wherein the classifier function is a binary classifier and/or the function is determined using a logistic regression model.

44. The system according to any one of claims 39 to 43 wherein features used in the computational model are selected from a group of one more measures of kinetic state listed in Table 4.

45. The system according to any one of claims 38 to 44, wherein the evaluation score indicates one or more of: a) whether the subject responds to therapy; b) time to maximum responsiveness to therapy; c) duration of response to received therapy; d) magnitude of response to received therapy; e) variability of response to received therapy over the assessment period; f) an extent to which the subject is treated/undertreated by received therapy; g) an extent to which the subject is untreatable by therapy; h) an extent to which the subject is a candidate for advanced therapy; and i) severity of motor function symptoms that contribute to the movement disorder.

46. The system according to any one of claims 38 to 45, wherein the movement disorder disease is Parkinson’s disease (PD) and the processor calculates automatically a Unified Parkinson’s Disease Rating Score (UPDRS) equivalent score for a subject with PD in the absence of clinician-directed motor tasks required to be performed by the subject according to a clinical UPDRS protocol.

47. The system according to claim 46, further comprising the processor determining automatically if the subject is a fluctuator, wherein the evaluation score indicates magnitude of response to received therapy, and wherein the processor determines the subject to be a fluctuator when the magnitude of response meets a significance threshold.

48. The system according to claim 47, wherein the significance threshold is selected from the group including: a) a 30% improvement in a measure of the kinetic state; b) an improvement of 14 points on the UPDRS scale; and c) an improvement of at least 1 and preferably 1.15 Severity Level points.

49. The system according to claim 47 or claim 48, further comprising the processor determining automatically if the fluctuator is a controlled fluctuator, by comparing the measure of kinetic state or evaluation score following a received therapy with a target value, wherein the processor determines the fluctuator to be a controlled fluctuator when the measure of kinetic state is less than the target value.

50. The system according to claim 49, further comprising the processor determining automatically if a controlled fluctuator has wearing off, by calculating a reduction in response to received therapy over a set period, wherein the processor determines the controlled fluctuator to have wearing off when, over the set period, the reduction is one or both of: a) 1 or more Severity Level points; and b) 14 UPDRSIII points; and optionally, wherein the set period is 90 to 210 minutes and preferably 90 to

51.The system according to any one claims 38 to 50, further comprising the processor determining automatically a PTB value calculated as proportion of time during the assessment period that the measures of kinetic state, or an evaluation score calculated from the measures of kinetic state, indicate that the subject exhibits movement disorder symptoms corresponding to bradykinesia.

52. The system according to claim 51, further comprising the processor automatically performing one or more of: a) identifying a subject having a calculated PTB value less than 30% as having controlled motor symptoms; and b) monitoring a change in PTB value over time for the subject and identifying a subject having PTB over 30% and increasing as requiring modified therapy.

53. The system according to any one of claims 38 to 52, further comprising the processor determining automatically a PTD value calculated as proportion of time during the assessment period that the measures of kinetic state, or an evaluation score calculated from the measures of kinetic state, indicate that the subject exhibits movement disorder symptoms corresponding dyskinesia.

54. The system according to any one of claims 38 to 53, wherein the processor calculates an evaluation score which characterises the subject’s response to a first therapy after waking on one or more days during the assessment period, and wherein the processor automatically translates the evaluation score to a clinical score on a clinically accepted scale.

55. The system according to any one of claims 38 to 54, wherein the processor automatically identifies in the data unreliable data segments, and optionally, wherein the processor transmits an operational signal to a report generating processor, and the report generating processor generates a report including a notification of unreliable reporting when the processor identifies in the motion data unreliable data segments.

56. The system according to any one of claims 38 to 55, further comprising a wrist- wearable device configured to collect the time-marked motion data.

57. The system according to any one of claims 38 to 56, wherein the processor receives time-marked therapy data indicative of therapy received by the subject during the assessment period.