# US 2014/0155784 A1 · Signal processing method and apparatus > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2014/0155784 A1 - Document ID: US20140155784A1 - Status: Published application - Kind: A1 - Application number: 13/980,320 - Filed: 2012-01-18 - Published: 2014-06-05 - Earliest claimed: 2011-01-18 - Assignees: Clearsky Medical Diagnostics Limited - Inventors: Stephen Smith - Web page: https://patentgrep.com/patent/US20140155784A1 ## CPC - G06F 18/2111: by using evolutionary computational techniques, e.g. genetic algorithms - A61B 5/1101: {Detecting tremor} - A61B 5/1125: {Grasping motions of hands} - G06F 2218/12: Classification; Matching - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D) ## Abstract The invention provides signal processing algorithms and apparatus for detecting bradykinesia, tremor, or other symptoms of neurological dysfunction in subjects, using three-dimensional sensors to tract finger and hand position. The invention provides Cartesian Genetic Programming networks and particular function blocks for such networks to enable identification of subjects exhibiting such symptoms. ## Claims ### Claim 1 A method of generating a signal processing algorithm for processing signals representing finger and thumb positions for identifying subjects exhibiting bradykinesia or tremor at a required discrimination confidence, said method comprising the steps of: (1) providing a plurality of patient position data sets, each set representing a time series of three-dimensional position data of finger and thumb during tapping tests from each of a plurality of subjects exhibiting bradykinesia or tremor; (2) providing a plurality of control position data sets, each set representing time series of three-dimensional position data of finger and thumb during tapping tests from each of a plurality of control subjects; (3) processing said position data sets to produce patient acceleration data sets and control acceleration data sets representing time series of acceleration of the finger relative to the thumb for each corresponding tapping test; (4) providing a set of function blocks for construction of a Cartesian Genetic Programming (CGP) network; (5) providing a fitness function giving a measure of the success of said CGP network in discriminating between said control acceleration data set and said patient acceleration data set; (6) configuring an initial CGP network using said function blocks, said network accepting a plurality of input data comprising adjacent values in an acceleration data set and producing a single output value putatively indicative of the presence or absence of bradykinesia or tremor; (7) evolving said initial CGP network by sequential evolutionary mutation and selection towards increased fitness until said desired discrimination confidence is achieved. ### Claim 2 (depends on 1) A method according to claim 1 wherein the processing of position data sets in step (3) comprises intermediate calculation of patient and control velocity data sets representing a time series of velocity data of the finger relative to the thumb for each corresponding tapping test, said velocity data sets being smoothed using a moving average filter of window size 2. ### Claim 3 (depends on 1) A method according to claim 1 wherein the processing of position data sets in step (3) comprises intermediate calculation of patient and control velocity data sets representing a time series of velocity data of the finger relative to the thumb for each corresponding tapping test, said velocity data sets being clipped to within one standard deviation of the mean velocity. ### Claim 4 A method of generating a signal processing algorithm for processing signals representing finger and thumb positions for identifying subjects exhibiting bradykinesia, tremor, or other symptoms of neurological dysfunction at a required discrimination confidence, said method comprising the steps of: (1) providing a plurality of patient position data sets, each set representing a time series of three-dimensional position data of a subject's fingers during motor function tests from each of a plurality of subjects exhibiting bradykinesia, tremor, or symptoms of other neurological dysfunction; (2) providing a plurality of control position data sets, each set representing time series of three-dimensional position data of a subject's fingers during motor function tests from each of a plurality of control subjects; (3) processing said position data sets to produce patient acceleration data sets and control acceleration data sets representing time series of acceleration of the fingers for each corresponding motor function test; (4) providing a set of function blocks for construction of a Cartesian Genetic Programming (CGP) network; (5) providing a fitness function giving a measure of the success of said CGP network in discriminating between said control acceleration data set and said patient acceleration data set; (6) configuring an initial CGP network using said function blocks, said network accepting a plurality of input data comprising adjacent values in an acceleration data set and producing a single output value putatively indicative of the presence or absence of bradykinesia, tremor, or other neurological dysfunction; (7) evolving said initial CGP network by sequential evolutionary mutation and selection towards increased fitness until said desired discrimination confidence is achieved. ### Claim 5 (depends on 4) A method according to claim 4 wherein said motor function test comprises a constructional task such as the tracing of a geometric figure. ### Claim 6 (depends on 4) A method according to claim 4 wherein said motor function test comprises a prehension task such as a “reach and grasp” task. ### Claim 7 (depends on 1) A method according to claim 1, wherein said function blocks comprise functions having two inputs (X and Y) and one output (OP), said output having a limited range of values, and wherein each of said function blocks is configured to have one of the following characteristics: (1) said output is a relatively large number if the absolute difference between X and Y is greater than a pre-set value; (2) said output is a number less than the value of the output in (1), but preferably greater than the mid-point of the range, if the absolute difference between X and Y is greater than a second pre-set value; (3) said output is a relatively small number if the absolute difference between X and Y is less than a third pre-set value; (4) said output is a number greater than the value of the output in (3), but less than the output of (2) if the absolute difference between X and Y is less than a fourth pre-set value, said fourth pre-set value being larger than the third pre-set value, and less than the second pre-set value; (5) said output is the average of X and Y. ### Claim 8 (depends on 7) A method according to claim 7 wherein the first to fourth pre-set values and the centre of the range of values i.e. approximately equidistant from a neighbouring value and ideally are integers for ease of handling. ### Claim 9 (depends on 1) A method according to claim 1 wherein said fitness function comprises the area under the receiver operating characteristics curve. ### Claim 10 (depends on 1) Apparatus for detecting bradykinesia in a subject comprising: a first three-dimensional position sensor adapted to be securable to a thumb of a subject; a second three-dimensional position sensor adapted to be securable to an opposing finger of a subject; a signal processor configured to process a time series of position data from said sensors using an algorithm generated by a method according of claim 1 or any preceding claim dependent on claim 1. ### Claim 11 (depends on 4) Apparatus for detecting bradykinesia, tremor, or other neurological dysfunction in a subject comprising: a three-dimensional position sensor adapted to be securable to a finger of a subject; a signal processor configured to process a time series of position data from said sensor using an algorithm generated by a method according of claim 4. ### Claim 12 (depends on 11) Apparatus according to claim 11 wherein said three-dimensional position sensor is incorporated in a data glove. ### Claim 13 (depends on 12) Apparatus according to claim 12 wherein said data glove comprises multiple such sensors, thereby enabling tracking movement of fingers and thumbs. ### Claim 14 (depends on 13) Apparatus according to claim 13, further comprising a three-dimensional position sensor adapted to be securable to the wrist of a subject.