# US 10,368,792 B2 · Method for detecting deception and predicting interviewer accuracy in investigative interviewing using interviewer, interviewee and dyadic physiological and behavioral measurements > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 10,368,792 B2 - Document ID: US10368792B2 - Status: Granted - Kind: B2 - Application number: 14/728,527 - Filed: 2015-06-02 - Published: 2016-12-08 - Granted: 2019-08-06 - Earliest claimed: 2015-06-02 - Assignees: The Charles Stark Draper Laboratory, Inc. - Applicants: The Charles Stark Draper Laboratory, Inc. - Inventors: Nirmal Keshava; Andrea K. Webb; Laura J. Mariano; Philip D. Parks; Joshua C. Poore - Web page: https://patentgrep.com/patent/US10368792B2 ## CPC - A61B 5/164: {Lie detection} - A61B 5/0077: {Devices for viewing the surface of the body, e.g. camera, magnifying lens} - A61B 5/0205: Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition - A61B 5/0533: Measuring galvanic skin response - A61B 5/1128: {using image analysis (A61B5/1127 takes precedence)} - A61B 5/332: Portable devices specially adapted therefor - A61B 5/0816: {Measuring devices for examining respiratory frequency (measuring frequency of electric signals G01R23/00)} - A61B 5/11: Measuring movement of the entire body or parts thereof, e.g. head or hand tremor or mobility of a limb {(A61B5/1038 takes precedence; motion detection to correct for motion artifacts in physiological signals A61B5/721)} - A61B 5/14551: {for measuring blood gases} - A61B 5/14552: {Details of sensors specially adapted therefor} - A61B 5/6823: {Trunk, e.g., chest, back, abdomen, hip} - A61B 5/6824: {Arm or wrist} - A61B 5/6828: {Leg} - A61B 5/6829: {Foot or ankle} - A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D) ## Abstract Existing approaches for deception detection are primarily based on polygraph systems that measure specific channels of physiology in highly structured interviews and that are interpreted by trained polygraph examiners. Existing approaches for predicting interviewer accuracy involve interviewers' own estimates of their performances which inevitably are biased. The methods and systems described herein provides objective, quantitative and automated metrics to detect deception and predict interviewer accuracy. Physiological information of the interviewer during the interview is recorded by at least a first sensor. The physiological information includes a time series of physiological data. An interview assessment is calculated by a computer. By processing the recorded physiological information, the interview assessment indicates at least one of whether a statement made by the interviewee is likely to be deceitful and whether the interviewer is likely to be accurate in estimating truthfulness of the interviewee. The interview assessment is output by the computer. ## Claims ### Claim 1 A method of detecting deception by an interviewee and predicting accuracy by an interviewer in an interview conducted by the interviewer, comprising: recording, by at least a first sensor, a first time series of physiological data comprising electrocardiogram data of the interviewer during the interview; recording, by at least a second sensor, a second time series of physiological data comprising electrocardiogram data of the interviewee during the interview; generating, by a computer, a time lagged time series comprising the second time series of physiological data shifted a predetermined length of time; determining, by the computer, a dyadic feature based on the first time series of physiological data comprising electrocardiogram data of the interviewer and the time lagged time series of the interviewee, wherein the dyadic feature indicates at least one of a time-based or frequency-based relationship between the first time series of physiological data comprising electrocardiogram data of the interviewer and the time lagged time series of the interviewee; determining, by the computer and based on the dyadic feature, a level of coherence between the first time series of physiological data and the second time series of physiological data; and generating, by the computer, an accuracy score for the interviewer based on the level of coherence between the first time series of physiological data and the second time series of physiological data being above a predetermined threshold. ### Claim 2 (depends on 1) The method according to claim 1, further comprising recording at least one of skin conductance data and pulse oximetry data of the interviewer or the interviewee. ### Claim 3 (depends on 1) The method according to claim 1, further comprising calculating a level of entropy, a heart rate, and time intervals between consecutive normal sinus beats in the first time series and the second time series. ### Claim 4 (depends on 1) The method according to claim 1, further comprising: applying an auto-regressive integrated moving average analysis to the first time series of physiological data of the interviewer and the second time series of physiological data of the interviewee. ### Claim 5 (depends on 1) The method according to claim 1, further comprising: recording a third time series comprising movement data of the interviewer during the interview; recording a fourth time series comprising movement data of the interviewee during the interview; determining, by the computer, the level of coherence between the third time series comprising the movement data of the interviewer and the fourth time series comprising the movement data of the interviewee. ### Claim 6 (depends on 5) The method according to claim 5, further comprising: recording video of the interview; and processing the recorded video to generate the third time series comprising the movement data of the interviewer and the fourth time series comprising the movement data of the interviewee. ### Claim 7 (depends on 5) The method according to claim 5, wherein the third time series comprising the movement data of the interviewer and the fourth time series comprising the movement data of the interviewee indicate times at which the interviewer and interviewee, respectively, moved their respective limbs or appendages. ### Claim 8 (depends on 5) The method according to claim 5, wherein the determining by the computer at least one of an influence relationship or the level of coherence between the third time series comprising the movement data of the interviewer and the fourth time series comprising the movement data of the interviewee comprises modelling the third time series comprising the movement data of the interviewer and the fourth time series comprising the movement data of the interviewee as a coupled Hidden Markov Model. ### Claim 9 A system of detecting deception by an interviewee and predicting accuracy by an interviewer in an interview conducted by the interviewer, comprising: a first sensor to record a first time series of physiological data comprising electrocardiogram data of the interviewer during the interview; a second sensor to record a second time series of physiological data comprising electrocardiogram data of the interviewee during the interview; a processor to: generate a time lagged time series comprising the second time series of physiological data shifted a predetermined length of time; determine a dyadic feature based on the first time series of physiological data comprising electrocardiogram data of the interviewer and the time lagged time series of the interviewee, wherein the dyadic feature indicates at least one of a time-based or frequency-based relationship between the first time series of physiological data comprising electrocardiogram data of the interviewer and the time lagged time series of the interviewee; determine, based on the dyadic feature, a level of coherence between the first time series of physiological data and the second time series of physiological data; and generate an accuracy score for the interviewer based on the level of coherence between the first time series of the physiological data and the second time series of physiological data being above a predetermined threshold. ### Claim 10 (depends on 9) The system according to claim 9, further comprising: a third sensor to record skin conductance data or pulse oximetry data of the interviewer or the interviewee. ### Claim 11 (depends on 9) The system according to claim 9, further comprising the processor to calculate a level of entropy, a heart rate, and time intervals between consecutive normal sinus beats in the first time series and the second time series. ### Claim 12 (depends on 9) The system according to claim 9, further comprising the processor to apply an auto-regressive integrated moving average analysis to the first time series of physiological data of the interviewer and the second time series of physiological data of the interviewee. ### Claim 13 (depends on 9) The system according to claim 9, wherein, when the processor determines the level of coherence between a third time series comprising bodily motions of the interviewer and a fourth time series comprising bodily motions of the interviewee. ### Claim 14 (depends on 13) The system according to claim 13, further comprising: a video recording device to record a video of the interview; and the processor to process the recorded video to generate the third time series comprising the bodily motions of the interviewer and the fourth time series comprising the bodily motions of the interviewee. ### Claim 15 (depends on 14) The system according to claim 14, further comprising the processor to determine at least one of the influence relationship or the level of coherence between the third time series comprising the bodily motions of the interviewer and the fourth time series comprising the bodily motions of the interviewee with a coupled Hidden Markov Model. ### Claim 16 (depends on 14) The system according to claim 14, wherein the third time series comprising the bodily motions of the interviewer and the fourth time series comprising the bodily motions of the interviewee indicate times at which the interviewer and interviewee, respectively, moved their respective limbs or appendages. ### Claim 17 A non-transitory computer readable medium storing a computer-readable program to detect deception by an interviewee and predicting accuracy by an interviewer in an interview conducted by the interviewer, wherein execution of the computer-readable program by at least one processor cause the at least one processor to: record, from at least a first sensor, a first time series of physiological data comprising electrocardiogram data of the interviewer during the interview; record, from at least a second sensor, a second time series of physiological data comprising electrocardiogram data of the interviewee during the interview; generate a time lagged time series comprising the second time series of physiological data shifted a predetermined length of time; determine a dyadic feature based on the first time series of physiological data comprising electrocardiogram data of the interviewer and the time lagged time series of the interviewee, wherein the dyadic feature indicates at least one of a time-based or frequency-based relationship between the first time series of physiological data comprising electrocardiogram data of the interviewer and the time lagged time series of the interviewee; determine, based on the dyadic feature, a level of coherence between the first time series of physiological data and the second time series of physiological data; and generate an accuracy score for the interviewer based on the level of coherence between the first time series of the physiological data and the second time series of physiological data being above a predetermined threshold. ### Claim 18 (depends on 17) The non-transitory computer readable medium according to claim 17, wherein execution of the computer-readable program by the at least one processor cause the at least one processor to determine the level of coherence between a third time series comprising movement data of the interviewer and a fourth time series comprising movement data of the interviewee.