METHOD AND SYSTEM FOR AUTOMATIC POLYGRAPH TESTING USING TWO MACHINE LEARNING MODELS

EA202593380A1Pending Publication Date: 2026-11-02PUBLICHNOE AKTSIONERNOE OBSHCHESTVO SBERBANK ROSSII (PAO SBERBANK)
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
EA202593380
Authority / Receiving Office
EA · EA
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-11-24
Filing Date
2025-12-18
Publication Date
2026-11-02
Estimated Expiration
2045-12-18

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

The invention relates to an automated method and system for automatic polygraph testing using machine learning algorithms. The technical result of implementing the claimed method is an increase in the accuracy of polygraph testing. This technical result is achieved by implementing a computer-implemented method performed using a computing system containing two machine learning models, wherein the method includes the following steps: obtaining polygraph test records, processing the received signals, processing a vector of physiological statistics using a first machine learning model (MO1), processing a vector of physiological statistics using a second machine learning model (MO2), and determining that the answer is false if the average value is above or equal to the threshold value, or that the answer is true if the average value is below the threshold value.
Need to check novelty before this filing date? Find Prior Art

Claims

1. A computer-implemented method for automatic polygraph testing, performed using a computing system containing two machine learning models, wherein the method performs the steps in which receive records of polygraph examinations containing at least sensor signals with time scales on which the beginning of the question, the end of the question, the beginning of the answer to the question and the text of the question are marked; carry out processing of the received signals, and during the course of this processing the following is carried out: defining time intervals for extracting variables based on the question start and end timestamps, the answer timestamp, and obtaining the question text, determination of the topic of the question based on the text of the question, extraction of characteristic descriptive physiological statistics describing the physiological state of the subject from each signal at certain time intervals, processing of the obtained physiological statistics, which involves normalization of signals and aggregation of the processed physiological statistics within the framework of a specific topic of the question, construction of a vector of physiological statistics, based on the results obtained in the previous stage, describing the state of the subject answering a question related to a specific topic; They process the vector of physiological statistics using the first machine learning model (ML1), and during this processing the following is carried out: reduction of the said vector to the dimension specified for MO1, by selecting the most significant physiological statistics that influence the accuracy of MO1 prediction, feeding the reduced vector to MO1 to obtain the output value of MO1; They process the vector of physiological statistics using the second machine learning model (ML2), and during this processing the following is carried out: reduction of the said vector to the dimension specified for MO2 by selecting the most significant physiological statistics that influence the accuracy of MO2 prediction, feeding the reduced vector into MO2 to obtain the MO2 output value; average the output values ​​of MO1 and MO2; compare the average values ​​of the models with a given threshold value; and determine that the answer is false if the mean value is above or equal to the threshold value, or the answer is true if the mean value is below the threshold value.

2. The method according to claim 1, characterized in that the records of polygraph tests contain signals from sensors, including at least one of: heart rate (HR), galvanic skin response (GSR), blood pressure, upper and lower respiration, piezoplethysmogram, photoplethysmogram.

3. The method according to paragraph 1, characterized in that the MO models are of the logistic regression type.

4. An automatic polygraph testing system containing at least one processor; at least one memory connected to the processor, which contains machine-readable instructions that, when executed by at least one processor, ensure the execution of the method according to any one of claims 1 to 3.