Information processing device, inference model generation method, and non-transitory computer-readable medium

An inference model generation method for causing at least one processor to execute:

• waveform detection processing of assigning coordinate values to pixels included in image data indicating the waveform of a biological signal, detecting pixel groups each including adjacent pixels from the image data, selecting, as a pixel group relevant to a portion of the waveform, a pixel group satisfying at least one of (i) a number of elements equal to or larger than a threshold and (ii) a distance from a left end pixel to a right end pixel equal to or more than a threshold, detecting, by pattern matching, a pixel group including a calibration wave, and excluding pixels of the pixel group including the calibration wave from the selected pixel group;
• digitizing processing of generating numerical data representing a time-series change of the biological signal in a predetermined unit relevant to the biological signal by converting coordinates of pixels in the selected pixel group based on a scale for converting a pixel dimension to the predetermined unit;
• and
• inference model generation processing of generating an inference model inferring a state of the subject from numerical data of a biological signal of the subject by machine learning using training data generated from the numerical data.

A non-transitory computer-readable medium storing a digitizing program for causing a computer to function as:

• a memory storing instructions;
• and
• a processor configured to execute the instructions to:
• assign coordinate values to pixels included in image data indicating a waveform of a biological signal;
• detect, from the image data, pixel groups each including adjacent pixels;
• select, as a pixel group relevant to a portion of the waveform, a pixel group satisfying at least one of (i) a number of elements equal to or larger than a threshold and (ii) a distance from a left end pixel to a right end pixel equal to or more than a threshold;
• detect, by pattern matching, a pixel group including a calibration wave;
• exclude pixels of the pixel group including the calibration wave from the selected pixel group;
• and
• generate numerical data representing a time-series change of the biological signal in a predetermined unit relevant to the biological signal by converting coordinates of pixels in the selected pixel group based on a scale for converting a pixel dimension to the predetermined unit.