PROBLEM TO BE SOLVED: To simply and accurately determine whether a person may suffer from disease that appears in a speech without depending on the speech.SOLUTION: A sound collection section 101 collects voice that a person uttered. A voice recognition section 201 performs voice recognition processing to voice data uttered by the person to convert it to text data. A compression section 301 performs reversible compression of the text data transmitted from a voice recognition server 200. A compressibility ratio calculation section 302 calculates a compressibility ratio of the text data compressed by the compression section 301. A determination section 303 determines that a person to be diagnosed may suffers from dementia or not when the compressibility ratio calculated by the compressibility ratio calculation section 302 is larger than a prescribed threshold value.SELECTED DRAWING: Figure 1
