FIGURE 3. Visualization and analysis of inference results for speech sound model output in PD. (A) The inference findings for instances identified as male, H&Y 4, UPDRS II-part 5 score of 3, and UPDRS III-part 18 score of 3 within the PD group. The corresponding class for each of 500 frames along the time axis of the waveform. (B) The trajectory for detecting words decoded from vocalized sounds is visually depicted. (C) The probability at the detection juncture in the time domain. (D) The decoded words are superimposed onto the waveform within the time domain. In contrast to the scenario observed in the HC group, a maximum time delay, 5.45 (t_max) seconds was identified.

PD: Parkinson’s disease, H&Y: Hoehn and Yahr, UPDRS: Unified Parkinson’s Disease Rating Scale, HC: healthy control, t_max: maximum delay time.