Voice Analysis System for Suicide Risk Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods lack standardized approaches for analyzing nonverbal behaviors, such as gestures, facial expressions, and voice characteristics, to effectively assess suicide risk in adolescents, which are crucial factors in clinical decision-making for psychiatry.
Innovation Solution
The development of systems and methods using computerized analysis to classify nonverbal behaviors and cues, specifically acoustic characteristics like prosodic and voice quality features, to categorize subjects as suicidal or non-suicidal, employing machine learning algorithms like hidden Markov models and support vector machines, and incorporating voice source and voice quality-related features associated with 'breathy voice' or 'breathy phonation' types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If standardized approaches for analyzing nonverbal behaviors are implemented, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent replaces manual clinical assessment of nonverbal behaviors with automated computer-based acoustic analysis. Machine learning algorithms process voice recordings to extract prosodic features and classify suicide risk, substituting the mechanical process of human observation and interpretation with an automated computational system that provides standardized, objective measurements of voice characteristics.
Solution Approach 2:
The patent introduces an intermediary computational layer between the raw voice data and clinical decision-making. The system uses feature extraction algorithms to transform complex acoustic signals into standardized prosodic parameters, which then serve as inputs for classification algorithms. This intermediary processing stage standardizes the analysis while managing complexity through modular architecture.
2Adaptability or versatility
If speaker-independent classification is used, then adaptability is improved, but measurement precision decreases
Solution Approach 1:
The patent transforms the classification problem from speaker-dependent to speaker-independent by changing the feature representation parameters. Instead of using raw acoustic features that vary between speakers, the system extracts normalized prosodic parameters (such as pitch contours, speech rate, pause patterns) that capture universal indicators of suicidal risk while being invariant to individual speaker characteristics. This parameter transformation enables generalization across speakers while maintaining diagnostic precision.
Data Source
AI summary
A method for assessing suicide risk for a human subject including receiving recorded voice data of the subject; and classifying the subject as suicidal or non-suicidal based upon a computerized analysis of one or more nonverbal characteristics of the speech data, especially features associated with a breathy phonation type. The analysis of the nonverbal characteristics of the voice data can include an analysis of acoustic characteristics of speech, and/or an analysis of prosodic and voice quality-related features of the voice data. Related apparatus, systems, techniques and articles are also described.