The technology introduces an AI-driven monitoring and prediction system designed to identify real and imminent suicide risk at an early stage. The system integrates clinical, behavioral, and digital signals to generate accurate and interpretable risk assessments, enabling timely, life-saving interventions across healthcare, defense, and community settings.
The Problem
- Current suicide-risk assessment tools are highly limited:
- They rely heavily on self-report questionnaires, which are often incomplete or unreliable.
- There is a severe lack of validated clinical datasets containing real suicide attempts.
- Existing AI models in the field function as black-box systems, limiting trust and clinical adoption.
- No practical solution today can integrate multiple data modalities, clinical notes, speech, text, images, and social-media behavior.
- The result: late detection, insufficient intervention windows, and a major public-health gap.
The Technology
The technology provides a multimodal AI system built to detect severe suicide risk using diverse data sources:
Integrated Multimodal Data
- Social-media text and images
- Voice recordings capturing stress and cognitive load
- Structured clinical interviews (e.g., SSRS-C, FNSSI)
- Validated psychological questionnaires (PHQ-9, GAD-7, PCL-5, UCLA Loneliness, etc.)
- Psychiatric diagnoses, medical records, and documented prior attempts
- This data is collected from three populations: individuals with recent suicide attempts, psychiatric patients at elevated risk, and the general public.
Advanced AI Models
- Machine-learning and deep-learning models trained on clinically verified outcomes
- Interpretability-driven algorithms (counterfactual and causal-inference models) that “open the black box”
- Ability to identify new risk factors beyond traditional clinical assessment
High-Accuracy Prediction
The target performance:
AUC > 0.9, representing a highly reliable early-warning capability.
Advantages
- Early and Accurate Risk Detection High-performance AI (target AUC > 0.9) identifies severe suicide risk earlier than traditional self-report tools.
- Multimodal, Clinically Validated Data Integration Combines clinical interviews, medical records, speech, questionnaires, and social-media signals into one coherent risk model.
- Interpretable and Clinician-Friendly AI Causality-based and counterfactual models reveal why the system predicts risk, supporting trust, safety, and ethical adoption.
- Scalable, Software-Only Implementation No hardware required; integrates with EHRs and digital-health platforms across healthcare, defense, and community systems.
- Personalized, High-Resolution Risk Profiling Generates a detailed, individualized risk signature that outperforms existing single-signal screening tools.
- Strong Commercial and Public-Health Impact Reduces emergency interventions and hospitalizations, supports national prevention programs, and aligns with growing global investment in mental-health technologies.
Applications
- Healthcare systems: hospitals, HMOs, emergency psychiatric services
- Defense & security organizations: monitoring high-stress personnel
- Digital health platforms: remote-care apps, chatbots, therapy interfaces
- Public-sector programs: national suicide-prevention initiatives
- Families & caregivers: early warning in at-risk populations
Business Development Contacts
Motti Koren
Director of Business Development, Life Sciences
