Hierarchical Dual-Head Model for Suicide Risk Assessment via MentalRoBERTa

TL;DR

This paper introduces a hierarchical dual-head neural network based on MentalRoBERTa for multi-level suicide risk assessment from social media posts, effectively handling class imbalance, temporal dynamics, and risk level relationships.

Contribution

It presents a novel dual-head model combining ordinal and categorical predictions, with temporal and time interval embeddings, trained using a mixed loss function for improved accuracy.

Findings

Achieved higher macro F1 scores compared to baseline models.

Effectively modeled temporal posting patterns and risk level relationships.

Reduced computational costs through layer freezing and mixed-precision training.

Abstract

Social media platforms have become important sources for identifying suicide risk, but automated detection systems face multiple challenges including severe class imbalance, temporal complexity in posting patterns, and the dual nature of risk levels as both ordinal and categorical. This paper proposes a hierarchical dual-head neural network based on MentalRoBERTa for suicide risk classification into four levels: indicator, ideation, behavior, and attempt. The model employs two complementary prediction heads operating on a shared sequence representation: a CORAL (Consistent Rank Logits) head that preserves ordinal relationships between risk levels, and a standard classification head that enables flexible categorical distinctions. A 3-layer Transformer encoder with 8-head multi-head attention models temporal dependencies across post sequences, while explicit time interval embeddings capture posting behavior dynamics. The model is trained with a combined loss function (0.5 CORAL + 0.3 Cross-Entropy + 0.2 Focal Loss) that simultaneously addresses ordinal structure preservation, overconfidence reduction, and class imbalance. To improve computational efficiency, we freeze the first 6 layers (50%) of MentalRoBERTa and employ mixed-precision training. The model is evaluated using 5-fold stratified cross-validation with macro F1 score as the primary metric.