Newest papers

NarraSpace: Multimodal Scene-Grounded Evidence for Picture-Description Dementia Classification

Qingming Li, Ruiyan Sun, Satoshi Nakamura · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.28616 · cited by 0 works (OpenAlex) · open-access PDF

French

La classification de la démence par description d'image utilise une scène visuelle partagée, mais une prédiction au niveau de la transcription seule n'identifie pas les preuves de scène sous-jacentes à ses entrées. NarraSpace relie les mots et les temps de parole aux éléments d'image nommés et récupère les déclarations pour les événements de scène prédéfinies. Les descripteurs narratifs-spatiaux résument la couverture et l'organisation temporelle. Un modèle de langue congelée contraste le texte de chaque événement sous des exemples de la maladie d'Alzheimer et des participants à la formation de contrôle de la parole en bonne santé, ainsi que des mesures de surprises de transcription et de couverture. Les enregistrements source retenus relient les scores

Machine translation (Meta M2M100); the English original follows.

Abstract

Picture-description dementia classification uses a shared visual scene, but a transcript-level prediction alone does not identify the scene evidence underlying its inputs. NarraSpace links words and speech times to named picture elements and retrieves utterances for predefined scene events. Narrative-spatial descriptors summarize coverage and temporal organization. A frozen language model contrasts each event's text under examples from Alzheimer's disease and healthy-control training participants, alongside transcript surprisal and coverage measurements. Retained source records link event scores to their source utterances, speech intervals, and lexical matches. On the official 48-participant test set of the Alzheimer's Dementia Recognition through Spontaneous Speech benchmark, a fixed support vector classifier achieves $80.88\%\pm2.56\%$ mean accuracy over fifty seeds; averaged probabilities yield $81.25\%$ accuracy and $91.32\%$ area under the receiver operating characteristic curve. NarraSpace makes the construction of scene-linked classification inputs explicit and inspectable by connecting event-indexed language-model measurements to their source utterances, speech timing, and extraction records. Such source-linked measurements could support clinician review of picture descriptions by presenting quantitative summaries alongside the original speech evidence.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Classifier (UML), Dementia, Natural language processing, Speech recognition, Artificial intelligence

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange