Adapting large language models to individual users remains challenging because systems must capture fine-grained personal preferences while serving many users efficiently. We introduce CARD, a hierarchical personalization framework that first learns preferences shared across user groups and then captures how individual users depart from those shared patterns. CARD trains a parameter-efficient adaptation module for each group, providing a reliable prior even when a user has limited history. It then learns a compact user preference vector by contrasting user-authored responses with outputs from the corresponding group model, without requiring manual annotations. During inference, both the backbone and the group adaptation module remain frozen, while the user vector applies a low-rank correction to the token logits during decoding. Experiments on personalization benchmarks show that CARD consistently outperforms competitive baselines while requiring only a compact vector for each user.

Simulating dementia patients with large language models (LLMs) is challenging due to the need to jointly model cognitive impairment, emotional dynamics, and nonverbal behaviors over long conversations. We present DemMA, an expert-guided dementia dialogue agent for high-fidelity multi-turn patient simulation. DemMA constructs clinically grounded dementia personas by integrating pathology information, personality traits, and subtype-specific memory-status personas informed by clinical experts. To move beyond text-only simulation, DemMA explicitly models nonverbal behaviors, including motion, facial expressions, and vocal cues. We further introduce a Chain-of-Thought distillation framework that trains a single LLM to jointly generate reasoning traces, patient utterances, and aligned behavioral actions within one forward pass, enabling efficient deployment without multi-agent inference.

Personalized large language models are often expected to follow explicit style instructions, yet we find that such instructions can undermine the user-specific characteristics that personalization methods aim to preserve. We call this failure mode personalization collapse: explicit style control can conflict with implicit user preferences. To address this challenge, we propose PsPLUG, a lightweight plug-in that learns a user-specific residual after accounting for the requested style. PsPLUG also allows us to tune personalization strength at inference time. Our experiments show that explicit style instructions can diminish personalization in existing methods, whereas PsPLUG better preserves user preferences while providing precise control over the balance between personalization and style adherence.

Pathological speech from patients with neurodegenerative and neuromotor disorders is often acoustically distorted and linguistically fragmented, making pathological speech reconstruction necessary to recover intended textual content from distorted and incomplete speech recordings. Crucially, such recordings are rarely uniformly degraded: some words or short phrases remain reliable and can serve as audible anchors for reconstructing the corrupted surrounding content. We introduce **Anchor-gated Phonetic Group Relative Policy Optimization (AP-GRPO)**, a GRPO framework with phonetic reward that aligns speech language models (SLMs) through audible-anchor preservation and inter-anchor phonetic compatibility to the original speech signal. AP-GRPO consists of: (i) an **anchor-gated reward** that matches reliable audible anchors in clear regions; and (ii) an **inter-anchor phonetic alignment reward** that evaluates whether recovered contents are phonetically supported by the corresponding corrupted inter-anchor speech span. Across four disease conditions, AP-GRPO improves faithful speech reconstruction, and the learned anchor constraint automatically adapts to each condition and thus reveals interpretable disease-specific profiles: conditions with severe articulatory degradation require stronger anchor enforcement, whereas milder impairment or linguistically impaired conditions rely more on phonetic alignment for inter-anchor recovery.

The growing prevalence of multi-sensor wearable devices generates vast amounts of long-term, multimodal time-series data, posing significant challenges for context-aware Large Language Models (LLMs). Current LLM-based health analysis methods typically rely on manually curated context, which becomes impractical with increasing data volume and sensor diversity. To overcome these limitations, we introduce Wearable As Graph (WAG), a novel framework that automates context retrieval for LLMs using personalized knowledge graphs. WAG constructs knowledge graphs mapping relationships between wearable modalities and incorporates user-specific data. We develop a data-driven retrieval pipeline that leverages both global (long-term) and local (short-term) relationships within metrics to identify the most relevant nodes for user queries. We evaluate WAG on a benchmark of over 10k data-associated queries created from multiple wearable datasets. WAG consistently delivers the strongest overall performance, while using substantially fewer input tokens than query-window full-context prompting in the diagnostic setting. These gains remain broadly consistent under human evaluation and across a cross-model generator–judge matrix, supporting structured, query-adaptive context retrieval for LLM-driven analysis of wearable data.