arXiv subjects

Jiening Zhan

Publications and source records attributed to Jiening Zhan.

10 recordsLinked to original sources

arXiv2026

Contrastive Reflection for Iterative Prompt Optimization

LLM agents are becoming central to information retrieval: they issue retrieval queries, synthesize answers, and increasingly serve as judges for IR evaluation. Improving the prompts that control these agents is an optimization problem, but in applied IR settings it often looks less like blind search and more like debugging. Engineers need to know which behavior failed, which nearby behavior still worked, what distinguishes the two, and whether a prompt edit improves held-out quality without introducing regressions. We present Contrastive Reflection, an iterative prompt-optimization framework for agentic IR workflows. The framework starts from a task-centric quality definition: QA agents expose retrieval or reasoning traces, and grading agents expose dimension-level scores and rationales. These structured traces are used to identify error-anchored behavioral slices, add nearby successful examples from the same region, and ask a Teacher LLM to propose a targeted prompt edit. Candidate edits are accepted only when validation performance improves, optionally subject to regression checks. We instantiate the framework with a tree-based slice selector, but the contribution is the contrastive reflection loop rather than the tree itself. On a public HotpotQA retrieval-augmented QA setup, one tree-selected contrastive repair improves held-out exact-match accuracy from 51.4% to 60.4%. Failure-only and random-evidence variants improve less and break more previously correct examples. A light instruction-only comparison places the method near modern prompt optimizers: MIPROv2 reaches 59.4% and GEPA 57.0%. The result is an interpretable optimization loop for IR agents, aimed at making prompt repair more inspectable and validation-driven.

cs.AI↗

arXiv2026

Beyond BMI: Smartphone Body Composition Phenotyping for Cardiometabolic Risk Assessment

Body Mass Index (BMI) is a widely accessible but imprecise proxy of cardiometabolic health. While assessing true body composition is superior, gold-standard methods like Dual-Energy X-ray Absorptiometry (DXA) are not scalable. We address this gap by developing and validating "PhotoScan," a method to estimate body composition from smartphone imagery. We pretrained a deep learning model on UK Biobank participants (N=35,323) and fine-tuned on a newly recruited clinical cohort (PhotoBIA cohort, N=677) with diverse ethnicity, age, and body fat distribution, achieving high accuracy against DXA for total body fat percentage (BF%, MAE = 2.15%), Android-to-Gynoid fat ratio (A/G, MAE = 0.11), and visceral-to-subcutaneous fat area ratio (V/S, MAE = 0.09). Generalizability of the model was demonstrated on an independent metabolic health study cohort (MetabolicMosaic cohort, N=132 participants), achieving MAEs of 2.13% for BF%, 0.09 for A/G, and 0.09 for V/S. We then evaluated the clinical utility of these metrics in the MetabolicMosaic cohort by predicting insulin resistance (IR). Adding PhotoScan-derived body composition metrics to baseline demographics model (Age, Sex, BMI) significantly improved insulin resistance classification (Area Under the Receiver Operating Characteristic Curve "AUROC" 76.0% vs 69.2%, DeLong test p=0.002, Net Reclassification Index "NRI" 0.593). Crucially, this accessible smartphone method achieved performance nearly equivalent to adding clinical-grade DXA data to baseline demographics model (AUROC 77.3% vs 69.2%, DeLong test p=0.004, NRI 0.748). These findings demonstrate that smartphone-based phenotyping captures clinically meaningful risk signals missed by BMI and anthropometrics, offering a scalable alternative to DXA for cardiometabolic risk stratification.

q-bio.QM↗

arXiv2025

The Anatomy of a Personal Health Agent

Health is a fundamental pillar of human wellness, and the rapid advancements in large language models (LLMs) have driven the development of a new generation of health agents. However, the application of health agents to fulfill the diverse needs of individuals in daily non-clinical settings is underexplored. In this work, we aim to build a comprehensive personal health agent that is able to reason about multimodal data from everyday consumer wellness devices and common personal health records, and provide personalized health recommendations. To understand end-users' needs when interacting with such an assistant, we conducted an in-depth analysis of web search and health forum queries, alongside qualitative insights from users and health experts gathered through a user-centered design process. Based on these findings, we identified three major categories of consumer health needs, each of which is supported by a specialist sub-agent: (1) a data science agent that analyzes personal time-series wearable and health record data, (2) a health domain expert agent that integrates users' health and contextual data to generate accurate, personalized insights, and (3) a health coach agent that synthesizes data insights, guiding users using a specified psychological strategy and tracking users' progress. Furthermore, we propose and develop the Personal Health Agent (PHA), a multi-agent framework that enables dynamic, personalized interactions to address individual health needs. To evaluate each sub-agent and the multi-agent system, we conducted automated and human evaluations across 10 benchmark tasks, involving more than 7,000 annotations and 1,100 hours of effort from health experts and end-users. Our work represents the most comprehensive evaluation of a health agent to date and establishes a strong foundation towards the futuristic vision of a personal health agent accessible to everyone.

cs.AI↗

arXiv2025

SensorLM: Learning the Language of Wearable Sensors

We present SensorLM, a family of sensor-language foundation models that enable wearable sensor data understanding with natural language. Despite its pervasive nature, aligning and interpreting sensor data with language remains challenging due to the lack of paired, richly annotated sensor-text descriptions in uncurated, real-world wearable data. We introduce a hierarchical caption generation pipeline designed to capture statistical, structural, and semantic information from sensor data. This approach enabled the curation of the largest sensor-language dataset to date, comprising over 59.7 million hours of data from more than 103,000 people. Furthermore, SensorLM extends prominent multimodal pretraining architectures (e.g., CLIP, CoCa) and recovers them as specific variants within a generic architecture. Extensive experiments on real-world tasks in human activity analysis and healthcare verify the superior performance of SensorLM over state-of-the-art in zero-shot recognition, few-shot learning, and cross-modal retrieval. SensorLM also demonstrates intriguing capabilities including scaling behaviors, label efficiency, sensor captioning, and zero-shot generalization to unseen tasks.

cs.LG↗

arXiv2024

Scaling Wearable Foundation Models

Wearable sensors have become ubiquitous thanks to a variety of health tracking features. The resulting continuous and longitudinal measurements from everyday life generate large volumes of data; however, making sense of these observations for scientific and actionable insights is non-trivial. Inspired by the empirical success of generative modeling, where large neural networks learn powerful representations from vast amounts of text, image, video, or audio data, we investigate the scaling properties of sensor foundation models across compute, data, and model size. Using a dataset of up to 40 million hours of in-situ heart rate, heart rate variability, electrodermal activity, accelerometer, skin temperature, and altimeter per-minute data from over 165,000 people, we create LSM, a multimodal foundation model built on the largest wearable-signals dataset with the most extensive range of sensor modalities to date. Our results establish the scaling laws of LSM for tasks such as imputation, interpolation and extrapolation, both across time and sensor modalities. Moreover, we highlight how LSM enables sample-efficient downstream learning for tasks like exercise and activity recognition.

cs.LG↗

arXiv2024

Transforming Wearable Data into Personal Health Insights using Large Language Model Agents

Deriving personalized insights from popular wearable trackers requires complex numerical reasoning that challenges standard LLMs, necessitating tool-based approaches like code generation. Large language model (LLM) agents present a promising yet largely untapped solution for this analysis at scale. We introduce the Personal Health Insights Agent (PHIA), a system leveraging multistep reasoning with code generation and information retrieval to analyze and interpret behavioral health data. To test its capabilities, we create and share two benchmark datasets with over 4000 health insights questions. A 650-hour human expert evaluation shows that PHIA significantly outperforms a strong code generation baseline, achieving 84% accuracy on objective, numerical questions and, for open-ended ones, earning 83% favorable ratings while being twice as likely to achieve the highest quality rating. This work can advance behavioral health by empowering individuals to understand their data, enabling a new era of accessible, personalized, and data-driven wellness for the wider population.

cs.AI↗

arXiv2024

Towards a Personal Health Large Language Model

In health, most large language model (LLM) research has focused on clinical tasks. However, mobile and wearable devices, which are rarely integrated into such tasks, provide rich, longitudinal data for personal health monitoring. Here we present Personal Health Large Language Model (PH-LLM), fine-tuned from Gemini for understanding and reasoning over numerical time-series personal health data. We created and curated three datasets that test 1) production of personalized insights and recommendations from sleep patterns, physical activity, and physiological responses, 2) expert domain knowledge, and 3) prediction of self-reported sleep outcomes. For the first task we designed 857 case studies in collaboration with domain experts to assess real-world scenarios in sleep and fitness. Through comprehensive evaluation of domain-specific rubrics, we observed that Gemini Ultra 1.0 and PH-LLM are not statistically different from expert performance in fitness and, while experts remain superior for sleep, fine-tuning PH-LLM provided significant improvements in using relevant domain knowledge and personalizing information for sleep insights. We evaluated PH-LLM domain knowledge using multiple choice sleep medicine and fitness examinations. PH-LLM achieved 79% on sleep and 88% on fitness, exceeding average scores from a sample of human experts. Finally, we trained PH-LLM to predict self-reported sleep quality outcomes from textual and multimodal encoding representations of wearable data, and demonstrate that multimodal encoding is required to match performance of specialized discriminative models. Although further development and evaluation are necessary in the safety-critical personal health domain, these results demonstrate both the broad knowledge and capabilities of Gemini models and the benefit of contextualizing physiological data for personal health applications as done with PH-LLM.

cs.AI↗

arXiv2023

Large Language Models are Few-Shot Health Learners

Large language models (LLMs) can capture rich representations of concepts that are useful for real-world tasks. However, language alone is limited. While existing LLMs excel at text-based inferences, health applications require that models be grounded in numerical data (e.g., vital signs, laboratory values in clinical domains; steps, movement in the wellness domain) that is not easily or readily expressed as text in existing training corpus. We demonstrate that with only few-shot tuning, a large language model is capable of grounding various physiological and behavioral time-series data and making meaningful inferences on numerous health tasks for both clinical and wellness contexts. Using data from wearable and medical sensor recordings, we evaluate these capabilities on the tasks of cardiac signal analysis, physical activity recognition, metabolic calculation (e.g., calories burned), and estimation of stress reports and mental health screeners.

cs.CL↗

arXiv2019

Temporal Graph Convolutional Networks for Automatic Seizure Detection

Seizure detection from EEGs is a challenging and time consuming clinical problem that would benefit from the development of automated algorithms. EEGs can be viewed as structural time series, because they are multivariate time series where the placement of leads on a patient's scalp provides prior information about the structure of interactions. Commonly used deep learning models for time series don't offer a way to leverage structural information, but this would be desirable in a model for structural time series. To address this challenge, we propose the temporal graph convolutional network (TGCN), a model that leverages structural information and has relatively few parameters. TGCNs apply feature extraction operations that are localized and shared over both time and space, thereby providing a useful inductive bias in tasks where one expects similar features to be discriminative across the different sequences. In our experiments we focus on metrics that are most important to seizure detection, and demonstrate that TGCN matches the performance of related models that have been shown to be state of the art in other tasks. Additionally, we investigate interpretability advantages of TGCN by exploring approaches for helping clinicians determine when precisely seizures occur, and the parts of the brain that are most involved.

cs.LG↗

arXiv2014

Integer-Forcing Linear Receivers

Linear receivers are often used to reduce the implementation complexity of multiple-antenna systems. In a traditional linear receiver architecture, the receive antennas are used to separate out the codewords sent by each transmit antenna, which can then be decoded individually. Although easy to implement, this approach can be highly suboptimal when the channel matrix is near singular. This paper develops a new linear receiver architecture that uses the receive antennas to create an effective channel matrix with integer-valued entries. Rather than attempting to recover transmitted codewords directly, the decoder recovers integer combinations of the codewords according to the entries of the effective channel matrix. The codewords are all generated using the same linear code which guarantees that these integer combinations are themselves codewords. Provided that the effective channel is full rank, these integer combinations can then be digitally solved for the original codewords. This paper focuses on the special case where there is no coding across transmit antennas and no channel state information at the transmitter(s), which corresponds either to a multi-user uplink scenario or to single-user V-BLAST encoding. In this setting, the proposed integer-forcing linear receiver significantly outperforms conventional linear architectures such as the zero-forcing and linear MMSE receiver. In the high SNR regime, the proposed receiver attains the optimal diversity-multiplexing tradeoff for the standard MIMO channel with no coding across transmit antennas. It is further shown that in an extended MIMO model with interference, the integer-forcing linear receiver achieves the optimal generalized degrees-of-freedom.

cs.IT↗