About

Professor Hassan Ghasemzadeh is the director of the Embedded Machine Intelligence Lab (EMIL) at Arizona State University. His research focuses on the design, development, and validation of algorithms, tools, and technologies that enhance the utilization and large-scale adoption of pervasive systems, particularly in the domain of digital health. At EMIL, Professor Ghasemzadeh and his team develop next-generation AI, sensing, and digital health technologies that bridge the gap between algorithmic innovation and real-world clinical impact. Their work emphasizes robust, interactive, efficient, and trustworthy machine learning methods for pervasive systems operating under dynamic and resource-constrained conditions. To ensure translational relevance and large-scale adoption, the lab collaborates closely with clinicians, healthcare systems, patients, caregivers, and community partners to co-design, evaluate, and refine these technologies through clinical studies and real-world deployment. This interdisciplinary approach enables the creation of scalable, evidence-based, and impactful solutions that improve health outcomes, reduce the burden of care, empower patients and clinicians, and advance the future of precision and participatory healthcare. Professor Ghasemzadeh's research addresses transformative applications in remote and mobile healthcare, chronic disease management, preventive medicine, digital therapeutics, aging and elderly care support, rehabilitation, mental and behavioral health, smart homes and environments, emergency response, fitness and performance monitoring, and human-centered assistive technologies.

Research topics

• Computer Science
• Artificial Intelligence
• Machine Learning
• Geography
• Business
• Mathematics
• Knowledge management
• Psychology

Selected publications

• GUIDE: Reinforcement Learning for Behavioral Action Support in Type 1 Diabetes

  arXiv (Cornell University) · 2026-04-01

  preprintOpen accessSenior author

  Type 1 Diabetes (T1D) management requires continuous adjustment of insulin and lifestyle behaviors to maintain blood glucose within a safe target range. Although automated insulin delivery (AID) systems have improved glycemic outcomes, many patients still fail to achieve recommended clinical targets, warranting new approaches to improve glucose control in patients with T1D. While reinforcement learning (RL) has been utilized as a promising approach, current RL-based methods focus primarily on insulin-only treatment and do not provide behavioral recommendations for glucose control. To address this gap, we propose GUIDE, an RL-based decision-support framework designed to complement AID technologies by providing behavioral recommendations to prevent abnormal glucose events. GUIDE generates structured actions defined by intervention type, magnitude, and timing, including bolus insulin administration and carbohydrate intake events. GUIDE integrates a patient-specific glucose level predictor trained on real-world continuous glucose monitoring data and supports both offline and online RL algorithms within a unified environment. We evaluate both off-policy and on-policy methods across 25 individuals with T1D using standardized glycemic metrics. Among the evaluated approaches, the CQL-BC algorithm demonstrates the highest average time-in-range, reaching 85.49% while maintaining low hypoglycemia exposures. Behavioral similarity analysis further indicates that the learned CQL-BC policy preserves key structural characteristics of patient action patterns, achieving a mean cosine similarity of 0.87 $\pm$ 0.09 across subjects. These findings suggest that conservative offline RL with a structured behavioral action space can provide clinically meaningful and behaviorally plausible decision support for personalized diabetes management.

  PublisherDOI

• Trustworthy AI in digital health: a comprehensive review of robustness and explainability

  Progress in Biomedical Engineering · 2026-03-06

  articleOpen accessSenior author

  Ensuring trust in artificial intelligence (AI) systems is essential for the safe and ethical integration of machine learning systems into high-stakes domains such as digital health. Key dimensions, including robustness, explainability, fairness, accountability, and privacy, need to be addressed throughout the AI lifecycle, from problem formulation and data collection to model deployment and human interaction. While various contributions address different aspects of trustworthy AI, a focused synthesis on robustness and explainability, especially tailored to the healthcare context, remains limited. This review addresses that need by organizing recent advancements into an accessible framework, highlighting both technical and practical considerations. We present a structured overview of methods, challenges, and solutions, aiming to support researchers and practitioners in developing reliable and explainable AI (XAI) solutions for digital health. This review article is organized into three main parts. First, we introduce core pillars of trustworthy AI and discuss the technical and ethical challenges they pose, particularly in the context of digital health. Second, we explore application-specific trust considerations across domains such as intensive care, mental health, metabolic disease, and public health surveillance, highlighting how explainability, clinical validation, and human oversight support trust. Lastly, we present recent advancements in techniques aimed at improving robustness under data scarcity and distributional shifts, as well as XAI methods ranging from feature attribution to gradient-based interpretations and counterfactual explanations. This paper is further enriched with detailed discussions of the contributions toward robustness and explainability in digital health, the development of trustworthy AI systems in the era of large language models, and various evaluation metrics for measuring trust and related parameters such as validity, fidelity, and diversity, offering a roadmap for building safer and more reliable AI systems.

  PublisherDOI

• Self-Supervised Learning and Opportunistic Inference for Continuous Monitoring of Freezing of Gait in Parkinson’s Disease

  ACM Transactions on Computing for Healthcare · 2026-03-24

  preprintOpen accessSenior author

  Parkinson’s disease (PD) significantly affects patients’ quality of life through debilitating motor symptoms, such as Freezing of Gait (FoG). Continuous, in-home monitoring of FoG is essential for timely clinical intervention but remains challenging due to high power consumption, annotation cost, and the controlled environments required by current wearables. We introduce LIFT-PD (the source code is available at: https://github.com/shovito66/LIFT-PD ), a novel self-supervised learning (SSL) framework for real-time, patient-independent FoG detection that uniquely utilizes a single waist-worn accelerometer—an approach traditionally considered less optimal due to weaker gait signatures. LIFT-PD leverages SSL on unlabeled data collected from uncontrolled, real-world settings and employs a novel Differential Hopping Windowing Technique (DHWT) to address gait variability and dataset imbalance. Additionally, an opportunistic inference module selectively activates the deep learning model only during patient movement, significantly reducing power consumption and enabling continuous monitoring ( \(>\) 48 hours). Experimental results show that LIFT-PD achieves a 7.25% increase in precision and 4.4% improvement in accuracy compared to supervised and semi-supervised baseline models while requiring approximately 40% fewer labeled training samples. Evaluations across diverse patient characteristics-including severity, medication state, age, and gender-confirm the model’s robustness and clinical applicability, positioning LIFT-PD as a practical, energy-efficient, and scalable solution for continuous real-world FoG monitoring in PD.

  PublisherDOI

• GlyRAG: Context-Aware Retrieval-Augmented Framework for Blood Glucose Forecasting

  arXiv (Cornell University) · 2026-01-08

  preprintOpen accessSenior author

  Accurate forecasting of blood glucose from CGM is essential for preventing dysglycemic events, thus enabling proactive diabetes management. However, current forecasting models treat blood glucose readings captured using CGMs as a numerical sequence, either ignoring context or relying on additional sensors/modalities that are difficult to collect and deploy at scale. Recently, LLMs have shown promise for time-series forecasting tasks, yet their role as agentic context extractors in diabetes care remains largely unexplored. To address these limitations, we propose GlyRAG, a context-aware, retrieval-augmented forecasting framework that derives semantic understanding of blood glucose dynamics directly from CGM traces without requiring additional sensor modalities. GlyRAG employs an LLM as a contextualization agent to generate clinical summaries. These summaries are embedded by a language model and fused with patch-based glucose representations in a multimodal transformer architecture with a cross translation loss aligining textual and physiological embeddings. A retrieval module then identifies similar historical episodes in the learned embedding space and uses cross-attention to integrate these case-based analogues prior to making a forecasting inference. Extensive evaluations on two T1D cohorts show that GlyRAG consistently outperforms state-of-the art methods, achieving up to 39% lower RMSE and a further 1.7% reduction in RMSE over the baseline. Clinical evaluation shows that GlyRAG places 85% predictions in safe zones and achieves 51% improvement in predicting dysglycemic events across both cohorts. These results indicate that LLM-based contextualization and retrieval over CGM traces can enhance the accuracy and clinical reliability of long-horizon glucose forecasting without the need for extra sensors, thus supporting future agentic decision-support tools for diabetes management.

  PublisherDOI

• Counterfactual Modeling with Fine-Tuned LLMs for Health Intervention Design and Sensor Data Augmentation

  arXiv (Cornell University) · 2026-01-21

  preprintOpen accessSenior author

  Counterfactual explanations (CFEs) provide human-centric interpretability by identifying the minimal, actionable changes required to alter a machine learning model's prediction. Therefore, CFs can be used as (i) interventions for abnormality prevention and (ii) augmented data for training robust models. We conduct a comprehensive evaluation of CF generation using large language models (LLMs), including GPT-4 (zero-shot and few-shot) and two open-source models-BioMistral-7B and LLaMA-3.1-8B, in both pretrained and fine-tuned configurations. Using the multimodal AI-READI clinical dataset, we assess CFs across three dimensions: intervention quality, feature diversity, and augmentation effectiveness. Fine-tuned LLMs, particularly LLaMA-3.1-8B, produce CFs with high plausibility (up to 99%), strong validity (up to 0.99), and realistic, behaviorally modifiable feature adjustments. When used for data augmentation under controlled label-scarcity settings, LLM-generated CFs substantially restore classifier performance, yielding an average 20% F1 recovery across three scarcity scenarios. Compared with optimization-based baselines such as DiCE, CFNOW, and NICE, LLMs offer a flexible, model-agnostic approach that generates more clinically actionable and semantically coherent counterfactuals. Overall, this work demonstrates the promise of LLM-driven counterfactuals for both interpretable intervention design and data-efficient model training in sensor-based digital health. Impact: SenseCF fine-tunes an LLM to generate valid, representative counterfactual explanations and supplement minority class in an imbalanced dataset for improving model training and boosting model robustness and predictive performance

  PublisherDOI

• Counterfactual Modeling with Fine-Tuned LLMs for Health Intervention Design and Sensor Data Augmentation

  arXiv (Cornell University) · 2026-01-21

  articleOpen accessSenior author

  Counterfactual explanations (CFEs) provide human-centric interpretability by identifying the minimal, actionable changes required to alter a machine learning model's prediction. Therefore, CFs can be used as (i) interventions for abnormality prevention and (ii) augmented data for training robust models. We conduct a comprehensive evaluation of CF generation using large language models (LLMs), including GPT-4 (zero-shot and few-shot) and two open-source models-BioMistral-7B and LLaMA-3.1-8B, in both pretrained and fine-tuned configurations. Using the multimodal AI-READI clinical dataset, we assess CFs across three dimensions: intervention quality, feature diversity, and augmentation effectiveness. Fine-tuned LLMs, particularly LLaMA-3.1-8B, produce CFs with high plausibility (up to 99%), strong validity (up to 0.99), and realistic, behaviorally modifiable feature adjustments. When used for data augmentation under controlled label-scarcity settings, LLM-generated CFs substantially restore classifier performance, yielding an average 20% F1 recovery across three scarcity scenarios. Compared with optimization-based baselines such as DiCE, CFNOW, and NICE, LLMs offer a flexible, model-agnostic approach that generates more clinically actionable and semantically coherent counterfactuals. Overall, this work demonstrates the promise of LLM-driven counterfactuals for both interpretable intervention design and data-efficient model training in sensor-based digital health. Impact: SenseCF fine-tunes an LLM to generate valid, representative counterfactual explanations and supplement minority class in an imbalanced dataset for improving model training and boosting model robustness and predictive performance

  PublisherOA PDF

• scaleTRIM: Scalable TRuncation-Based Integer Approximate Multiplier With Linearization and Compensation

  IEEE Access · 2026-01-01

  articleOpen access

  In this paper, we propose a scalable approximate multiplier design, scaleTRIM, that approximates the multiplication operation using fitted linear functions, also referred to as linearization. We show that multiplication operations can be completely replaced by low-cost addition and bit-wise shift operations by exploiting linearization. Moreover, our proposed design utilizes a lookup table (LUT)-based compensation unit as a novel error-reduction method. In essence, input operands are truncated to a reduced bit-width representation (i.e., <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">h</i> bits) based on their leading-one positions. Then, a curve-fitting method is employed to map the product term to a linear function. Additionally, a piecewise constant error-correction term is used to reduce the approximation error. To compute the piecewise constant, we divide the function space into <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">M</i> segments and average the errors within each segment. In particular, our multiplier supports various degrees of truncation and error compensation to offer a range of accuracy-efficiency trade-offs. The proposed multiplier improves the Mean Relative Error Distance (MRED) by about 15.2% while satisfying the efficiency constraint and improves the Power Delay Product (PDP) by about 22.8% while satisfying the accuracy and efficiency constraints compared to different state-of-the-art approximate multipliers. From a usability perspective, our evaluation of the proposed design for image classification using Deep Neural Networks (DNNs) demonstrates that scaleTRIM offers a better accuracy-efficiency trade-off than state-of-the-art approximate multiplier designs.

  PublisherDOI

• GUIDE: Reinforcement Learning for Behavioral Action Support in Type 1 Diabetes

  ArXiv.org · 2026-04-01

  articleOpen accessSenior author

  Type 1 Diabetes (T1D) management requires continuous adjustment of insulin and lifestyle behaviors to maintain blood glucose within a safe target range. Although automated insulin delivery (AID) systems have improved glycemic outcomes, many patients still fail to achieve recommended clinical targets, warranting new approaches to improve glucose control in patients with T1D. While reinforcement learning (RL) has been utilized as a promising approach, current RL-based methods focus primarily on insulin-only treatment and do not provide behavioral recommendations for glucose control. To address this gap, we propose GUIDE, an RL-based decision-support framework designed to complement AID technologies by providing behavioral recommendations to prevent abnormal glucose events. GUIDE generates structured actions defined by intervention type, magnitude, and timing, including bolus insulin administration and carbohydrate intake events. GUIDE integrates a patient-specific glucose level predictor trained on real-world continuous glucose monitoring data and supports both offline and online RL algorithms within a unified environment. We evaluate both off-policy and on-policy methods across 25 individuals with T1D using standardized glycemic metrics. Among the evaluated approaches, the CQL-BC algorithm demonstrates the highest average time-in-range, reaching 85.49% while maintaining low hypoglycemia exposures. Behavioral similarity analysis further indicates that the learned CQL-BC policy preserves key structural characteristics of patient action patterns, achieving a mean cosine similarity of 0.87 $\pm$ 0.09 across subjects. These findings suggest that conservative offline RL with a structured behavioral action space can provide clinically meaningful and behaviorally plausible decision support for personalized diabetes management.

  PublisherOA PDF

• Understanding Associations Between Behavioral Factors and Glycemic Outcomes in Patients with Type 1 Diabetes Using Automated Insulin Delivery Systems: Longitudinal Observational Study (Preprint)

  2026-03-05

  articleOpen accessSenior author

  <sec> <title>BACKGROUND</title> Despite advances in continuous glucose monitoring (CGM) and automated insulin delivery (AID) systems, many individuals with type 1 diabetes (T1D) fail to achieve recommended glycemic targets. Although behavioral factors (e.g., physical activity, sleep, diet, insulin timing) influence glucose outcomes, the behavioral context under free-living conditions remains insufficiently characterized. </sec> <sec> <title>OBJECTIVE</title> To examine associations between real-world behavioral patterns and glycemic outcomes in individuals with type 1 diabetes using automated insulin delivery systems by analyzing multimodal data from wearable sensors, mobile food logs, and continuous glucose monitoring. </sec> <sec> <title>METHODS</title> We conducted a prospective observational study involving 19 adults with T1D using AID systems over a 30-day period. Participants wore a smartwatch to capture behavioral metrics, including step counts, exercise duration, and sleep duration, and used a custom mobile application to log time-stamped food intake. Wearable and mobile app data were integrated with AID system data to construct a multimodal dataset. Behavioral–glycemic relationships were analyzed using a complementary framework combining unsupervised clustering and correlation analyses across individuals. </sec> <sec> <title>RESULTS</title> Clustering revealed distinct groups with similar overall activity and intake patterns but different percentages of time-in-range (TIR ≈ 69–86%), indicating that comparable behavioral profiles were associated with different levels of glycemic control. Insulin timing relative to meals consistently differentiated glycemic profiles, whereas physical activity and carbohydrate intake alone showed weaker separation. Correlation analysis identified average meal–bolus delay as one of the strongest behavioral correlates of glycemic outcomes, with a negative association with TIR (ρ ≈ −0.67). Sleep duration showed a moderate positive association with TIR and lower variability, while activity- and intake-related measures were strongly interrelated but less directly associated with glycemic metrics. </sec> <sec> <title>CONCLUSIONS</title> Glycemic differences appear to be more closely associated with how behaviors are coordinated—particularly insulin timing relative to meals—than with exercise or carbohydrate intake alone. These findings highlight the importance of incorporating behavioral context to explain heterogeneity in real-world diabetes management and support personalized, behavior-aware strategies. </sec>

  PublisherDOI

• Gated Adaptation for Continual Learning in Human Activity Recognition

  arXiv (Cornell University) · 2026-03-08

  preprintOpen accessSenior author

  Wearable sensors in Internet of Things (IoT) ecosystems increasingly support applications such as remote health monitoring, elderly care, and smart home automation, all of which rely on robust human activity recognition (HAR). Continual learning systems must balance plasticity (learning new tasks) with stability (retaining prior knowledge), yet AI models often exhibit catastrophic forgetting, where learning new tasks degrades performance on earlier ones. This challenge is especially acute in domain-incremental HAR, where on-device models must adapt to new subjects with distinct movement patterns while maintaining accuracy on prior subjects without transmitting sensitive data to the cloud. We propose a parameter-efficient continual learning framework based on channel-wise gated modulation of frozen pretrained representations. Our key insight is that adaptation should operate through feature selection rather than feature generation: by restricting learned transformations to diagonal scaling of existing features, we preserve the geometry of pretrained representations while enabling subject-specific modulation. We provide a theoretical analysis showing that gating implements a bounded diagonal operator that limits representational drift compared to unconstrained linear transformations. Empirically, freezing the backbone substantially reduces forgetting, and lightweight gates restore lost adaptation capacity, achieving stability and plasticity simultaneously. On PAMAP2 with 8 sequential subjects, our approach reduces forgetting from 39.7% to 16.2% and improves final accuracy from 56.7% to 77.7%, while training less than 2% of parameters. Our method matches or exceeds standard continual learning baselines without replay buffers or task-specific regularization, confirming that structured diagonal operators are effective and efficient under distribution shift.

  PublisherDOI

Frequent coauthors

• Roozbeh Jafari

  Massachusetts Institute of Technology

  83 shared

• Majid Sarrafzadeh

  71 shared

• Lorraine S. Evangelista

  University of Nevada, Las Vegas

  57 shared

• Jung‐Ah Lee

  University of California, Irvine

  53 shared

• Carol M. Mangione

  University of California, Los Angeles

  50 shared

• Alison Moore

  University of California, San Diego

  50 shared

• Marjan Motie

  University of California, Irvine

  49 shared

• Ramin Fallahzadeh

  Stanford University

  38 shared

Labs

• Embedded Machine Intelligence LabPI

  We research design, development, and validation of algorithms, tools, and technologies that enhance utilization and large-scale adoption of pervasive systems.

Education

• Ph.D., Computer Science

  University of California Los Angeles

• M.S., Computer Engineering

  University of Texas at Dallas

• B.S., Computer Engineering

  University of Tehran

• B.S.

  Sharif University of Technology

Awards & honors

• Best Poster Award, ASU College of Health Solutions Faculty R…
• Best Poster Award Runner Up, IEEE Body Sensor Networks (BSN)…
• Best Poster Award Honorable Mention, IEEE Biomedical & Healt…
• Nominee for Outstanding Teaching – Graduate, ASU College of…
• Best Poster Award, ASU College of Health Solutions Faculty R…