arXiv subjects

Vassilis Kostakos

Publications and source records attributed to Vassilis Kostakos.

At least 19 recordsLinked to original sources

arXiv2026

VitalAgent: A Tool-Augmented Agent for Reactive and Proactive Physiological Monitoring over Wearable Health Data

Wearable devices enable continuous monitoring of physiological signals such as ECG and PPG, but existing mHealth systems are largely limited to task-specific prediction pipelines or reactive question answering over static summaries. They lack the ability to support temporal reasoning, persistent physiological context, and proactive monitoring over long-term signal streams. We propose VitalAgent, a tool-augmented agentic framework for ECG/PPG-based mHealth that supports both reactive question answering and proactive monitoring. VitalAgent is built on a longitudinal physiological memory and a tool-augmented reasoning interface that enables dynamic computation over raw signals. We further introduce VitalBench, a longitudinal physiological monitoring benchmark dataset comprising 1,862 QA pairs for reactive question answering and 90.2 hours of continuous ECG/PPG recordings for proactive monitoring, covering cardiac, physical activity, and stress-related tasks. Experiments demonstrate that VitalAgent achieves over 25% improvement over prompt-based and ReAct baselines in reactive evaluation and supports proactive alert monitoring over long-term physiological signals, highlighting the importance of dynamic tool use and long-term physiological monitoring.

cs.AI↗

arXiv2026

Do LLMs Need to See Everything? A Benchmark and Study of Failures in LLM-driven Smartphone Automation using Screentext vs. Screenshots

With the rapid advancement of large language models (LLMs), mobile agents have emerged as promising tools for phone automation, simulating human interactions on screens to accomplish complex tasks. However, these agents often suffer from low accuracy, misinterpretation of user instructions, and failure on challenging tasks, with limited prior work examining why and where they fail. To address this, we introduce DailyDroid, a benchmark of 75 tasks in five scenarios across 25 Android apps, spanning three difficulty levels to mimic everyday smartphone use. We evaluate it using text-only and multimodal (text + screenshot) inputs on GPT-4o and o4-mini across 300 trials, revealing comparable performance with multimodal inputs yielding marginally higher success rates. Through in-depth failure analysis, we compile a handbook of common failures. Our findings reveal critical issues in UI accessibility, input modalities, and LLM/app design, offering implications for future mobile agents, applications, and UI development.

cs.HC↗

arXiv2025

Menta: A Small Language Model for On-Device Mental Health Prediction

Mental health conditions affect hundreds of millions globally, yet early detection remains limited. While large language models (LLMs) have shown promise in mental health applications, their size and computational demands hinder practical deployment. Small language models (SLMs) offer a lightweight alternative, but their use for social media--based mental health prediction remains largely underexplored. In this study, we introduce Menta, the first optimized SLM fine-tuned specifically for multi-task mental health prediction from social media data. Menta is jointly trained across six classification tasks using a LoRA-based framework, a cross-dataset strategy, and a balanced accuracy--oriented loss. Evaluated against nine state-of-the-art SLM baselines, Menta achieves an average improvement of 15.2\% across tasks covering depression, stress, and suicidality compared with the best-performing non--fine-tuned SLMs. It also achieves higher accuracy on depression and stress classification tasks compared to 13B-parameter LLMs, while being approximately 3.25x smaller. Moreover, we demonstrate real-time, on-device deployment of Menta on an iPhone 15 Pro Max, requiring only approximately 3GB RAM. Supported by a comprehensive benchmark against existing SLMs and LLMs, Menta highlights the potential for scalable, privacy-preserving mental health monitoring. Code is available at: https://hong-labs.github.io/menta-project/

cs.AI↗

arXiv2025

HealthSLM-Bench: Benchmarking Small Language Models for Mobile and Wearable Healthcare Monitoring

Mobile and wearable healthcare monitoring play a vital role in facilitating timely interventions, managing chronic health conditions, and ultimately improving individuals' quality of life. Previous studies on large language models (LLMs) have highlighted their impressive generalization abilities and effectiveness in healthcare prediction tasks. However, most LLM-based healthcare solutions are cloud-based, which raises significant privacy concerns and results in increased memory usage and latency. To address these challenges, there is growing interest in compact models, Small Language Models (SLMs), which are lightweight and designed to run locally and efficiently on mobile and wearable devices. Nevertheless, how well these models perform in healthcare prediction remains largely unexplored. We systematically evaluated SLMs on health prediction tasks using zero-shot, few-shot, and instruction fine-tuning approaches, and deployed the best performing fine-tuned SLMs on mobile devices to evaluate their real-world efficiency and predictive performance in practical healthcare scenarios. Our results show that SLMs can achieve performance comparable to LLMs while offering substantial gains in efficiency and privacy. However, challenges remain, particularly in handling class imbalance and few-shot scenarios. These findings highlight SLMs, though imperfect in their current form, as a promising solution for next-generation, privacy-preserving healthcare monitoring.

cs.AI↗

arXiv2025

Beyond Scale: Small Language Models are Comparable to GPT-4 in Mental Health Understanding

The emergence of Small Language Models (SLMs) as privacy-preserving alternatives for sensitive applications raises a fundamental question about their inherent understanding capabilities compared to Large Language Models (LLMs). This paper investigates the mental health understanding capabilities of current SLMs through systematic evaluation across diverse classification tasks. Employing zero-shot and few-shot learning paradigms, we benchmark their performance against established LLM baselines to elucidate their relative strengths and limitations in this critical domain. We assess five state-of-the-art SLMs (Phi-3, Phi-3.5, Qwen2.5, Llama-3.2, Gemma2) against three LLMs (GPT-4, FLAN-T5-XXL, Alpaca-7B) on six mental health understanding tasks. Our findings reveal that SLMs achieve mean performance within 2\% of LLMs on binary classification tasks (F1 scores of 0.64 vs 0.66 in zero-shot settings), demonstrating notable competence despite orders of magnitude fewer parameters. Both model categories experience similar degradation on multi-class severity tasks (a drop of over 30\%), suggesting that nuanced clinical understanding challenges transcend model scale. Few-shot prompting provides substantial improvements for SLMs (up to 14.6\%), while LLM gains are more variable. Our work highlights the potential of SLMs in mental health understanding, showing they can be effective privacy-preserving tools for analyzing sensitive online text data. In particular, their ability to quickly adapt and specialize with minimal data through few-shot learning positions them as promising candidates for scalable mental health screening tools.

cs.CL↗

arXiv2025

Real-Time Detection of Robot Failures Using Gaze Dynamics in Collaborative Tasks

Detecting robot failures during collaborative tasks is crucial for maintaining trust in human-robot interactions. This study investigates user gaze behaviour as an indicator of robot failures, utilising machine learning models to distinguish between non-failure and two types of failures: executional and decisional. Eye-tracking data were collected from 26 participants collaborating with a robot on Tangram puzzle-solving tasks. Gaze metrics, such as average gaze shift rates and the probability of gazing at specific areas of interest, were used to train machine learning classifiers, including Random Forest, AdaBoost, XGBoost, SVM, and CatBoost. The results show that Random Forest achieved 90% accuracy for detecting executional failures and 80% for decisional failures using the first 5 seconds of failure data. Real-time failure detection was evaluated by segmenting gaze data into intervals of 3, 5, and 10 seconds. These findings highlight the potential of gaze dynamics for real-time error detection in human-robot collaboration.

cs.HC↗

arXiv2025

Gazing at Failure: Investigating Human Gaze in Response to Robot Failure in Collaborative Tasks

Robots are prone to making errors, which can negatively impact their credibility as teammates during collaborative tasks with human users. Detecting and recovering from these failures is crucial for maintaining effective level of trust from users. However, robots may fail without being aware of it. One way to detect such failures could be by analysing humans' non-verbal behaviours and reactions to failures. This study investigates how human gaze dynamics can signal a robot's failure and examines how different types of failures affect people's perception of robot. We conducted a user study with 27 participants collaborating with a robotic mobile manipulator to solve tangram puzzles. The robot was programmed to experience two types of failures -- executional and decisional -- occurring either at the beginning or end of the task, with or without acknowledgement of the failure. Our findings reveal that the type and timing of the robot's failure significantly affect participants' gaze behaviour and perception of the robot. Specifically, executional failures led to more gaze shifts and increased focus on the robot, while decisional failures resulted in lower entropy in gaze transitions among areas of interest, particularly when the failure occurred at the end of the task. These results highlight that gaze can serve as a reliable indicator of robot failures and their types, and could also be used to predict the appropriate recovery actions.

cs.RO↗

arXiv2024

Efficient and Personalized Mobile Health Event Prediction via Small Language Models

Healthcare monitoring is crucial for early detection, timely intervention, and the ongoing management of health conditions, ultimately improving individuals' quality of life. Recent research shows that Large Language Models (LLMs) have demonstrated impressive performance in supporting healthcare tasks. However, existing LLM-based healthcare solutions typically rely on cloud-based systems, which raise privacy concerns and increase the risk of personal information leakage. As a result, there is growing interest in running these models locally on devices like mobile phones and wearables to protect users' privacy. Small Language Models (SLMs) are potential candidates to solve privacy and computational issues, as they are more efficient and better suited for local deployment. However, the performance of SLMs in healthcare domains has not yet been investigated. This paper examines the capability of SLMs to accurately analyze health data, such as steps, calories, sleep minutes, and other vital statistics, to assess an individual's health status. Our results show that, TinyLlama, which has 1.1 billion parameters, utilizes 4.31 GB memory, and has 0.48s latency, showing the best performance compared other four state-of-the-art (SOTA) SLMs on various healthcare applications. Our results indicate that SLMs could potentially be deployed on wearable or mobile devices for real-time health monitoring, providing a practical solution for efficient and privacy-preserving healthcare.

cs.CL↗

arXiv2024

AutoJournaling: A Context-Aware Journaling System Leveraging MLLMs on Smartphone Screenshots

Journaling offers significant benefits, including fostering self-reflection, enhancing writing skills, and aiding in mood monitoring. However, many people abandon the practice because traditional journaling is time-consuming, and detailed life events may be overlooked if not recorded promptly. Given that smartphones are the most widely used devices for entertainment, work, and socialization, they present an ideal platform for innovative approaches to journaling. Despite their ubiquity, the potential of using digital phenotyping, a method of unobtrusively collecting data from digital devices to gain insights into psychological and behavioral patterns, for automated journal generation has been largely underexplored. In this study, we propose AutoJournaling, the first-of-its-kind system that automatically generates journals by collecting and analyzing screenshots from smartphones. This system captures life events and corresponding emotions, offering a novel approach to digital phenotyping. We evaluated AutoJournaling by collecting screenshots every 3 seconds from three students over five days, demonstrating its feasibility and accuracy. AutoJournaling is the first framework to utilize seamlessly collected screenshots for journal generation, providing new insights into psychological states through digital phenotyping.

cs.HC↗

arXiv2024

ScreenTK: Seamless Detection of Time-Killing Moments Using Continuous Mobile Screen Text and On-Device LLMs

Smartphones have become essential to people's digital lives, providing a continuous stream of information and connectivity. However, this constant flow can lead to moments where users are simply passing time rather than engaging meaningfully. This underscores the importance of developing methods to identify these "time-killing" moments, enabling the delivery of important notifications in a way that minimizes interruptions and enhances user engagement. Recent work has utilized screenshots taken every 5 seconds to detect time-killing activities on smartphones. However, this method often misses to capture phone usage between intervals. We demonstrate that up to 50% of time-killing instances go undetected using screenshots, leading to substantial gaps in understanding user behavior. To address this limitation, we propose a method called ScreenTK that detects time-killing moments by leveraging continuous screen text monitoring and on-device large language models (LLMs). Screen text contains more comprehensive information than screenshots and allows LLMs to summarize detailed phone usage. To verify our framework, we conducted experiments with six participants, capturing 1,034 records of different time-killing moments. Initial results show that our framework outperforms state-of-the-art solutions by 38% in our case study.

cs.HC↗

arXiv2024

Predicting Affective States from Screen Text Sentiment

The proliferation of mobile sensing technologies has enabled the study of various physiological and behavioural phenomena through unobtrusive data collection from smartphone sensors. This approach offers real-time insights into individuals' physical and mental states, creating opportunities for personalised treatment and interventions. However, the potential of analysing the textual content viewed on smartphones to predict affective states remains underexplored. To better understand how the screen text that users are exposed to and interact with can influence their affects, we investigated a subset of data obtained from a digital phenotyping study of Australian university students conducted in 2023. We employed linear regression, zero-shot, and multi-shot prompting using a large language model (LLM) to analyse relationships between screen text and affective states. Our findings indicate that multi-shot prompting substantially outperforms both linear regression and zero-shot prompting, highlighting the importance of context in affect prediction. We discuss the value of incorporating textual and sentiment data for improving affect prediction, providing a basis for future advancements in understanding smartphone use and wellbeing.

cs.HC↗

arXiv2024

Enabling On-Device LLMs Personalization with Smartphone Sensing

This demo presents a novel end-to-end framework that combines on-device large language models (LLMs) with smartphone sensing technologies to achieve context-aware and personalized services. The framework addresses critical limitations of current personalization solutions via cloud LLMs, such as privacy concerns, latency and cost, and limited personal information. To achieve this, we innovatively proposed deploying LLMs on smartphones with multimodal sensor data through context-aware sensing and customized prompt engineering, ensuring privacy and enhancing personalization performance. A case study involving a university student demonstrated the capability of the framework to provide tailored recommendations. In addition, we show that the framework achieves the best trade-off in privacy, performance, latency, cost, battery and energy consumption between on-device and cloud LLMs. To the best of our knowledge, this is the first framework to provide on-device LLMs personalization with smartphone sensing. Future work will incorporate more diverse sensor data and involve extensive user studies to enhance personalization. Our proposed framework has the potential to substantially improve user experiences across domains including healthcare, productivity, and entertainment.

cs.HC↗

arXiv2023

Safeguarding Crowdsourcing Surveys from ChatGPT with Prompt Injection

ChatGPT and other large language models (LLMs) have proven useful in crowdsourcing tasks, where they can effectively annotate machine learning training data. However, this means that they also have the potential for misuse, specifically to automatically answer surveys. LLMs can potentially circumvent quality assurance measures, thereby threatening the integrity of methodologies that rely on crowdsourcing surveys. In this paper, we propose a mechanism to detect LLM-generated responses to surveys. The mechanism uses "prompt injection", such as directions that can mislead LLMs into giving predictable responses. We evaluate our technique against a range of question scenarios, types, and positions, and find that it can reliably detect LLM-generated responses with more than 93% effectiveness. We also provide an open-source software to help survey designers use our technique to detect LLM responses. Our work is a step in ensuring that survey methodologies remain rigorous vis-a-vis LLMs.

cs.HC↗

arXiv2023

Quality of Service (QoS)-driven Edge Computing and Smart Hospitals: A Vision, Architectural Elements, and Future Directions

The Internet of Things (IoT) paradigm is drastically changing our world by making everyday objects an integral part of the Internet. This transformation is increasingly being adopted in the healthcare sector, where Smart Hospitals are now relying on IoT technologies to track staff, patients, devices, and equipment, both within a hospital and beyond. This paradigm opens the door to new innovations for creating novel types of interactions among objects, services, and people in smarter ways to enhance the quality of patient services and the efficient utilisation of resources. However, the realisation of real-time IoT applications in healthcare and, ultimately, the development of Smart Hospitals are constrained by their current Cloud-based computing environment. Edge computing emerged as a new computing model that harnesses edge-based resources alongside Clouds for real-time IoT applications. It helps to capitalise on the potential economic impact of the IoT paradigm of $11 trillion per year, with a trillion IoT devices deployed by 2025 to sense, manage and monitor the hospital systems in real-time. This vision paper proposes new algorithms and software systems to tackle important challenges in Edge computing-enabled Smart Hospitals, including how to manage and execute diverse real-time IoT applications and how to meet their diverse and strict Quality of Service (QoS) requirements in hospital settings. The vision we outline can help tackle timely challenges that hospitals increasingly face.

cs.DC↗

arXiv2023

MicroFog: A Framework for Scalable Placement of Microservices-based IoT Applications in Federated Fog Environments

MicroService Architecture (MSA) is gaining rapid popularity for developing large-scale IoT applications for deployment within distributed and resource-constrained Fog computing environments. As a cloud-native application architecture, the true power of microservices comes from their loosely coupled, independently deployable and scalable nature, enabling distributed placement and dynamic composition across federated Fog and Cloud clusters. Thus, it is necessary to develop novel microservice placement algorithms that utilise these microservice characteristics to improve the performance of the applications. However, existing Fog computing frameworks lack support for integrating such placement policies due to their shortcomings in multiple areas, including MSA application placement and deployment across multi-fog multi-cloud environments, dynamic microservice composition across multiple distributed clusters, scalability of the framework, support for deploying heterogeneous microservice applications, etc. To this end, we design and implement MicroFog, a Fog computing framework providing a scalable, easy-to-configure control engine that executes placement algorithms and deploys applications across federated Fog environments. Furthermore, MicroFog provides a sufficient abstraction over container orchestration and dynamic microservice composition. The framework is evaluated using multiple use cases. The results demonstrate that MicroFog is a scalable, extensible and easy-to-configure framework that can integrate and evaluate novel placement policies for deploying microservice-based applications within multi-fog multi-cloud environments. We integrate multiple microservice placement policies to demonstrate MicroFog's ability to support horizontally scaled placement, thus reducing the application service response time up to 54%.

cs.DC↗

arXiv2023

Placement of Microservices-based IoT Applications in Fog Computing: A Taxonomy and Future Directions

The Fog computing paradigm utilises distributed, heterogeneous and resource-constrained devices at the edge of the network for efficient deployment of latency-critical and bandwidth-hungry IoT application services. Moreover, MicroService Architecture (MSA) is increasingly adopted to keep up with the rapid development and deployment needs of fast-evolving IoT applications. Due to the fine-grained modularity of the microservices and their independently deployable and scalable nature, MSA exhibits great potential in harnessing Fog and Cloud resources, thus giving rise to novel paradigms like Osmotic computing. The loosely coupled nature of the microservices, aided by the container orchestrators and service mesh technologies, enables the dynamic composition of distributed and scalable microservices to achieve diverse performance requirements of the IoT applications using distributed Fog resources. To this end, efficient placement of microservice plays a vital role, and scalable placement algorithms are required to utilise the said characteristics of the MSA while overcoming novel challenges introduced by the architecture. Thus, we present a comprehensive taxonomy of recent literature on microservices-based IoT applications placement within Fog computing environments. Furthermore, we organise multiple taxonomies to capture the main aspects of the placement problem, analyse and classify related works, identify research gaps within each category, and discuss future research directions.

cs.DC↗

arXiv2023

Using Thermal Imaging to Measure Hand Hygiene Quality

Hand hygiene has long been promoted as the most effective way to prevent the transmission of infection. However, due to the low compliance and quality of hand hygiene reported in previous studies, constant monitoring of healthcare workers' hand hygiene compliance and quality is crucial. In this study, we investigate the feasibility of using a thermal camera together with an RGB camera to detect hand coverage of alcohol-based formulation, thereby monitoring handrub quality. The system yields promising results in terms of accuracy (93.5%) and Dice coefficient (87.1%) when observations take place 10 seconds after performing handrub. In addition, we also examine the system performance change over a 60-second observation period, and the accuracy and Dice coefficient still remain at about 92.4% and 85.7% when observation happens at the 60-second time point. Given these encouraging results, thermal imaging shows its potential feasibility in providing accurate, constant, and systematic hand hygiene quality monitoring.

cs.HC↗

arXiv2022

DDoD: Dual Denial of Decision Attacks on Human-AI Teams

Artificial Intelligence (AI) systems have been increasingly used to make decision-making processes faster, more accurate, and more efficient. However, such systems are also at constant risk of being attacked. While the majority of attacks targeting AI-based applications aim to manipulate classifiers or training data and alter the output of an AI model, recently proposed Sponge Attacks against AI models aim to impede the classifier's execution by consuming substantial resources. In this work, we propose \textit{Dual Denial of Decision (DDoD) attacks against collaborative Human-AI teams}. We discuss how such attacks aim to deplete \textit{both computational and human} resources, and significantly impair decision-making capabilities. We describe DDoD on human and computational resources and present potential risk scenarios in a series of exemplary domains.

cs.HC↗