JSJohn A. Stankovic

John A. Stankovic

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


John A. Stankovic has a total of 176 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Medical Cyber-Physical Systems

    by Insup Lee, Homa Alemzadeh, Rahul Mangharam, John Stankovic, James Weimer

    Medical Cyber-Physical Systems (MCPS) are closed-loop systems that integrate sensing, computation, communication, and control with clinicians, patients, users, and/or care workflows to support health monitoring and intervention. They span regulated clinical platforms and consumer-facing health technologies, combining Internet of Medical Things (IoMT) connectivity, intelligent algorithms, and human physiological dynamics with varying safety, security, and efficacy constraints. This paper presents a twenty-year review of MCPS research and practice, covering community evolution, foundational methods, devices and datasets, and representative deployments across hospital, home, and wellness contexts. We synthesize key advances in modeling, verification, interoperability, data quality, and learning-enabled decision support, and we examine translational case studies that expose the gap between laboratory prototypes and real-world adoption. We conclude with open research directions for scalable engineering methods, trustworthy data and evaluation infrastructure, real-time integration, and sustained post-deployment assurance to advance MCPS toward safe, secure, effective, and broadly deployable systems that improve health outcomes.

  • Open Access

    The Landscape of Cyber-Physical Systems Education

    by John A. Stankovic, Bhaskar Krishnamachari, André Platzer, Janos Sztipanovits

    The rapid rise of cyber-physical systems (CPS) has created a significant need for new classes and curricula that teach the foundations, principles, and expertise needed to develop CPS and conduct CPS research. Universities and faculty have responded overwhelmingly at both the undergraduate and graduate levels. Examples of major worldwide accomplishments of the CPS education field are described in this paper. One of the first major accomplishments is the results of a three-year study performed by the National Academy of Engineering. Highlights of the study include foundations, principles, skills required for CPS, and example classes and curricula. Some universities have implemented an entire curriculum and others multiple classes. The paper presents examples of both situations. These examples provide details on class objectives, content, tools, testbeds, class projects, challenges, lessons learned, successes, and next steps. The presence of textbooks is a sign of an emerging and stabilizing discipline. A brief survey is given of CPS textbooks that treat fundamentals or that focus on a particular application domain or key CPS topic. Finally, a brief discussion on the future of CPS education is presented.

  • Open Access

    Can (Should) We Incorporate Common Sense into AI/ML Solutions?

    by J. A. Stankovic

    Artificial Intelligence (AI) and Machine Learning (ML) have been addressing how to incorporate common sense into solutions for many decades. While some progress has been made, due to the incredibly complex issues involved, we are far from a general solution. This paper describes personal opinions about common sense from two perspectives: is it even feasible, and if so, should or how can such solutions be applied? A proposed solution direction with three examples, first responders, smart health, and smart cities is presented. Promising research directions and key open questions are also discussed.

  • Open Access

    Feasibility and Acceptability of the Smarthealth Intervention for Dementia Caregivers. A Qualitative Analysis of a Single‐Group Pilot Study

    by Eunjung Ko, Karen M. Rose, Kristina C. Gordon, Sooyoung Kim, Ye Gao, Peng Wang, Lahiru Wijayasingha, Hongning Wang, John A. Stankovic, Kathy D. Wright

    Aim(s): To explore the feasibility and acceptability of acoustic monitoring and real‐time recommendations for stress detection and management (i.e., smarthealth intervention). Design: This qualitative study used a framework of acceptability for healthcare interventions. Methods: From January 2021 to December 2023 in the U.S.A., we interviewed 10 family caregivers who had completed the 4‐month smarthealth intervention. The caregivers shared their user experiences and feedback on the system's feasibility and acceptability. Data were analysed using abductive thematic analysis, incorporating the framework of acceptability for healthcare interventions and the collected data. Results: Seven themes and 19 categories emerged: attitudes, burden, ethicality, intervention adherence, intervention coherence, perceived effectiveness and suggestions. Feedback on the smarthealth intervention was mixed. Some found it beneficial, citing accuracy, ease of use and increased awareness. However, others felt burdened during its use, primarily due to time constraints. Conclusion: The smarthealth intervention can potentially improve caregivers' awareness of themselves and caregiving situations. Implications for the Profession and/or Patient Care: Future directions should involve adapting the smarthealth intervention to consider diverse caregiving scenarios and incorporating a larger sample of caregivers. Impact: This is the first study to offer a voice detection system and real‐time stress management recommendations to caregivers of people living with dementia. An individualised approach should be considered to improve the system's effectiveness. This includes providing personalised intervention components, considering caregivers' time and establishing a user‐friendly system with high accessibility. The findings can be a cornerstone for smarthealth interventions influencing dementia caregivers' self‐care and emotional regulation. Reporting Method: Standards for Reporting Qualitative Research. Patient or Public Contribution: Members of the public and service users from a memory clinic and social media platforms contributed to the study by reviewing recruitment materials. Trial Registration: This trial's study protocol was registered with ClinicalTrials.gov (ID No. NCT04536701) on 3 September 2020 ( https://classic.clinicaltrials.gov/ct2/show/NCT04536701 )

  • Open Access

    Recruitment Challenges and Strategies in a Technology-Based Intervention for Dementia Caregivers: Descriptive Study

    by Eunjung Ko, Ye Gao, Peng Wang, Lahiru Wijayasingha, Kathy D Wright, Kristina C Gordon, Hongning Wang, John A Stankovic, Karen M Rose

    Background: Researchers have encountered challenges in recruiting unpaid caregivers of people living with Alzheimer disease and related dementias for intervention studies. However, little is known about the reasons for nonparticipation in in-home smart health interventions in community-based settings. Objective: This study aimed to (1) assess recruitment rates in a smart health technology intervention for caregivers of people living with Alzheimer disease and related dementias and reasons for nonparticipation among them and (2) discuss lessons learned from recruitment challenges and strategies to improve recruitment. Methods: The smart health intervention was a 4-month, single-arm trial designed to evaluate an in-home, technology-based intervention that monitors stressful moments for caregiving dyads through acoustic signals and to provide the caregivers with real-time stress management strategies. The recruitment involved two main methods: on-site engagement by a recruiter from a memory clinic and social media advertising. Caregivers were screened for eligibility by phone between January 2021 and September 2023. The recruitment rates, reasons for nonparticipation, and participant demographics were analyzed using descriptive statistics. Results: Of 201 caregivers contacted, 11 were enrolled in this study. Eighty-two caregivers did not return the screening call, and others did not participate due to privacy concerns (n=30), lack of interest (n=29), and burdensome study procedures (n=26). Our recruitment strategies included addressing privacy concerns, visualizing collected data through a dashboard, boosting social media presence, increasing the recruitment budget, updating advertisements, and preparing and deploying additional study devices. Conclusions: This study highlighted barriers to participation in the smart health intervention. Despite several recruitment strategies, enrollment rates remained below expectations. These findings underscore the need for future research to explore alternative methods for increasing the recruitment of informal dementia caregivers in technology-based intervention studies.


  • Open Access

    Real-Time Multimodal Cognitive Assistant for Emergency Medical Services

    by Keshara Weerasinghe, Saahith Janapati, Xueren Ge, Sion Kim, Sneha Iyer, John A. Stankovic, Homa Alemzadeh

    Emergency Medical Services (EMS) responders often operate under time-sensitive conditions, facing cognitive overload and inherent risks, requiring essential skills in critical thinking and rapid decision-making. This paper presents CognitiveEMS, an end-to-end wearable cognitive assistant system that can act as a collaborative virtual partner engaging in the real-time acquisition and analysis of multimodal data from an emergency scene and interacting with EMS responders through Augmented Reality (AR) smart glasses. CognitiveEMS processes the continuous streams of data in real-time and leverages edge computing to provide assistance in EMS protocol selection and intervention recognition. We address key technical challenges in real-time cognitive assistance by introducing three novel components: (i) a Speech Recognition model that is fine-tuned for real-world medical emergency conversations using simulated EMS audio recordings, augmented with synthetic data generated by large language models (LLMs); (ii) an EMS Protocol Prediction model that combines state-of-the-art (SOTA) tiny language models with EMS domain knowledge using graph-based attention mechanisms; (iii) an EMS Action Recognition module which leverages multimodal audio and video data and protocol predictions to infer the intervention/treatment actions taken by the responders at the incident scene. Our results show that for speech recognition we achieve superior performance compared to SOTA (WER of 0.290 vs. 0.618) on conversational data. Our protocol prediction component also significantly outperforms SOTA (top-3 accuracy of 0.800 vs. 0.200) and the action recognition achieves an accuracy of 0.727, while maintaining an end-to-end latency of 3.78s for protocol prediction on the edge and 0.31s on the server.


  • Open Access

    ViObject

    by Wenqiang Chen, Shupei Lin, Zhencan Peng, Farshid Salemi Parizi, Seongkook Heo, Shwetak Patel, Wojciech Matusik, Wei Zhao, John Stankovic

    Knowing the object grabbed by a hand can offer essential contextual information for interaction between the human and the physical world. This paper presents a novel system, ViObject, for passive object recognition that uses accelerometer and gyroscope sensor data from commodity smartwatches to identify untagged everyday objects. The system relies on the vibrations caused by grabbing objects and does not require additional hardware or human effort. ViObject's ability to recognize objects passively can have important implications for a wide range of applications, from smart home automation to healthcare and assistive technologies. In this paper, we present the design and implementation of ViObject, to address challenges such as motion interference, different object-touching positions, different grasp speeds/pressure, and model customization to new users and new objects. We evaluate the system's performance using a dataset of 20 objects from 20 participants and show that ViObject achieves an average accuracy of 86.4%. We also customize models for new users and new objects, achieving an average accuracy of 90.1%. Overall, ViObject demonstrates a novel technology concept of passive object recognition using commodity smartwatches and opens up new avenues for research and innovation in this area.


  • Open Access

    Camera-Independent Single Image Depth Estimation from Defocus Blur

    by Lahiru Wijayasingha, Homa Alemzadeh, John A. Stankovic

    Monocular depth estimation is an important step in many downstream tasks in machine vision. We address the topic of estimating monocular depth from defocus blur which can yield more accurate results than the semantic based depth estimation methods. The existing monocular depth from defocus techniques are sensitive to the particular camera that the images are taken from. We show how several camera-related parameters affect the defocus blur using optical physics equations and how they make the defocus blur depend on these parameters. The simple correction procedure we propose can alleviate this problem which does not require any retraining of the original model. We created a synthetic dataset which can be used to test the camera independent performance of depth from defocus blur models. We evaluate our model on both synthetic and real datasets (DDFF12 and NYU depth V2) obtained with different cameras and show that our methods are significantly more robust to the changes of cameras. Code: https://github.com/sleekEagle/defocus_camind.git

  • Open Access

    Robust Finger Interactions with COTS Smartwatches via Unsupervised Siamese Adaptation

    by Wenqiang Chen, Ziqi Wang, Pengrui Quan, Zhencan Peng, Shupei Lin, Mani Srivastava, Wojciech Matusik, John Stankovic

    Wearable devices like smartwatches and smart wristbands have gained substantial popularity in recent years. However, their small interfaces create inconvenience and limit computing functionality. To fill this gap, we propose ViWatch, which enables robust finger interactions under deployment variations, and relies on a single IMU sensor that is ubiquitous in COTS smartwatches. To this end, we design an unsupervised Siamese adversarial learning method. We built a real-time system on commodity smartwatches and tested it with over one hundred volunteers. Results show that the system accuracy is about 97% over a week. In addition, it is resistant to deployment variations such as different hand shapes, finger activity strengths, and smartwatch positions on the wrist. We also developed a number of mobile applications using our interactive system and conducted a user study where all participants preferred our unsupervised approach to supervised calibration. The demonstration of ViWatch is shown at https://youtu.be/N5-ggvy2qfI.