Tahera Hossain

Research Assistant Professor

Nagoya University, Japan

Email: taheramoni@gmail.com

tahera@ucl.nuee.nagoya-u.ac.jp and hossain.tahera.k9@f.mail.nagoya-u.ac.jp

Google Scholar

Twitter

Curriculum Vitae (CV)

ABOUT ME

My research goal is to create technologies that will automatically infer user goals, predict future user behaviors, describe common user behaviors, and guide users to follow their healthy routine life. Broadly, my research focuses on modeling and understanding human behaviors in the healthcare domain, and I research in the fields of Applied Machine Learning, Computational Modeling, and Human-Computer Interaction (HCI). Specifically, I model behaviors of people based on the data collected from people’s mobile, smart wearable devices, and their instrumented environments. One line of my research is analyzing how activities and behaviors of patients/elderly correlate with their disease symptoms and how technology can coach them to modify their complex routines to be productive, healthy, and safe. I also work on the design and development of robust human behavior recognition methods, which are able to detect people's behavior in real-world settings by combating the challenges faced in such settings. My research has been published in top-tier venues such as TOCHI and IMWUT UbiComp.

My current research looks at ways to understand and measure the health status and outcomes of individuals. I analyze real-world data collected from everyday situations, such as behavioral and physiological data, as well as nursing care and patient records, to gain insights into the subject matter. I use machine learning to understand patients/elderly people’s real-life complex activities and modeling of human ambulatory multimodal time-series data, including physiological, biological, and behavioral data for measuring, predicting, improving, and understanding human physiology and behavior for health, wellbeing, and performance.

RESEARCH HIGHLIGHTS (for a complete list see my Google Scholar)

Human Behavior Modeling and Computational Interaction

A Bayesian Approach for Quantifying Data Scarcity when Modeling Human Behavior via Inverse Reinforcement Learning

Tahera Hossain, Wanggang Shen, Anindya Antar, Snehal Prabhudesai, Sozo Inoue, Xun Huan, and Nikola Banovic

ACM Transactions on Computer-Human Interaction 30, 1, pp. 27 pages, 2023.

Publisher Talk

Abstract

Computational models that formalize complex human behaviors enable study and understanding of such behaviors. However, collecting behavior data required to estimate the parameters of such models is often tedious and resource intensive. Thus, estimating dataset size as part of data collection planning (also known as Sample Size Determination) is important to reduce the time and effort of behavior data collection while maintaining an accurate estimate of model parameters. In this article, we present a sample size determination method based on Uncertainty Quantification (UQ) for a specific Inverse Reinforcement Learning (IRL) model of human behavior, in two cases: (1) pre-hoc experiment design—conducted in the planning stage before any data is collected, to guide the estimation of how many samples to collect; and (2) post-hoc dataset analysis—performed after data is collected, to decide if the existing dataset has sufficient samples and whether more data is needed. We validate our approach in experiments with a realistic model of behaviors of people with Multiple Sclerosis (MS) and illustrate how to pick a reasonable sample size target. Our work enables model designers to perform a deeper, principled investigation of the effects of dataset size on IRL model parameters.

Interpretable Behavior Modeling for Mental Health and Well-Being

Office Workers Behaviors, Mental Health, and Productivity Prediction Using Multimodal Data

Yusuke Nishimura, Tahera Hossain, Akane Sano, Shota Isomura, Yutaka Arakawa, and Sozo Inoue

Activity and Behavior Computing (ABC), pp. 26 pages, Springer, 2021.

Publisher

Abstract

In recent years, many organizations have prioritized efforts to detect and treat mental health issues. In particular, office workers are affected by many stressors, and physical and mental exhaustion, which is also a social problem. To improve the psychological situation in the workplace, we need to clarify the cause. In this paper, we conducted a 14-day experiment to collect wristband sensor data as well as behavioral and psychological questionnaire data from about 100 office workers. We developed machine learning models to predict psychological indexes using the data. In addition, we analyzed the correlation between behavior (work content and work environment) and psychological state of office workers to reveal the relationship between their work content, work environment, and behavior. As a result, we showed that multiple psychological indicators of office workers can be predicted with more than 80% accuracy using wearable sensors, behavioral data, and weather data. Furthermore, we found that in the working environment, the time spent in “web conferencing”, “working at home (living room)”, and “break time (work time)’ had a significant effect on the psychological state of office workers.

Toward Human Thermal Comfort Sensing: New Dataset and Analysis of Heart Rate Variability (HRV) Under Different Activities

Tahera Hossain, Yusuke Kawasaki, Kazuki Honda, Kizito Nkurikiyeyezu, Guillaume Lopez

Activity and Behavior Computing (ABC), pp. 27 pages, Taylor & Francis, 2022.

Publisher

Abstract

Thermal comfort is the mental state of feeling comfortable and satisfied with one’s surrounding thermal environment and is essential for overall well-being. In this research, primarily physiological indicators such as heart rate, body movement, body temperature, and skin potential are measured utilizing a wearable terminal having a variety of sensors designed to collect sympathetic nervous system activity data in different thermal environments that can be encountered in daily life. We investigate not only the effect of the thermal environment on biological information but also the effect of the work when it overlaps with the mental or physical burden. During the experiment, we collected 33 participants’ 10 days of data in varying temperatures and humidity levels for different work conditions, i.e., reading, typewriting, and gymnastics activities data focusing on hot thermal conditions. We present a thermal comfort providing approach using heart rate variability (HRV) data. We conducted a comparative analysis of several machine learning algorithms including K-nearest neighbors (KNN), Extra Trees Classifier (ET), and LightGBM classifier, as well as a convolutional neural network (1-D CNN), to predict the subjective thermal sensation state based on individuals’ HRV indices. Out of all the machine learning models assessed, Extra Trees Classifier (ET) demonstrated the highest accuracy rate of 99%, while the convolutional neural network (1-D CNN) achieved an accuracy rate of 97.6%. This study aimed to compare the environmental thermal comfort and personal thermal assessment states and explore the potential use of low-granularity signals for thermal comfort provision models.

Automatic Nurse Care Record Creation with Activity Recognition for Proactive Care Management

Integrating Activity Recognition and Nursing Care Records: The System, Deployment, and a Verification Study

Sozo Inoue, Paula Lago, Tahera Hossain, Tittaya Mairittha, Nattaya Mairittha

In Proc of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), Volume 3, 86, pp. 24 pages, 2019.

Publisher

Abstract

In this paper, we introduce a system of integrating activity recognition and collecting nursing care records at nursing care facilities as well as activity labels and sensors through smartphones, and describe experiments at a nursing care facility for 4 months. A system designed to be used even by staff not familiar with smartphones could collected enough number of data without losing but improving their workload for recording. For collected data, we revealed the nature of the collected data as for activities, care details, and timestamps, and considering them, we show a reference accuracy of recognition of nursing activity which is durable to time skewness, overlaps, and class imbalances. Moreover, we demonstrate the near future prediction to predict the next day's activities from the previous day's records which could be useful for proactive care management. The dataset collected is to be opened to the research community, and can be the utilized for activity recognition and data mining in care facilities.

Featured

Presenting at the CHI2023

Delighted to give my first talk at CHI2023, representing the CompHCI Lab, University of Michigan. I talked about how to conduct sample size determination when modeling human behavior using Inverse Reinforcement Learning (IRL). ACM CHI Conference on Human Factors in Computing Systems is the premier international conference of Human-Computer Interaction (HCI) (CHI2023, Hamburg, Germany).

Representing CompHCI Lab (University of Michigan) at CHI2023

CompHCI lab at CHI2023 in one place, taking a break after the paper presentation.

Activity and Behavior Computing Conference (ABC 2022)

Delighted to receive the Excellent Paper Award in the Activity and Behavior Computing Conference, 2022, London, UK.

Best Ph.D. Forum Presentation Award at IEEE PerCom 2019

17th IEEE International Conference on Pervasive Computing and Communications (PerCom), Kyoto, Japan. Best Ph.D. Forum Presentation for 'Sensor-based Daily Activity Understanding in Caregiving Center'. PerCom is the premier annual scholarly venue in pervasive computing and communications.

Best Poster Paper Award at UbiComp, 2018

ACM International Conference on Pervasive and Ubiquitous Computing (Ubicomp’18), Singapore. UbiComp / ISWC is a premier interdisciplinary venue in which leading international researchers, designers, developers, and practitioners in the field present and discuss novel results in all aspects of ubiquitous, pervasive, and wearable computing.

Highlighted Research Collaborators

Get in touch at [taheramoni@gmail.com]

Page updated

Google Sites

Report abuse