Hyunju Kim

Dr. Hyunju Kim

orcid

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 March 2023
Verified
Conference Paper
Open access
TLTae-Hoon Lee
Tae-Hoon Lee
HKHyunju Kim
Hyunju Kim
DLDongman Lee
Dongman Lee
Human activity recognition (HAR) plays a key role in intelligent systems. Ambient sensors are utilized to avoid privacy concerns and to collect data streams in a less intrusive manner in smart homes. In scenarios requiring immediate intervention, the systems must perform HAR in real-time. Since it is difficult to segment the exact transition point in real-time, data unrelated to the target activity can appear at the beginning of the time series, which we call unrefined data. It leads us to a new challenge that the HAR model recognizes a user's activity as early as possible with unrefined data. To explore the impact of unrefined data on real-time HAR, we design an experimental system that consists of a Transformer-based filtering network and an LSTM-based early classifier. We evaluate the experimental system with 3 public datasets collected on testbeds with ambient sensors installed. Our results reveal that unrefined data degrade HAR performance in terms of accuracy and earliness, and the use of the filtering network that filters out unrefined data improves recognition performance.

21 May 2024 in Scientific Data
Verified
Journal Article
Open access
PDF file
HKHyunju Kim
Hyunju Kim
GKGeon Kim
Geon Kim
TLTaehoon Lee
Taehoon Lee
KKKisoo Kim
Kisoo Kim
DLDongman Lee
Dongman Lee
As IoT technology advances, using machine learning to detect user activities emerges as a promising strategy for delivering a variety of smart services. It is essential to have access to high-quality data that also respects privacy concerns and data streams from ambient sensors in the surrounding environment meet this requirement. However, despite growing interest in research, there is a noticeable lack of datasets from ambient sensors designed for public spaces, as opposed to those for private settings. To bridge this gap, we design the DOO-RE dataset within an actual meeting room environment, equipped with three types of ambient sensors: those triggered by actuators, users, and the environment itself. This dataset is compiled from the activities of over twenty students throughout a period of four months. DOO-RE provides reliable and purpose-oriented activity data in a public setting, with activity labels verified by multiple annotators through a process of cross-validation to guarantee data integrity. DOO-RE categorizes nine different types of activities and facilitates the study of both single and group activities. We are optimistic that DOO-RE will play a significant role in advancing human activity recognition technologies, enhancing smart automation systems, and enabling the rapid setup of smart spaces through ambient sensors.


23 May 2026 in Sensors
Verified
Journal Article
Open access
PDF file
YZYuezhou Zhang
Yuezhou Zhang
AFAmos Akinola Folarin
Amos Akinola Folarin
RZRongrong Zhong
Rongrong Zhong
HKHyunju Kim
Hyunju Kim
CSCallum Stewart
Callum Stewart
RDRichard J. B. Dobson
Richard J. B. Dobson
We characterized long-term, objectively measured physical activity trajectories surrounding total knee arthroplasty (TKA) and total hip arthroplasty (THA) and examined factors associated with wearable-derived physical activity recovery. In this observational study within the All of Us Research Program, linked electronic health records and Fitbit step count data spanning the two years before and the two years after surgery were analyzed using piecewise linear mixed-effects models to characterize preoperative and postoperative trajectories. Recovery of physical activity was defined relative to two preoperative baselines—activity measured immediately before surgery and a more remote baseline approximating longer-term habitual activity—and associated factors were examined using Cox proportional hazards models. Among 238 participants (147 TKA, 91 THA; mean age 64.9 [SD 8.3] years), both procedures showed progressive preoperative decline, with accelerated decline beginning earlier in TKA than in THA. Postoperative recovery followed a staged pattern, with rapid early improvement, slower intermediate gains, and later stabilization. Recovery to the immediate preoperative baseline occurred earlier than recovery to the remote baseline. Higher activity during the 4 weeks before surgery was associated with a greater likelihood of recovery to the remote baseline. These findings support long-term wearable monitoring as a complementary measure of physical activity recovery after arthroplasty.

9 February 2026 in JMIR Preprints
Verified
Preprint
Open access
YZYuezhou Zhang
Yuezhou Zhang
AFAmos Akinola Folarin
Amos Akinola Folarin
HEHugh Logan Ellis
Hugh Logan Ellis
RZRongrong Zhong
Rongrong Zhong
CSCallum Stewart
Callum Stewart
RDRichard Jb Dobson
Richard Jb Dobson
BACKGROUND Low physical activity is a well-established risk factor for major depressive disorder (MDD). However, how physical activity changes in the months preceding a first clinical diagnosis of MDD has not been well explored, particularly when assessed using long-term, objective measures collected in real-world settings. OBJECTIVE This study aimed to characterize trajectories of wearable-measured physical activity during the year preceding incident MDD diagnosis. METHODS We conducted a retrospective nested case-control study using linked electronic health record (EHR) and consumer wearable data (Fitbit) from the All of Us Research Program. Adults with at least 6 months of valid Fitbit physical activity data in the 12 months preceding diagnosis were eligible. Incident MDD cases were identified based on a first recorded EHR diagnosis and matched to MDD-free controls on age, sex, body mass index, and time of diagnosis, with up to four controls per case. Daily step counts and moderate-to-vigorous physical activity (MVPA) were aggregated into monthly averages. Linear mixed-effects models were used to compare pre-diagnostic physical activity trajectories between cases and controls over a retrospective time scale from −12 to 0 months. Among cases, within-person contrasts were used to identify when physical activity levels first showed statistically significant deviations relative to levels observed 12 months before diagnosis. Exploratory analyses examined whether pre-diagnostic trajectories differed by sex, age, and body mass index. RESULTS The analytic cohort included 4,104 participants (829 incident MDD cases and 3,275 matched controls; 81.7% women; median age 48.4 years). Compared with controls, individuals who developed MDD exhibited consistently lower overall physical activity levels and significant downward trajectories in both daily step counts and MVPA during the year preceding diagnosis (global trajectory tests, P < .001 for both outcomes). Among cases, statistically significant changes in daily step counts emerged approximately 4 months before diagnosis, whereas significant changes in MVPA emerged approximately 5 months before diagnosis. Furthermore, exploratory analyses suggested heterogeneity in pre-diagnostic trajectories across demographic subgroups, including steeper declines among men, more pronounced reductions in activity intensity at older ages, and persistently lower activity levels with flatter trajectories among individuals with obesity. CONCLUSIONS In this large cohort with linked wearable and clinical data, sustained within-person declines in physical activity emerged several months before incident MDD diagnosis. These findings suggest that longitudinal monitoring of physical activity using consumer wearable devices may provide clinically relevant early signals to support risk stratification, targeted prevention, and earlier intervention for MDD.

4 May 2026 in Journal of Medical Internet Research
Verified
Journal Article
Open access
YZYuezhou Zhang
Yuezhou Zhang
AFAmos Akinola Folarin
Amos Akinola Folarin
HEHugh Logan Ellis
Hugh Logan Ellis
RZRongrong Zhong
Rongrong Zhong
CSCallum Stewart
Callum Stewart
RDRichard Jb Dobson
Richard Jb Dobson
Background: Low physical activity (PA) is a well-established risk factor for major depressive disorder (MDD). However, the temporal dynamics of PA preceding an incident clinical diagnosis of MDD remain poorly characterized, particularly using long-term, objective measures collected in real-world settings. Objective: This study aimed to characterize trajectories of wearable-measured PA during the year preceding incident MDD diagnosis and identify the timing of within-person changes. Methods: We conducted a retrospective nested case-control study using linked electronic health record and wearable (Fitbit) data from the All of Us Research Program. Adults with at least 6 months of valid Fitbit PA data in the 12 months preceding diagnosis were included. Incident MDD cases were identified based on a first electronic health record–recorded diagnosis and matched to MDD-free controls on age, sex, BMI, and calendar time of diagnosis, with up to 4 controls per case. Daily steps and moderate to vigorous PA (MVPA) were aggregated into monthly averages. Linear mixed-effects models were used to compare prediagnostic PA trajectories between cases and controls over a retrospective time scale from −12 to 0 months. Among cases, within-person contrasts were used to identify when PA levels first showed statistically significant deviations relative to levels observed 12 months before diagnosis. Exploratory analyses assessed heterogeneity by demographic factors. Results: The analytic cohort included 4104 participants (n=829, 20.2% incident MDD cases and n=3275, 79.8% matched controls; n=3355, 81.7% women; median age 48.4, IQR 36.3-61.3 years). Compared with controls, individuals who developed MDD exhibited consistently lower overall PA and significant downward trajectories in both daily steps and MVPA during the year preceding diagnosis (global trajectory tests; P <.001 for both outcomes). Differences widened progressively over time, indicating accelerating declines as diagnosis approached. Among cases, statistically significant changes in daily step counts emerged approximately 4 months before diagnosis (−145, 95% CI −253 to −37 steps vs month −12; P =.02) and reached −428 (95% CI −531 to −326) steps at diagnosis ( P <.001). Declines in MVPA emerged approximately 5 months before diagnosis (−2.48, 95% CI −4.32 to −0.64 minutes; P =.02) and reached −5.61 (95% CI −7.35 to −3.86) minutes at diagnosis ( P <.001). Furthermore, exploratory analyses suggested heterogeneity in prediagnostic trajectories across demographic subgroups, including steeper declines among men, more pronounced reductions in activity intensity at older ages, and persistently lower activity levels with flatter trajectories among individuals with obesity. Conclusions: Unlike prior studies lacking objective PA assessment before MDD diagnosis, this study linked wearable and clinical data to characterize long-term prediagnostic trajectories in real-world settings. We observed sustained within-person declines emerging 4 to 5 months before diagnosis, providing insights into temporal dynamics preceding clinical recognition. These findings suggest that wearable-based monitoring may offer scalable early signals for risk stratification, prevention, and intervention for MDD.