Jai-Chang Park

Mr. Jai-Chang Park

Angel Robotics
orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

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

Representative Publications

View All Publications
24 February 2021 in Applied Sciences
Verified
Journal Article
Open access
PDF file
JPJai-Chang Park
Jai-Chang Park
SKSeongbeom Kim
Seongbeom Kim
JLJeHoon Lee
JeHoon Lee
Diabetes mellitus is a severe chronic disease, and the number of patients has increased. To manage blood glucose levels, patients should frequently measure their blood glucose and analyze which lifestyle habits affect blood glucose levels. However, it is hard to record and analyze the relationship between their blood glucose levels and lifestyle. The internet of things (IoT) is useful to interconnect, monitor, obtain, and process data between various devices used in everyday life to fulfill a common objective. This paper proposes an intelligent self-care platform using IoT technology that helps patients with chronic diabetes manage their blood glucose levels in their target range. In particular, we developed various devices called the self-care IoT pack. It consists of five different types of devices to obtain blood glucose levels, physical activities, food intake, medication, sleeping, and so on. They can collect blood glucose levels with lifestyles that automatically impact the patient’s blood glucose level. We also devised a self-care application to display and analyze the data obtained from the IoT pack. Consequently, the proposed self-care IoT platform collects the blood glucose levels and the lifestyles without any burden of record. By reviewing the accumulated information, the patients can find bad habits in blood glucose management and improve their lifestyle.

29 July 2021 in Electronics
Verified
Journal Article
Open access
PDF file
JLJeHoon Lee
JeHoon Lee
JPJai-Chang Park
Jai-Chang Park
SKSeongbeom Kim
Seongbeom Kim
Exercise enables continuous glycemic control for diabetic patients, and it is effective in preventing diabetic complications and maintaining emotional stability. However, it is difficult for diabetic patients to know the appropriate intensity and duration of exercise. Excessive exercise causes sudden hypoglycemia, and patients avoid therapeutic exercise or perform it conservatively owing to the repeated hypoglycemia symptoms. In this paper, we propose a new therapeutic exercise platform that supports type 2 diabetes patients to exercise regularly according to the exercise prescription received from the hospital. The proposed platform includes the following three significant contributions. First, we develop a hardware platform that automatically tracks and records all aerobic exercise performed by a patient indoors or outdoors using a wearable band and aerobic exercise equipment. Second, we devise a patient-specific exercise stress test to know whether the patient is exercising according to his or her usual exercise regimen. Finally, we develop a mobile application that informs patients in real-time whether they are exercising appropriately for their exercise regimen each time they exercise. For platform evaluation and future improvement, we received satisfaction ratings and functional improvements through a questionnaire survey on 10 type 2 diabetes patients and 10 persons without a diabetes diagnosis who had used the proposed platform for more than 3 months. Most users were (1) satisfied with automatic exercise recording, and (2) exercise time increased. Diabetics reported that their fasting blood glucose was dropped, and they were more motivated to exercise. These results prove that exercise must be combined with medication for blood glucose management in chronic diabetic patients. The proposed platform can be helpful for patients to continue their daily exercise according to their exercise prescription.

25 February 2026 in TechRxiv
Verified
Preprint
Open access
PDF file
JPJai-Chang Park
Jai-Chang Park
This paper presents flutter_kiwi_nlp, a production-oriented Flutter plugin for Korean morphological analysis built on Kiwi. The package exposes one stable Dart API while internally operating two runtime stacks: Native (Dart FFI + C bridge + Kiwi shared library) and Web (Dart JS interop + kiwi-nlp WASM). This design enables a single application codebase across Android, iOS, macOS, Linux, Windows, and Web. Unlike ONNX-export deployment paths, this integration does not require adding an extra generic inference runtime layer because Kiwi already provides the analysis engine and model format used here. The implementation is explicitly aligned with on-device AI requirements: local inference execution, reduced text egress by default, predictable latency without per-request network dependence, and operational fallback controls for model provisioning. Empirically, repeated benchmark trials show a warm-path throughput reversal (steady-state path, initialization excluded): Flutter warm throughput can exceed the Python-native baseline: 3.12x on desktop, 3.87x on iOS simulator (debug), and 1.52x on Android emulator (release). Added Linux/Windows CI rows show clear execution-mode sensitivity: single mode is 0.64x of Python throughput on both OSes, while batch mode is 1.54-1.55x faster than Python on the same runners. The updated benchmark pipeline also adds boundary-decomposed measurements (pure processing vs full JSON path), quantifying serialization/parsing overhead at 13.89-47.93% depending on runtime profile. This warm-path reversal does not imply end-to-end session superiority: cold-start initialization and short-session effective-throughput remain weaker than the Python baseline on all measured platforms. Cross-runtime linguistic agreement remains close on gold corpora (88.39% vs 88.58% token agreement; 84.90% vs 85.55% POS agreement). On native targets, inference can run fully offline after one-time model provisioning. This paper provides an implementation-complete, reproducible specification of API contract, runtime parity rules, build automation hooks, security boundaries, failure taxonomy, benchmark protocol, and quantified ecosystem survey signals.

25 February 2026 in TechRxiv
Verified
Preprint
Open access
PDF file
JPJai-Chang Park
Jai-Chang Park
This report presents kiwi-rs, a Rust library that exposes the public Kiwi C API through a high-level and safety-oriented interface for Korean NLP. The implementation combines dynamic symbol loading, ownership-safe handle wrappers, runtime capability checks, and autobootstrap for library/model assets, and it includes an agent-skill package for reliable LLMassisted onboarding. We evaluate kiwi-rs against kiwipiepy under matched conditions using repeated-input (warm-cache) and varied-input (cache-active, mixed-reuse) workloads with bootstrap confidence intervals and sink-parity checks. On a dataset of 192 Korean texts across 8 categories, kiwi-rs shows measurable speedups on selected varied-input features (e.g., join: 3.85×, tokenize_many_batch: 23.18×, split_into_sents_with_tokens: 67.75×) while other paths remain close to parity or statistically inconclusive under conservative decision rules. Repeatedinput numbers are reported as warm-cache capacity bounds only. We additionally report cacheminimized stress results and path-pinned reproducibility runs to bound interpretation of headline ratios.