생체 정보 측정을 통한 사용자 맞춤형 척추 마사지 기기의 개발 = Development of a Personalized Spinal Massage Device Based on Biometric Data Measurement

다국어 초록 (Multilingual Abstract)

As of 2019, the average life expectancy in Korea reached 83.3 years, exceeding the OECD average for both men and women. With increasing longevity, interest in improving and maintaining health-related quality of life has grown, with particular emphasis on pain as a key indicator. Manual therapy is widely recognized as an effective intervention for spinal dysfunction, as it relaxes fascia, reduces muscle tension, and corrects joint alignment. These changes in joint mobility activate peripheral analgesic mechanisms, promote recovery through stretching, and enhance the expression of mediators that alleviate pain. Acupressure further contributes to functional recovery by improving the circulation of qi and blood, thereby restoring muscle contraction and strength. As muscle strength increases, improvements in pain and motor function typically follow.

However, both manual therapy and acupressure have notable limitations: their effectiveness depends heavily on the practitioner’s skill and physical capacity, and consistent reproducibility is difficult to achieve across repeated treatments. Mechanical massage devices have emerged as alternatives capable of delivering standardized acupressure-like stimulation; yet, most rely on ceramic-roller rotation mechanisms, which pose risks of entrapment and difficulty in delivering localized pressure.

To address these shortcomings, we developed a system that applies localized, stepwise vertical upward pressure. Unlike fixed vertical-pressure systems reported in prior studies, the proposed system adaptively adjusts both height and width based on body-pressure mapping acquired while the user is lying down. Using 2,736 pressure-distribution datasets from normal and spinal-abnormal groups, the system demonstrated high classification performance (accuracy 97.36%, precision 97.80%, sensitivity 94.86%, specificity 98.79%), enabling customized massage sequencing based on user condition. Posture-estimation error remained within ±0.1009°, supporting its applicability to acupressure-bar control algorithms. Weight and pulse measurements showed errors within ±1.9 kg and ±1.6 bpm, respectively, indicating robust capability for continuous physiological monitoring.

Overall, these findings demonstrate the feasibility and practical potential of an adaptive, user-responsive vertical-pressure massage system. Nonetheless, as current results are derived from a limited participant pool, broader field testing and expanded data acquisition are necessary to further validate reliability prior to commercial deployment.

목차 (Table of Contents)

• I. 서론 1
• 1.1 연구 배경 1
• Ⅱ. 본론 5
• 2.1 연구의 방법 5
• 2.1.1 체압 및 체중 측정 모듈의 제작 7
• 2.1.2 맥박 측정 모듈의 제작 9
• 2.1.3 AI 기반 자세 분석 알고리즘 개발 11
• 2.1.4 하드웨어 및 소프트웨어 설계 및 제작 21
• 2.1.5 침대 본체의 제작 24
• Ⅲ. 결과 26
• Ⅳ. 결론 32
• 참 고 문 헌 33
• Abstract 36