DOI QR코드

DOI QR Code

A Personalized Dietary Coaching Method Using Food Clustering Analysis

음식 군집분석을 통한 개인맞춤형 식이 코칭 기법

  • 오유리 (숙명여자대학교 컴퓨터과학부) ;
  • 최지은 (숙명여자대학교 컴퓨터과학부) ;
  • 김윤희 (숙명여자대학교 컴퓨터과학부)
  • Received : 2016.01.05
  • Accepted : 2016.02.22
  • Published : 2016.06.30

Abstract

In recent times, as most people develop keen interest in health management, the importance of cultivating dietary habits to prevent various chronic diseases is emphasized. Subsequently, dietary management systems using a variety of mobile and web application interfaces have emerged. However, these systems are difficult to apply in real world and also do not provide personalized information reflective of the user's situation. Hence it is necessary to develop a personalized dietary management and recommendation method that considers user's body state information, food analysis and other essential statistics. In this paper, we analyze nutrition using self-organizing map (SOM) and prepare data about nutrition using clustering. We provide a substitute food recommendation method and also give feedback about the food that user wants to eat based on personalized criteria. The experiment results show that the distance between input food and recommended food of the proposed method is short compared to the recommended food results using general methods and proved that nutritional similar food is recommended.

현대인의 건강관리에 대한 관심이 증가하고 다양한 만성질환을 야기하는 식습관에 대한 중요성이 강조되고 있는 상황이다. 이에 따라 여러가지 모바일 및 웹시스템을 이용한 식단 관리 방법이 등장하고 있지만 이는 실제로 적용하기 어렵고 사용자의 상황을 반영하는 맞춤형 정보를 제공하지 않는다. 따라서 개인의 신체정보 및 상황을 반영하고 음식을 분석하여 실질적으로 사용자가 섭취 가능한 맞춤형 식단관리 및 추천 방법이 필요하다. 본 논문에서는 자기조직화지도를 이용하여 음식을 분석하고 이를 군집화하여 음식에 대한 데이터를 준비한다. 그리고 사용자의 신체정보 및 상황을 고려한 개인 맞춤형 기준을 반영하여 섭취하고 싶은 음식에 대한 피드백 및 대체음식 추천방법을 제안한다. 또한 실험을 통하여 일반적인 방법을 이용한 추천된 음식결과와 비교하여 제안된 방법의 입력 음식과 추천 음식의 거리가 짧다는 것을 통하여 영양적으로 유사한 음식이 추천됨을 증명하였다.

Keywords

References

  1. J. E. Yu and Y. J. Kim, "Healthy Eating Index (HEI) Indicates that High Diet Quality is Associated with Low Prevalence of Hypertension and Type 2 Diabetes in Koreans: the Korean Genome and Epidemiology Study (KoGES)," Public Health Weekly Report, Vol.8, No.3, pp.51-58, 2015. (in Korean).
  2. Jongsu Lee, "A Study on Application Service for Dietary Habit Improvement of Patients with Chronic Disease - Focused on Patients with Hypertension and Diabetes -," Journal of Korea Society of Design Forum, Vol.48, pp.71-82, 2015. (in Korean).
  3. Hae-Ok Jeon and Ok-Soo Kim, "The effects of an internet based coaching program for obesity management in hypertensive patients," Korean Journal of Adult Nursing, Vol.23, No.2, pp.146-159, 2011.
  4. B. K. Lim, et al., "DietAdvisor: A Personalized eHealth Agent in Mobile Computing Environment," Conference on KIISE, Vol.38, No.2(D), pp.115-118, 2011.
  5. Phanich, Maiyaporn, Phathrajarin Pholkul, and Suphakant Phimoltares, "Food recommendation system using clustering analysis for diabetic patients," Information Science and Applications (ICISA), 2010 International Conference on. IEEE, 2010.
  6. Woan-Tyng Lin, et al., "FML-Based Recommender System for Restaurants," Technologies and Applications of Artificial Intelligence (TAAI), 2013 Conference on. IEEE, 2013.
  7. Food Nutrients Database [Internet], http://www.foodnara.go.kr/kisna/index.do.
  8. The Korean Nutrition Society, Dietary Reference Intakes for Koreans, revised Version, The Korean Nutrition Society, Korea, 2010.
  9. Ron Wehrens and MC Buydens Lutgarde, "Self-and super-organizing maps in R: the Kohonen package," Journal of Statistical Software, Vol.21, No.5, pp.1-19, 2007.
  10. D. L. Davies and D. W. Bouldin, "A cluster separation measure," IEEE Trans. Pattern Anal. Machine Intell., Vol. 1, No.4, pp.224-227, 1979.
  11. Elbow Method [Internet], https://en.wikipedia.org/wiki/Determining_the_number_of_clusters_in_a_data_set.