Authors:
Masatoshi Nagata
;
Kazunori Matsumoto
and
Masayuki Hashimoto
Affiliation:
KDDI R&D Labs, Japan
Keyword(s):
Sequential Latent Dirichlet Allocation, LDA, Sequential LDA, Lifestyle-related Disease, Medical Cost.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Biomedical Engineering
;
Business Analytics
;
Data Engineering
;
Data Mining
;
Databases and Information Systems Integration
;
Datamining
;
Enterprise Information Systems
;
Health Information Systems
;
Pattern Recognition and Machine Learning
;
Sensor Networks
;
Signal Processing
;
Soft Computing
Abstract:
Foreseeing the medical expenditure is beneficial for both insurance companies and individuals. In this paper we propose a new methodology to predict disease risk and medical cost. Based on sequential latent dirichlet allocation (SeqLDA), which classifies hierarchical sequential data into segments of topics, we tried to predict the number of people with diseases and the one-year cost of lifestyle-related diseases. Using the health checkup information and medical claims of 6500 people for three years, we achieved that prediction error was less than conventional LDA, and for accuracy rate, AUC was more than 0.71. The results suggest that the SeqLDA method serve to predict the number of people with diseases and the related medical costs using time series healthcare data.