Browsing by Author "Chin, Chu-Yu"

Now showing 1 - 3 of 3

eDRAM: Effective early disease risk assessment with matrix factorization on a large-scale medical database: A case study on rheumatoid arthritis
(2018-11-26)
Chin, Chu-Yu
;
Hsieh, Sun-Yuan
;
Tseng, Vincent S.
;
資訊科學與工程研究所
;
Institute of Computer Science and Engineering
Recently, a number of analytical approaches for probing medical databases have been developed to assist in disease risk assessment and to determine the association of a clinical condition with others, so that better and intelligent healthcare can be provided. The early assessment of disease risk is an emerging topic in medical informatics. If diseases are detected at an early stage, prognosis can be improved and medical resources can be used more efficiently. For example, if rheumatoid arthritis (RA) is detected at an early stage, appropriate medications can be used to prevent bone deterioration. In early disease risk assessment, finding important risk factors from large-scale medical databases and performing individual disease risk assessment have been challenging tasks. A number of recent studies have considered risk factor analysis approaches, such as association rule mining, sequential rule mining, regression, and expert advice. In this study, to improve disease risk assessment, machine learning and matrix factorization techniques were integrated to discover important and implicit risk factors. A novel framework is proposed that can effectively assess early disease risks, and RA is used as a case study. This framework comprises three main stages: data preprocessing, risk factor optimization, and early disease risk assessment. This is the first study integrating matrix factorization and machine learning for disease risk assessment that is applied to a nation-wide and longitudinal medical diagnostic database. In the experimental evaluations, a cohort established from a large-scale medical database was used that included 1007 RA-diagnosed patients and 921,192 control patients examined over a nine-year follow-up period (2000-2008). The evaluation results demonstrate that the proposed approach is more efficient and stable for disease risk assessment than state-of-the-art methods.
Effective Risk Assessment of Type 2 Diabetes Using Diagnostic Information Retrieval
(2018-01-01)
Chin, Chu-Yu
;
Hsieh, Sun-Yuan
;
Tseng, Vincent S.
;
交大名義發表
;
資訊工程學系
Data in electronic medical records (EMRs) have been widely employed owing to rapid advances in disease assessment technologies. Accordingly, the challenging issue of how to effectively retrieve meaningful data from large-scale medical databases for disease assessment has risen. Furthermore, the manner in which early disease risk assessment models can detect disease symptoms is an issue of concern because early detection leads to early treatment. In this paper, with the aim of detecting diseases sooner and more effectively, a novel early disease risk assessment method is proposed, and type 2 diabetes mellitus (T2DM) is used as a case study. The proposed method is to improve the quality and meaning of diagnostic data using novel features and early strategy. To apply EMRs to construct a relationship matrix between patients and diseases, a retrieval method for generalized diagnostic coded information with extracted occurrence numbers was proposed. In order to identify diseases earlier, a disease risk assessment strategy from 7, 60, and 120 days before the onset of T2DM was established. The experimental results showed that the proposed method to improve disease risk assessment achieved high accuracy in terms of AUC-ROC and AUC-PR values. These results also demonstrate that the EMR information retrieval methods play an important role for disease assessment, and assessments can be performed at an earlier stage based on large-scale diagnostic databases.
Music Recommendation Based on Information of User Profiles, Music Genres and User Ratings
(2018-01-01)
Su, Ja-Hwung
;
Chin, Chu-Yu
;
Yang, Hsiao-Chuan
;
Tseng, Vincent S.
;
Hsieh, Sun-Yuan
Music data has been becoming bigger and bigger in recent years. It makes online music stores hard to provide the users with good personalized services. Therefore, a number of past studies were proposed for effectively retrieving the user preferences on music. However, they countered problems such as new user, new item and rating sparsity. To cope with these problems, in this paper, we propose a creative method that integrates information of user profiles, music genres and user ratings. In terms of solving problem of new user, the user similarities can be calculated by the profiles instead of ratings. By the user similarities, the unknown ratings can be predicted using user-based Collaborative Filtering. In terms of solving problem of rating sparsity, the unknown ratings are initialized by ratings of music genres. Even facing new music items, the rating data will not be sparse due to imputing the initialized ratings. Because the rating data is enriched, the user preference can be retrieved by item-based Collaborative Filtering. The experimental results reveal that, our proposed method performs more promising than the compared methods in terms of Root Mean Squared Error.