目前的轉診程序被認為過於復雜且沒有效率，醫務人員通常需要很長時間才能將病人轉移到另一家醫院。因此，我們設計了一個智慧匹配系統來改善這種情況。 我們想要實現的主要目標有兩個。首先，我們希望減少轉診所需的時間並自動化一些原本需要人力的步驟。此系統內建置了一個應用程式，可以檢查所有醫院的資源，方便醫務人員完成轉診程序。其次，我們還希望建立一種機制來決定轉診目標。我們使用的匹配算法由信任度計算模組和機器學習模組組成，基於這兩個模組的結果生成推薦目標建議清單。本文主要介紹了智慧匹配系統及其中所使用的演算法。

Nowadays, the current referral procedure is considered to be too complicated and inefficient, medical staff often take a long time to transfer a patient to another hospital. Therefore, we design a smart matching system to improve this situation. There are two main goals that we want to achieve. First, we want to reduce the needed time and automate some steps that would otherwise require manpower. An application is built in this system, which can check all hospitals' resource and make it convenient for medical staff to complete the referral procedure. Second, we also want to establish a mechanism to determine the referral target. The matching algorithm we use is consist of trust value calculating model and machine learning model, generating a referral target suggestion list based on the results of these two models. This paper mainly introduce the smart matching system and the algorithm used in it.

摘要 i
Abstract ii
Contents iii
List of Figures v
List of Tables vi
1 Introduction 1
2 Related Work 3
3 System Architecture 5
3.1 Client layer 5
3.2 Application Server layer 5
3.3 Database layer 7
4 The Smart Matching Application 8
4.1 ArcGIS and Web Appbuilder for ArcGIS 8
4.2 Referral request form widget 10
4.3 Hospital information widget 10
4.4 Main map 11
5 The Smart Matching Algorithm 12
5.1 Filtering 12
5.2 Trust Value Calculating 14
5.2.1 Interaction Events 14
5.2.2 Interaction Value 17
5.2.3 Reliability of the Interaction Value 17
5.2.4 Reputation value 18
5.2.5 Distance Value 19
5.3 Machine Learning 20
6 Experiment Results 22
6.1 The Result of Trust Value Calculating 22
6.2 The Result of Machine Learning 25
7 Conclusion 28

[1] Michael A. Nielsen, ”Neural Networks and Deep Learning”, Determination Press,
2015
[2] Grus J. Data science from scratch. Sebastopol: O’Reilly; 2015.
[3] Method and system for data aggregation for real-time emergency resource management
[4] Method and a system for effecting transfer of a patient from a hospital via a computer
network
[5] Method and system for online secure patient referral system
[6] Patient referral and physician-to-physician marketing method and system
[7] ”Node.js,” [Online]. Available: https://nodejs.org/.
[8] ”MongoDB,” [Online]. Available: https://www.mongodb.com/.
[9] ”ArcGIS,” [Online]. Available: https://www.arcgis.com/.
[10] Huynh, T.D., Jennings, N. R. and Shadbolt, N. (2004) Developing an integrated trust
and reputation model for open multi-agent systems. At 7th International Workshop
on Trust in Agent Societies 7th International Workshop on Trust in Agent Societies,
United States.
[11] *Lin, Phone, Chung, P.-C., and Fang, Y. P2P-iSN: A Peer-to-Peer Architecture for
Heterogeneous Social Networks. IEEE Network Magazine, 28(1): 56-64, JanuaryFebruary
2014.