Title Page
Contents
ABSTRACT 12
국문초록 15
Chapter 1. Introduction 17
1.1. Research Background 17
1.1.1. Applying IoT, AI, and Mobile Terminal to Medical Information Platform 20
1.1.2. Authentication of Medical Field 23
1.2. Research Motivation 25
1.3. Contribution 27
1.4. Approach Direction of The Research 28
1.5. Research Scope and Organization 30
1.5.1. Research Scope 30
1.5.2. Organization 30
Chapter 2. Related Works 31
2.1. System Authentication in Medical Field 31
2.1.1. Authentication and Authorization 31
2.1.2. Rule-Based Access Control Model 34
2.1.3. System Authentication in Medical Field 37
2.2. Bigdata Processing and AI in Medical Information Platform 38
2.3. Edge Computing in Medical Field 41
2.3.1. Cloud Computing and Edge Computing 41
2.3.2. Research on Edge Computing in Medical Field 46
Chapter 3. Design of Edge Computing-Supported Medical Information Platform for Executing Automatic Authentication Based on Patient's Emergency Situations Applying Deep Learning 48
3.1. Architecture of The AutoAuthMIP 48
3.1.1. The Smart Space Device Part 50
3.1.2. Edge Computing Part 50
3.1.3. Medical Information and Management Part 51
3.1.4. Mobile Terminal Part 53
3.2. Data Flow of The AutoAuthMIP 53
3.2.1. Platform Scenario 53
3.2.2. Patient Emergency Situation Decision 57
3.3. The Automatic Authentication Based on Patient's Situations 60
3.3.1. Motivation 60
3.3.2. Procedure for The Automatic Authentication 61
3.3.3. DB Scheme for The AutoAuthMIP 62
3.3.4. Patient's Medical Information and Personal Information Classification According to Sensitivity 66
3.3.5. Setting Access Right According to Patient's Situations and The Role of Medical Staff 67
3.4. Summary 68
Chapter 4. Implement of AutoAuthMIP 69
4.1. Implementation of AutoAuthMIP's Parts 69
4.1.1. Implementation Hardware Description 69
4.2. Determining Patient Emergency Situations Using AI Models and Sending Emergency Message 72
4.2.1. Machine Learning Models and NHIS Dataset for Determining Underlying Disease 73
4.2.2. Deep Learning Model and MIT-BIH Dataset for Arrhythmia Detection 77
4.2.3. Interactive Process of Edge Computing and Medical Information Server for Patient-Customized Emergency Situation Decision 83
4.2.4. Implementation of Determining Emergency Situation 85
4.2.5. Implementation of Medical Information Platform Procedures in Emergency Situation 89
4.3. Implementation of The Automatic Authentication 92
4.3.1. Implementation of The Automatic Authentication Procedure 92
4.3.2. Access to Patient Information According to Access Right 94
4.4. Summary 97
Chapter 5. Executability Evaluation of The AutoAuthMIP 99
5.1. Evaluation of The Automatic Authentication 99
5.1.1. Structural Evaluation Between Existing Authentication and the Automatic Authentication 99
5.1.2. Evaluation of Execution Time of Automatic Authentication 102
5.2. Structural Evaluation of The Edge Computing 104
5.2.1. Comparison of Transmission Delay Between Two Platforms 104
5.2.2. The System-Wide Comparison for Determining Emergency Situations Between Two Platforms 106
5.3. Summary 108
Chapter 6. Conclusion and Prospects 109
6.1. Conclusion 109
6.2. Prospects 110
Reference 112
Table 3-1. Overview of The AutoAuthMIP Command Set 56
Table 3-2. Patient Emergency Conditions and Decision Model Reference 59
Table 3-3. An Example Graded Medical and Personal Information of Patient 66
Table 3-4. An Example of Access Rights by Patient's Situation and The Role of Medical Staff 67
Table 4-1. Details of Blood Pressure and Blood Sugar Data from NHIS 74
Table 4-2. On-Hot Encoding of Sex and DIS Attribute 74
Table 4-3. The Training Accuracy, Test Accuracy, and Test Data Execution Time of Machine Learning Models for The Determination of The Patient's... 76
Table 4-4. Five Classifications For MIT-BIH Arrhythmia Data Label Based on AAMI EC57 79
Table 4-5. Precision, Recall, F1-Score of The Arrhythmia Classification Model 83
Table 5-1. Structural Comparison of Authentication Studies 101
Table 5-2. System-wide Comparison Between Existing Platform and AutoAuthMIP for Emergency Situation Determination 107
Figure 1-1. Architecture of General Medical Information Platform 18
Figure 1-2. Growth Trend of The Healthcare Market 19
Figure 1-3. Medical Information Platform Applied with IoT, AI, and Terminal 21
Figure 1-4. The Average Total Cost of A Data Breach by Industry From IBM's "Cost of A Data Breach Report 2022" 24
Figure 1-5. The Edge Computing Infrastructure 26
Figure 1-6. Approach Direction of This Research 29
Figure 2-1. The Three Elements of Authentication: The Subject, Factors, and System 32
Figure 2-2. Three Authentication factors: Knowledge-based, Ownership-based, Biometric-based 33
Figure 2-3. Rule-based Access Control Model 36
Figure 2-4. Worldwide Data Generation Trend by IDC 42
Figure 2-5. Existing Cloud Computing Model 43
Figure 2-6. Edge Computing Model 44
Figure 3-1. The Architecture of The AutoAuthMIP 49
Figure 3-2. The Flow Chart for The Whole Scenario of AutoAuthMIP 55
Figure 3-3. National Early Warning Score (NEWS) 58
Figure 3-4. Database Logical E-R Diagram for Medical Information and Management Parts and Edge Computing Parts 64
Figure 4-1. The Overall Implementation for The AutoAuthMIP 69
Figure 4-2. BMS-AE-DK and S-Patch ECG Sensors in Smart Space Device Part 70
Figure 4-3. The BMS-AE-DK Measuring SpO2 and NIBP 71
Figure 4-4. Raspberry Pi 4 8 GB Units and Sensor Hat 71
Figure 4-5. Smartphone and Tablet Device 72
Figure 4-6. The Actual Value of The NHIS Data 75
Figure 4-7. Illustration of ECG Signal and A Sample Labeled R from The MIT-BIH Database 77
Figure 4-8. MIT-BIH Data: Horizontal Line is one signal, Vertical Line is One Frame 78
Figure 4-9. The Architecture of The 1-D CNN Model 80
Figure 4-10. Pytorch Code for Implementation of 1D-CNN 81
Figure 4-11. The Interactive Processing Between The Edge Node and The Medical Information Server 84
Figure 4-12. Real-Time SpO2 Value Measured by BMS 86
Figure 4-13. Sending Biometric Data from The Smart Space Device Part to The Edge Node 86
Figure 4-14. Emergency Situation Decision Code According to Sensor Type in Edge Node 88
Figure 4-15. Mobile Terminal Registration Process and Emergency Message Transmission Using Google FCM 90
Figure 4-16. The Process of Emergency Message Transmission 91
Figure 4-17. Login Screen in Normal Situation (Left) and Emergency Situations (Right) 92
Figure 4-18. The Implementation Details of Checking Emergency Message Confirmation, Attempting Automatic Authentication, and Accessing Patient... 93
Figure 4-19. The Terminal Log of Automatic Authentication Process in The Authentication Server 94
Figure 4-20. Accessing Medical Information According to The Patient's Situation and The Role of Medical Staff 96
Figure 5-1. Proceed of Manual Authentication and Automatic Authentication 102
Figure 5-2. Test Bed for Comparison of Transmission Delay of Two Platforms. (a) Existing Medical Information Platform without Edge Computing;... 105
Figure 5-3. Dotted Line Graph as A Result of Measuring Biometric Data Measurement Time t1 and Execution Time of Emergency Situation Decision t2 10 Times in The... 107