AI-ML Based Cyber-Resilience for Emergent Healthcare Systems
Location
Poster #12
Start Date
2-5-2025 12:00 PM
Department
Computer Science
Abstract
The proposed study researches cybersecurity in emergent healthcare by developing a Real-Time Adaptive Cyber Resilience Software System using hybrid AI models. There is a distinct lack of research focusing on adaptive, real-time resilience mechanisms that dynamically adjust to the evolving nature of emergent healthcare scenarios. This research will investigate how hybrid AI models, combining reinforcement learning (RL) with traditional anomaly detection and clustering techniques, can be used to enhance cyber resilience in highly dynamic and critical healthcare environments. The study will focus on designing a software framework that integrates reinforcement learning with existing anomaly detection (e.g., Isolation Forests) and clustering techniques (e.g., K-means). K-means clustering algorithm will be used to group data into distinct clusters based on similarities in network traffic patterns, system logs, and user behaviors. By segmenting data into clusters, the system can establish baseline profiles of normal system behavior. This allows for the identification of deviations from expected patterns, which could signal potential security threats or anomalies. Isolation Forests will be employed to detect anomalies in data that deviate significantly from the norm. This technique isolates observations by randomly selecting features and recursively partitioning the data. Anomalies are identified as data points that are isolated faster than others. This effective method helps in identifying outliers that could represent malicious activities or system comprises.The combination of Isolation Forests with K-means clustering and adapting reinforcement learning will enhance the system’s ability to detect both known and unknown threats.This hybrid approach aims to continuously learn and adapt to new cyber threats and evolving patterns in healthcare settings. The research is timely in light of recent cyberattacks on healthcare institutions, such as the January 2024 incident at Lurie Children's Hospital in Chicago, which disrupted critical systems and exposed data from over 775,000 individuals. Healthcare breaches provide cybercriminals access to valuable information, which can be exploited for ransom and gain media attention, furthering the hacker's notoriety. Using a comprehensive datasets, including historical cyber incident data, network traffic logs, and electronic health records, the research demonstrates the framework’s effectiveness in maintaining operational integrity and safeguarding patient data.
Faculty Sponsor
Yi Yang
AI-ML Based Cyber-Resilience for Emergent Healthcare Systems
Poster #12
The proposed study researches cybersecurity in emergent healthcare by developing a Real-Time Adaptive Cyber Resilience Software System using hybrid AI models. There is a distinct lack of research focusing on adaptive, real-time resilience mechanisms that dynamically adjust to the evolving nature of emergent healthcare scenarios. This research will investigate how hybrid AI models, combining reinforcement learning (RL) with traditional anomaly detection and clustering techniques, can be used to enhance cyber resilience in highly dynamic and critical healthcare environments. The study will focus on designing a software framework that integrates reinforcement learning with existing anomaly detection (e.g., Isolation Forests) and clustering techniques (e.g., K-means). K-means clustering algorithm will be used to group data into distinct clusters based on similarities in network traffic patterns, system logs, and user behaviors. By segmenting data into clusters, the system can establish baseline profiles of normal system behavior. This allows for the identification of deviations from expected patterns, which could signal potential security threats or anomalies. Isolation Forests will be employed to detect anomalies in data that deviate significantly from the norm. This technique isolates observations by randomly selecting features and recursively partitioning the data. Anomalies are identified as data points that are isolated faster than others. This effective method helps in identifying outliers that could represent malicious activities or system comprises.The combination of Isolation Forests with K-means clustering and adapting reinforcement learning will enhance the system’s ability to detect both known and unknown threats.This hybrid approach aims to continuously learn and adapt to new cyber threats and evolving patterns in healthcare settings. The research is timely in light of recent cyberattacks on healthcare institutions, such as the January 2024 incident at Lurie Children's Hospital in Chicago, which disrupted critical systems and exposed data from over 775,000 individuals. Healthcare breaches provide cybercriminals access to valuable information, which can be exploited for ransom and gain media attention, furthering the hacker's notoriety. Using a comprehensive datasets, including historical cyber incident data, network traffic logs, and electronic health records, the research demonstrates the framework’s effectiveness in maintaining operational integrity and safeguarding patient data.