**4.2 Predictive analytics of threat models**

Machine Learning (ML) assist in the mathematical models construction which has the ability to explain and showcase complicated behavior without the need for programming [50]. These techniques have a way of improving the HMI in the interface design and further improve the intelligence of the design. The usage of ML in Human Machine Interface/Interaction (HMI) design is not very trivial. Moustakis and Herrmann [66] affirmed that the misperceptions about ML, inadequate familiarity with ML's latent capacity and research shortage are the main sources for declined utilization of ML in HMI design. Though, as of today ML has gained recognitions and advance applications can enhance user capability in UI design which helps in mitigating against cyber security threats. Integration of models that use logs to reactively discriminate transactions based on user' history is essential for cyber security threat modeling.

The challenge usually face during threat analysis is now being fazed away by applying AI coupled with machine learning algorithms, which feeds on data to detect abnormalities in systems. Predictive analysis is AI-driven by data and uses large data to understand malicious activities, identify patterns, and provide insights into potential attacks much quicker [67]. Standardization does not exist, and thus the choice of threat models is deterministic of the project needs like targeted risk area, allocated time, expertise, and stakeholder's involvement [68]. Furthermore,

*Application of Artificial Intelligence in User Interfaces Design for Cyber Security Threat Modeling DOI: http://dx.doi.org/10.5772/intechopen.96534*

it is advantageous to apply threat models at the requirement and design stage of the project life-cycle [65] and the use of a well-formulated model for Software Development Life Cycle (SDLC) [69] for efficient threat modeling is desirous. Though AI sheds a positive light on cyber security, it also presents alarming intrusion possibilities for cyber criminals. This ordeal does not limit the impact of AI but rather re-enforces its significance, especially in cyber security threat modeling.
