Multi-factor authentication is a widely used approach to prevent impersonation and account theft, particularly in mobile applications, surpassing standard passwords. However, explicit multi-factor authentication, especially second-factor authentication, requires additional costs in terms of user effort. Behavioural biometrics techniques offer new ways of verifying users, including multi-credential feature authentication. However, multi-feature behavioural biometrics integrated with multi-factor authentication processes increase the memory and processing power needed for implementation. A promising solution can be achieved by efficiently integrating machine learning with application access patterns. This paper presents a model that uses machine learning to statistically analyze application access time behaviour to reduce the need for second-factor authentication in subsequent access attempts. Additionally, this model has been analytically evaluated using a dataset collected from thirty-five users, and the findings demonstrate high accuracy. Based on these results, the presented approach can adapt to changing conditions to enhance authentication systems and improve user convenience.