gems

gems simulates disease progression and predicts outcomes using generalized event-based multistate models represented as directed acyclic graphs to evaluate interventions.

Key Features:

• Disease Progression Simulation: Represents progression through events such as diagnosis, treatment, and death within a directed acyclic graph (DAG) where vertices denote states and edges denote transitions.
• Generalized Multistate Model: Employs generalized multistate models articulated as a DAG with continuous, transition-specific hazard functions, avoiding constraints of traditional Markov models.
• Customizable Hazard Functions: Allows specification of arbitrary hazard functions and their parameters for individual transitions.
• Incorporation of Parameter Uncertainty: Accounts for parameter uncertainty within the modeling framework to reflect estimation variability.
• Non-Markovian Flexibility: Incorporates the history of previous events so transition behavior can depend on event history rather than only the current state.
• Broad Applicability: Applies to medical contexts and other domains where multistate simulation of event-driven processes is relevant.

Scientific Applications:

• Outcome Prediction: Predicts patient outcomes under different intervention scenarios by simulating multistate pathways and associated risks.
• Intervention Evaluation: Evaluates effects of health interventions by comparing simulated trajectories and outcome distributions across scenarios.
• Epidemiological Planning: Supports health planners and evaluators in assessing population-level impacts of interventions using simulation-based projections.

Methodology:

Uses multistate models represented as a DAG with continuous, transition-specific hazard functions, supports user-defined arbitrary hazard functions and parameters, incorporates parameter uncertainty, and permits non-Markovian dependence on event history.