Technical note: Optimal allocation of limited proton therapy resources using model-based patient selection.

Area of Science:

• Radiation Oncology
• Medical Physics
• Operations Research

Background:

• Proton therapy offers advantages over photon therapy for certain cancers.
• Limited daily slots in proton therapy clinics necessitate careful patient selection.
• Maximizing patient benefit requires balancing slot availability with treatment efficacy.

Purpose of the Study:

• To develop an optimal patient selection strategy for proton therapy under limited resource constraints.
• To minimize expected normal tissue complications and maximize proton therapy benefits.
• To determine dynamic benefit thresholds for proton therapy allocation.

Main Methods:

• Extended normal tissue complication probability (NTCP) models to a limited resource scenario.
• Formulated patient selection as a Markov decision process (MDP).
• Utilized value-policy iteration to determine optimal benefit thresholds.

Main Results:

• Optimal benefit thresholds are influenced by slot availability, patient load, and benefit distribution.
• Thresholds dynamically adjust based on current facility utilization.
• Lower thresholds are optimal when slots become available sooner.

Conclusions:

• Markov decision process (MDP) methodology enhances NTCP-based selection for limited proton therapy slots.
• Optimal thresholds are dynamic and depend on real-time facility utilization.
• The proposed policy offers robustness against patient load variations with minimal nominal benefit loss.