Health Care Cost Reductions with Machine Learning-Directed Evaluations during Radiation Therapy - An Economic...

Area of Science:

• Oncology
• Health Services Research
• Medical Informatics

Background:

• Machine learning (ML) can optimize healthcare by identifying patients for preventative interventions.
• The SHIELD-RT study used ML to identify high-risk patients during radiation therapy (RT) for supplemental evaluations.
• This intervention successfully reduced acute care utilization.

Purpose of the Study:

• To conduct an economic analysis of the ML-directed intervention in the SHIELD-RT study.
• To determine the cost-effectiveness of using ML to target supplemental clinical evaluations for high-risk RT patients.

Main Methods:

• A post hoc economic analysis of the SHIELD-RT randomized controlled trial was performed.
• High-risk patients identified by ML were randomized to standard care or mandatory twice-weekly evaluations.
• Total medical costs, including acute care and intervention costs, were analyzed using negative binomial regression.

Main Results:

• The intervention group showed a reduction in acute care visits (0.31 vs. 0.47 per course, P=0.04).
• Total mean adjusted costs were significantly lower in the intervention group ($1494 vs. $3110 per course, P=0.03).
• Supplemental evaluation costs were $155 per course in the intervention group.

Conclusions:

• Mandatory supplemental evaluations for ML-identified high-risk patients reduced total medical costs and improved clinical outcomes.
• The ML-directed intervention demonstrated a cost-effective approach to managing high-risk patients during RT.
• Further research is needed to confirm the generalizability of these economic findings.