HORIZON-HLTH-2026-01-CARE-03 Contribution

Our proposed approach focuses on developing advanced models to support evidence-based decision-making and reduce low-value care across various healthcare conditions. The solution consists of two complementary components:

1. Anomaly Detection Model
   Analyze existing historical treatment data to identify anomalies—potential indicators of low-value care such as overuse, misuse, or unwarranted variations. This step ensures that questionable cases are flagged before data cleaning and model training.

2. Risk Classification Model
   Using the cleaned and validated dataset, implement a risk classification model that assists healthcare professionals in determining the most appropriate treatment pathway for patients. This model aims to minimize unnecessary interventions, improve patient outcomes, and enhance resource allocation.

This generalized approach can be applied to any disease or condition (as long as sufficient training data is provided), enabling healthcare systems to identify and reduce low-value care practices, improve efficiency, and support patient-centered, high-quality care. It aligns with the call’s objectives by fostering data-driven strategies, scalability, and transferability across diverse health and care systems.

We can fulfill the below scope/outcomes of the call:

-          Healthcare providers and policymakers make use of evidence-based indicators and methodologies to identify low-value care329 practices, as well as opportunities for improvement and tools to monitor such improvements.

-          Identify instances of overuse, misuse, underuse and unwarranted variation in specific healthcare contexts across different stages of the healthcare process. This analysis should provide actionable insights for policymakers, healthcare providers and healthcare professionals to evaluate the potential of possible strategies for reducing low-value care, allowing for more informed decision-making and improved care practices.