ATLAS: AI Framework for Personalized Medication Safety in Older Adults
Researchers have introduced a new system called ATLAS to enhance medication safety for older adults with multiple health issues. This innovative framework uses large language model agents to help with medication reviews between doctor visits, addressing the fact that medication choices can often be affected by various conditions and risks that users might miss. ATLAS creates a medication-safety graph that organizes guideline evidence and updates it based on the patient’s specific situation, leading to a personalized medication conflict graph (PMCG). A multi-agent policy then uses this graph to spot potential issues, suggest safer alternatives, and finalize the medication plan. Additionally, they launched GeriMedBench, an interactive tool for evaluating safety-critical information retrieval. Their results are published on arXiv, reference number 2608.09443.
Key facts
- ATLAS is a coupled graph-policy distillation framework for patient-adaptive medication safety.
- It uses large language model (LLM) agents to support medication review between clinical visits.
- ATLAS structures guideline evidence as a medication-safety graph.
- Targeted questions update the patient state and distill relevant relations into a patient-specific medication conflict graph (PMCG).
- A risk-first multi-agent policy uses the PMCG to screen contraindications, assess cautions and monitoring needs, identify safer alternatives, and verify the final medication plan.
- GeriMedBench is an interactive benchmark that tests safety-critical information acquisition and evidence-based decision revision.
- The framework was evaluated on a European non-interactive multimorbidity benchmark and an Asian interactive multimorbidity benchmark.
- The study is available on arXiv under the identifier 2608.09443.
Entities
Institutions
- arXiv
Locations
- Europe
- Asia