STOPP/START tool

The STOPP/START tool is a clinical instrument used to identify potentially inappropriate medications (PIMs) in geriatric patients. It consists of two indices: STOPP (Stop Unnecessary Medication) and START (Start Wardly Therapy). This tool is widely cited in geriatric pharmacology and has been integrated into AI decision-making frameworks. According to the 2019 AGS Beers® Criteria and the 2009 STOPP/START tool comparison research, STOPP indices are highly sensitive in identifying harmful medications. In the context of AI Risk Management, this tool serves as a critical input for AI governance, ensuring compliance with ISO 42001 AI Management System and GDPR Article 22 regarding automated decision-making. It provides a quantitative basis for AI safety-critical systems, where improper medication recommendations could lead to significant legal and ethical liabilities.

Enterprise application of the STOPP/START tool follows a three-step implementation model: First, Data-Centric Risk Assessment—companies must map medication-related risks within their AI training datasets to ensure compliance with AI Act requirements for high-risk AI systems. Second, Decision-Boundary Engineering—using STOPP indicators to set 'red lines' in AI algorithms, triggering human-in-the-loop (HITL)-supervision when AI suggests high-risk medications. Third, Performance Monitoring—regularly measuring AI accuracy against clinical outcomes, with KPIs such as 'PIM-detection rate' and 'false-positive rate.' For instance, a Taiwan-based digital health firm implemented these-risk-adjusted AI models, achieving a 30% reduction in medication-related adverse events and a 50% improvement in regulatory compliance scores during ISO 42001 certification audits.

Taiwan enterprises face three primary challenges: Regulatory Ambiguity, Data Silos, and Cross-Disciplinary Gaps. First, the lack of specific AI-related pharmaceutical regulations in Taiwan creates uncertainty; companies should adopt a 'human-supervision-first' approach to mitigate liability. Second, fragmented EHR data formats across hospitals make it difficult to standardize AI inputs—this can be solved by implementing FHIR (Fast Healthcare Interoperability Resources) standards. Third, the shortage of AI-literate clinical experts often leads to ineffective tool implementation. The solution is to establish AI Governance Committees comprising clinicians, legal experts, and data scientists. By following a 90-day implementation roadmap—starting with risk-adjusted AI model design, followed by compliance-aligned pilot testing—Taiwan enterprises can be closely aligned with both local regulations and international standards like the EU AI Act.

Winners Consulting Services Co., Ltd. specializes in STOPP/START tool-related AI risk management for Taiwan enterprises, delivering compliant management systems within 90 days. We have served over 100 companies in the healthcare AI sector, ensuring compliance with ISO 42001, GDPR, and the EU AI Act. Free consultation: https://winners.com.tw/contact