EHR-Embedded Predictive Decision Support Must Display Transparency Metrics (ONC HTI-1)
In the healthcare industry, the use of electronic health records (EHRs) with embedded predictive decision support systems is becoming increasingly prevalent. However, for these systems to be effective in providing accurate and safe patient care, they must display transparency metrics that demonstrate their validity.
What is happening:
The Office of the National Coordinator for Health Information Technology (ONC), as part of its effort to improve health information technology (HIT) standards, has introduced the “Health IT Infrastructure” initiative. This initiative aims to provide a framework for healthcare providers and organizations to evaluate the effectiveness of HIT systems in delivering high-quality patient care.
Why it demands action now:
The introduction of transparency metrics into EHR-embedded predictive decision support systems poses several risks for healthcare organizations that deploy these technologies. Firstly, there is the risk that these systems may not do what they claim to do, resulting in inaccurate or delayed diagnosis and treatment decisions. Secondly, there is the risk that healthcare providers may be unable to defend their use of AI-powered EHRs when regulators or courts question their decision-making processes. Finally, healthcare organizations are accountable for the actions taken by AI systems, even if these actions result from incorrect data or biased algorithms.
The action:
To mitigate these risks, prudent healthcare organizations should take the following steps:
1. Evaluate the accuracy of predictive decision support systems through rigorous testing and validation.
2. Ensure that these systems align with industry standards and best practices for patient care.
3. Develop clear policies and procedures for AI-powered EHRs to guide staff use and training.
How Straven helps:
Straven & Co provides independent validation services that assess the claims made by vendors of predictive decision support systems against real-world operations, governance obligations, legal exposure, and regulatory requirements. By examining these factors, we deliver a plain verdict: proceed, proceed with conditions, or do not, giving healthcare organizations peace of mind when implementing AI-powered EHRs. Our independence is crucial in this context, as it ensures that our validation services are free from vendor bias and provide unbiased assessments of the effectiveness and safety of predictive decision support systems.
Straven & Co examines AI decisions before they are acted on: stravenandco.com