How to Use AI for Patient Portals

Summary

This video discusses how AI is being used in healthcare to manage the growing volume of patient portal messages. The speaker highlights three key AI-driven solutions: urgency tagging to prioritize critical messages, AI-generated response suggestions to enhance empathy and efficiency, and workflow automation to streamline message routing. These models are highly accurate, easy to implement, and reduce the workload for healthcare staff while improving patient response times. Overall, AI helps optimize digital patient interactions and enhances both patient and staff experiences.

Takeaways

• 📈 There has been a 30% increase in patient portal messages over the past few years, causing a need for better management systems.
• 💡 AI is well-suited to solve the challenge of handling the high volume of patient portal messages efficiently.
• 🚨 4-10% of patient messages are urgent, which can pose risks if left unattended due to the typical 24-48 hour response time.
• 📊 Some health systems are using AI to automatically categorize the urgency of messages, allowing them to prioritize urgent ones faster.
• 🏆 AI models for urgency detection in patient portals have achieved high accuracy, with some systems reporting 95% accuracy.
• 🤖 AI-generated response suggestions can make messaging more empathetic, helping healthcare staff respond faster and with more care.
• 🏥 Health systems can use machine learning to automatically classify and route patient messages to the correct departments, reducing delays.
• 📉 Automating message classification reduces overall workload and helps in improving the turnaround time for responses.
• 👍 AI-enhanced workflows with human oversight are low-risk and help health systems improve productivity without compromising care quality.
• 🔐 These AI models are typically smaller and can be run on local systems, reducing concerns about data privacy and cloud dependency.

Q & A

What is the primary problem discussed in the video regarding patient portal messages?

The primary problem is the 30% increase in patient portal messages over the past few years, creating a high volume of digital interactions between health systems, physicians, and patients.

What is the current ratio of patient portal messages to visits in health systems?

The ratio has increased from one message per patient visit to six messages per visit.

What percentage of patient portal messages are urgent, and why is this a problem?

4-10% of patient portal messages are urgent. This is problematic because these messages may remain unread for 24-48 hours, as health systems typically have a turnaround time of up to two days, potentially delaying critical responses.

How accurate are AI models in detecting urgent messages in patient portals?

AI models can be trained to accurately detect urgent messages with a 95% accuracy rate (AU), effectively helping prioritize messages that require immediate attention.

What is the 'urgency tagging' solution in AI for patient portal messages?

Urgency tagging involves training AI models to identify whether a patient message is urgent. It adds a column or a tag to the inbox, allowing healthcare workers to sort and prioritize urgent messages.

How does AI improve the empathy in proposed responses to patient messages?

AI can generate empathetic responses by structuring the message to include compassionate language at the beginning and end, improving the patient's experience while enhancing the productivity of healthcare workers.

What role does AI play in improving workflow for patient portal messages?

AI can automate the classification and forwarding of messages to the correct department or specialist, reducing delays and workload by sending messages directly to the appropriate team.

How are AI models trained to handle message workflows accurately?

AI models are trained using historical data from previous message forwarding activities, learning to accurately classify and route messages based on past patterns.

Why is this AI-based solution considered lower risk compared to other AI applications in healthcare?

This solution is lower risk because a human is involved in the loop for proposed responses, and every message is still read and handled by staff. Additionally, the AI assists with prioritization and routing, reducing errors.

What feedback have healthcare systems provided about the AI solutions for patient portal messages?

Healthcare systems have reported positive feedback from both patients and contact center specialists. The AI has improved response times and reduced workloads, leading to higher satisfaction.