Switchboard, how MD replaces real-time printed call in clinical contact centers with Amazon Nova Sonic

In High Healthcare communication centers, all patient conversations carry clinical and operational significance, making accurate real-time documentation necessary for automated workflows. Accurate, fast documentation enables some smart capabilities without sacrificing clarity or care, for teams to record medical records (EMR) recording, workflow workflows, and complete data entry. By eliminating routine process steps, staff can remain fully focused on patient interactions, improving both experience and outcome. As health systems seek to measure efficiency with compassion, real-time documentation has become the ability to deliver responsive, high-quality care.

Switchboard, MD is an AI and data science LED and DATA company with a mission to prioritize human communication in medicine. Its service improves patient engagement and outcomes, while reducing inefficiency and fatigue. By designing and deploying smart solutions, Switchboard, MD helps providers and operators work together effectively to deliver a positive experience for patients and staff. One of its main solutions organizes the contact center using AI Voice automation, the matching of the medical record in time, and has led to the following reduction, which has led to a significant reduction in queue times and abandonment rates.

With over 20,000 calls handled every month, Switchboard, MD supports healthcare providers in delivering timely, personalized communications. Its AI platform is already helping to reduce Rube's drive times, improve patient engagement, and streamline contact center operations for clinical centers and health systems. Customers using Switchboard have seen results such as:

• 75% reduction in queue times
• 59% reduction in call abandonment rates

Despite these early successes, the switchboard faced a critical challenge: Their coding method could not scale economically while maintaining the accuracy required in clinical practice. Costs and word error rates (WER) were only functional kitchen machines – which can make the critical power of automatic measurement and increase the impact of the switchboard according to the patient's interaction.

In this post, we examine the specific challenges of the switchboard, MD is facing to find the accuracy of the writing and cost effectiveness in the clinical areas, their evaluation process to choose the right writing solution, and the construction of the technology they use using Amazon Connect and Amazon Kinesis streaming. This post works on the impressive results obtained and shows how they were able to use this foundation to use EMR matching and give healthcare workers more time to focus on patient care. Finally, we'll look at the broader implications of healthcare automation and how other organizations can implement similar solutions using Amazon Bedrock.

Choosing an accurate, scalable, and cost-effective documentation model for Computer Center automation

The switchboard, MD requires a written solution that delivers high precision at a stable cost. In clinical settings, transcription accuracy is important because errors can delay the comparison of EMR records, affect referral systems, and disrupt automated workflows. At the same time, the limited support for thousands of calls each week meant that the cost of measurement could not be ignored.

The switchboard was initially tested in a number of ways, including testing open source models such as Open Ai's locally hosted Whisper Model. But these options have been introduced by trade-offs – either in performance, cost, or complexity of integration.

After testing, the team decided that the Amazon Nova Sonic provided the perfect combination of quality writing and efficiency needed to support the health care use case. A model made faithfully to the live audience, even in noisy or dynamic environments. Posted by:

• 80-90% lower written costs
• The average error rate is 4% on switchboards
• Low emissions that match their need for real-time processing

Most importantly, Nova Sonic is well integrated into the switchboard of existing architecture, reducing the increase in engineering and speeding up deployment. With this foundation, the team reduced the manual steps and even had zero, real-time automation across thousands of patient interactions.

“Our vision is to restore the human connection to medicine by removing the administrative barriers that get in the way of meaningful interactions – so our clients can focus on what's happening more:
– Dr. Blake Anderson, Founder, CEO, and CTO, SCWROBOARD, MD

Construction and operation

The switchboard design uses Amazon Connect to capture live audio from both patients and representatives. The switchboard processes the audio stream through Amazon Kinesis Video Streams, which handles the actual conversion of the media before sending the information to be included in the services that contain AWS Lammbda services. Lambda's Lambda functions establish a bidirectional streaming connection with Amazon Nova Sonic using bedrockruntimeclieclieclient's InvokeModelWithBidirectionalStream API. This novel system creates a written transcript for each participant in the conversation, which updates the stackboard and duplicates it to create a complete transcript record. The entire processing pipeline operates in a seamless environment, providing scalable performance designed to handle thousands of concurrent calls while leveraging Nova Sonic's real-time speech processing capabilities.

Nova Sonic Integration: Real-time speech processing

Tying the Amazon Nova Sonic's addiod streaming and processing, the switchboard has been developed and built the ability to separate and reproduce the speakers and the streams and texts that are transmitted to them. This makes the Amazon Nova Sonic ideal for switchboard healthcare applications, where accurate labeling and speaker identification are critical.

Amazon Nova Sonic offers adjustable settings that can be optimized for different health use cases, with flexibility to prioritize or transfer text or respond based on specific needs. A key cost-effectiveness feature is the ability to convert speech output tokens – Organizations can set lower token values ​​when the focus is on writing, resulting in significant cost savings while maintaining high accuracy. This flexibility and cost flexibility make the Amazon Nova Sonic an important switchboard for healthcare organizations looking to implement voice-enabled solutions.

Why Serdless: Health Performance Benefits

Switchboard's selection of Services that use Amazon Connect, Amazon Kinesis video streaming, and Lambda-based operations represent a strategic decision that maximizes infrastructure performance while reducing infrastructure overhead. A seamless approach eliminates the need to provision, manage, and monitor the underlying infrastructure, so the switchboard engineering team can focus on developing Clinical Automation features rather than server management. This design provides built-in fault and high availability for critical health communications without requiring extensive configuration from the switchboard team.

The event-driven architecture of the switchboard, shown in the following, enables the system to scale from handling thousands of domestic calls, with AWS automatically managing the delivery of skills behind the scenes. The Pay-as-You-Go billing model helps Switchboard Pay only for computing resources used during the drive, minimizing costs while eliminating the risk of low-volume supply servers.

Lasting

Switchboard, MD's use of Amazon Nova Sonic shows how the right writing technology can transform healthcare. By achieving 80-90% cost reductions while maintaining clinical grade accuracy, they have created a sustainable foundation for measuring powerful patient interactions in the healthcare industry.

By building on Amazon Bedrock, the switchboard now has the flexibility to automatically expand across multiple use cases and provider networks. Their success exemplifies how perforcare integrators can combine accuracy, speed, and efficiency to transform how care teams communicate with patients – one conversation at a time.

Start with Amazon Nova in the Amazon Bedrock Console. Learn more about Amazon Nova Models on the Amazon Nova product page.

About the writers

Tanner Jones is a technical account manager in AWS business support, where he helps customers navigate and deploy their production systems to opportunities. He focuses on helping consumers develop applications that include AI Agents, with a particular focus on building secure agent applications.

Anuj Jauhari is the manager of SR. Product Marketing with AWS, where they help new customers

Jonathan Woods Is building solutions on AWS based in Nashville currently working with SMB customers. He is passionate about delivering AWS technology to businesses in a way that makes it easier for customers to create opportunities. Outside of work, she tries to keep up with her three children.

Nauman Zulfiqar is a senior account manager based in New York working with SMB clients. He enjoys building and maintaining strong customer relationships, understanding their business challenges and serving as the primary business advocate for the business within AWS.