Exom, a digital-first contract research organization (CRO), worked with AWS Partner Storm Reply to build a reliable and compliant chatbot using artificial intelligence (AI) on Amazon Web Services (AWS). The solution guides remote participants using smartphones through a verbal informed consent process for retrospective clinical trials. Using Exom’s Genius platform, hospitals and pharma and biotech customers can bypass the need for in-person informed consent through chatbots enabled by Amazon Lex and integrated with Amazon EventBridge, Amazon Simple Notification Service (SNS) and Amazon Simple Email Service (Amazon SES) to allow consent to be easily given through email and text, and boost participation rates in clinical trials.
Exom Group’s Chatbot Supports Remote Patients and Decentralized Clinical Trials
Exom Group is a globally active, full-service, digital contract research organization (CRO) in the healthcare sector. The company offers its customers, such as hospitals, a range of clinical trial applications, including data analytics, virtualization, cost and compliance management, and apps to capture informed consent from patients.
Before a clinical trial can begin, investigators must first obtain informed consent from patients. This process usually happens on-site but many patients hesitate to travel long distances to provide consent in person. This results in lower participation rates that affect the speed and cost of bringing new drugs to market.
Genius Engage, Exom’s existing informed consent mobile app, was already running on AWS. It uses video and document workflows to explain a study and its associated risks to a patient but still requires patients to visit the study site to provide consent on the investigator’s tablet, or remotely on the patient’s device.
In late 2022, Exom Group chief executive officer Luigi Visani saw the need for its Genius solution to have an always-available, easy-to-use, remote consent system to help hospitals. The company approached AWS Partner Storm Reply to design and implement the service.
I was amazed by how simple it is to build a chatbot on AWS. I know how much work it took to develop my own chatbot systems a decade or so ago. But today, setting up Connect and Lex is achievable for nearly everybody.
Developing a Verbal Consent Chatbot Prototype in 1 Week
The catalyst for developing a remote consent application were hospitals near an Italian city that were struggling to obtain consent from patients from the surrounding countryside. In this case, investigators needed only oral consent for a retrospective study that would reuse patient data from a previous study.
The hospitals’ existing processes for obtaining consent remotely were time-consuming and ineffective. Patients seldom answered or returned phone calls from investigators and even fewer responded to letters in the post. “The outcome to postal outreach was more or less zero,” says Peter Rottmann, chief technology officer at Exom Group. “When patients receive a friendly letter from the hospital asking them to participate again, most read it and forget it. They don’t go to the post office and send an answer or make a phone call.” In addition, older patients found navigating the existing remote consent procedure via the internet to be challenging.
In late 2022, Exom and Storm Reply began exploring the feasibility of building a remote verbal consent application using Amazon Lex, which lets you build chatbots with conversational AI, and Amazon Connect, an easy-to-use cloud contact center.
In just 1 week, the team had a prototype running in Exom’s AWS sandbox environment. “I was amazed by how simple it is to build a chatbot on AWS,” says Rottmann. “I know how much work it took to develop my own chatbot systems a decade or so ago. But today, setting up Amazon Connect and Amazon Lex is achievable for nearly everybody.”
Storm Reply collaborated with Exom’s developers between December 2022 and April 2023 to design, build, and deploy the Genius vEngage solution for remote verbal consent on AWS as a managed chatbot with a conversational interface. Using the service, patients can provide consent online at their convenience via a mobile device. The system can resume an interrupted session, suspend a call, or request a call with the investigator using Amazon Kinesis, which cost-effectively processes and analyzes streaming data at any scale as a fully managed service.
Rottmann says that Storm Reply met Exom’s high-level system requirements for safeguarding patient health information in the cloud. To meet European Union (EU) data regulations, Exom must both securely manage and delete the data as needed. Its EU patient data is stored in the AWS Frankfurt region in Germany using AWS Regions and Availability Zones. In addition, sensitive data like names and mobile numbers are deleted after the patient’s consent is processed.
Our partnership with Storm Reply gives us the security and awareness to be confident our production environment is running smoothly.
Eliminating Human Error Paves Way to Generative AI in Clinical Trials
Exom Genius vEngage is an automated, event-driven, microservices web application with business logic defined in AWS Lambda, which lets you run code without thinking about servers or clusters. The AWS Lambda functions control when the application reads, modifies, or deletes data stored in Amazon DynamoDB, a fast, flexible NoSQL database service for single-digit millisecond performance at any scale.
The company stores its data in Amazon Simple Storage Service (Amazon S3), object storage built to retrieve any amount of data from anywhere.
Exom’s AWS Lambda functions are responsible for initiating backend processes triggered by various events during the consent gathering process, such as launching Amazon Connect only when remote verbal consent is required. AWS Lambda functions are also used to verify patient voice consent in Amazon DynamoDB, adapt chatbot workflows, and automatically generate PDFs of call summaries and consent forms stored in Amazon S3.
Genius vEngage has given Exom a fully automated, serverless web application that standardizes remote informed consent for customers. But vEngage is also the springboard for Exom to explore generative AI to improve patient and doctor engagement on calls. “We see potential in using AI to analyze answers or generate questions, or use models taken automatically from a database scheme that we can present to the patient or doctor,” says Rottmann. He adds that Exom is always looking for new AWS services to help it develop new products when it identifies a gap in the market.
Amazon Lex’s conversational capabilities also eliminate human error when documenting consent conversations. Instead, these conversations are automatically transcribed and generated as PDFs that investigators can verify for accuracy and then electronically sign.
The chatbot application is integrated with Amazon EventBridge—a serverless event bus that helps you receive, filter, transform, route, and deliver events. Reports and messages are routed to investigators and patients through the Amazon Simple Notification Service (Amazon SNS) and Amazon Simple Email Service (Amazon SES) for email and SMS messaging. This allows investigators to easily share a text consent form and a verbal explanation of the study with the patient using smartphones. Ultimately, the application makes it possible for clinical studies to attract more patients and scale up as needed.
Infrastructure as Code Simplifies Deployment to New Environments
Storm Reply delivered the application as infrastructure as code so Exom can easily replicate, customize, and deploy it to new environments. Exom can also test new code before deployment in AWS CodePipeline, which automates continuous delivery pipelines for fast and reliable updates.
Following the project’s completion, Exom continues to rely on its partner. Using Storm Reply’s 24×7 human-operated, managed support, Exom knows it has a stable production environment, which frees up its developers to work on products and features with the greatest business value. “Our partnership with Storm Reply gives us the security and insight to be confident our production environment is running smoothly,” says Rottman. “When we need some new requirements and changes, we know they are here to support us—and to help us improve and deliver our ideas.”