CLARIO IMPROVE THE QUALITY OF THE DOCUMENTATION PROCESS OF THE CLINICAL TEST WITH Amazon Bedrock

This publication is co -written with Kimen and Shyam Banuprakash of Clio.

CLARIO is a leading provider of final point data solutions to the clinical trial industry, generating high quality clinical evidence for life science companies that seek to bring new therapies to patients. Since the foundation of Clio more than 50 years ago, the company's final data solutions have supported clinical trials more than 26,000 times with more than 700 regulatory approvals in more than 100 countries. One of the critical challenges facing a clrary when supporting its clients is the process that requires a lot of time to generate documentation for clinical trials, which can take weeks.

The commercial challenge

When medical image analysis is part of a clinical trial that is compatible, climate prepares a process of the medical image letter describing the format and requirements of the central review of clinical trial images (the letter). Based on the letter, the climate image team creates several subsequent documents (as shown in the following figure), including the specification of commercial requirements (BRS), training slides and auxiliary documents. The content of these documents is greatly derived from the letter, with a significant reformate and reformulation required. This process takes a long time, may be subject to an inadvertent manual error and entails the risk of inconsistent or redundant information, which can negatively delay or affect the clinical trial.

The clied image team recognized the need to modernize the document generation process and expedite the processes used to create workflows of extreme to extreme documents. CLARIO committed to his AWS accounts team and the AWS generative innovation center to explore how generative ai could help optimize the process.

The solution

The AWS team worked in close collaboration with a clrary to develop a prototype solution that uses AWS ai services to automate the BRS generation process. The solution implies the following key services:

• amazon Simo Storage Service (amazon S3): A scalable object storage service used to store the BRS documents derived and generated by the letter.
• amazon OpenSearch without server server: an without server configuration for amazon Opensarch Service used as a vector store.
• amazon Bedrock: amazon Bedrock is a fully managed service that offers a selection of basic high performance models (FMS) of the main artificial intelligence companies through a single API, together with a wide set of capabilities that you need to build generative applications of IA with security, privacy and responsible. Using amazon's mother rock, you can experiment and evaluate the best FM for use, customize them privately with their data using techniques such as adjustment and increased recovery generation (RAG) and build agents that execute tasks using their business systems and data sources.

The solution is shown in the following figure:

Architecture tutorial

1. The documents derived from the letter are processed in a local script in preparation to load.
2. The files are sent to AWS using AWS Direct Connect.
3. The script fragments the documents and calls an inlaid model to produce the incrustation document. Then store the inlays in a database of Opensarch Vector for the recovery of our application. CLARIO uses a model of Text Invertations of amazon Titan offered by amazon Bedrock. Each fragment is called to produce an incrustation.
4. amazon Opensearch Serverlessis used as the lasting vector store. The integrities of the documents fragment are stored in an OpenSearch vector index, which allows the application to seek the most relevant semantically relevant documents. CLARY also stores attributes for the source document and the associated test to allow a richer search experience.
5. A custom compilation user interface is the main access point for users to access the system, start generation and interact work with a chat user interface. The user interface is integrated with the workflow engine that manages the orchestration process.
6. The work flow engine calls the amazon and Orchestra Rock API the process of generating document specification documents. The engine:
   • It uses a global specification that stores the indications that will be used as an entry when calling the large language model.
   • OpenSearch consultations for the relevant image letter.
   • Loops through all commercial requirements.
   • Call the Claude 3.7 Large Language Model from amazon Bedrock to generate answers.
7. Issues the document specification document to the User Interface, where a commercial requirement writer can review the answers to produce a final document. CLARIO USA THE SONTO CLAUDE 3.7 OF amazon Bedrock for the question question and the application of conversational.
8. The final documents are written to amazon S3 to be consumed and published by workflows of additional documents that will be built in the future.
9. An ai chat agent according to what is necessary to allow the discovery based on documents and allow users to talk with one or more documents.

Benefits and results

By using AWS ai services, Clio has significantly simplified the complicated BRS generation process. The prototype solution demonstrated the following benefits:

• Improved precision: The use of generative ai models minimized the risk of errors and translation inconsistencies, reducing the need for delays and delays in the study.
• Scalability and flexibility: The architecture without server provided by AWS Services allows the solution to be perfectly scale as the demand increases, while the modular design allows direct integration with other climate systems.
• Security: The climate data security strategy revolves around limiting all its information within the AWS Safe ecosystem using the safety features of amazon Bedrock. By maintaining isolated data within AWS infrastructure, Clio helps guarantee protection against external threats and unauthorized access. This approach allows closing to meet compliance requirements and provide customers confidentiality and integrity of their confidential data.

Lessons learned

The successful implementation of this prototype solution reinforced the value of using generative ai models for specific domain applications such as those that prevail in the life science industry. He also highlighted the importance of involving commercial stakeholders at the beginning of the process and having a clear understanding of the commercial value to be carried out. After the success of this project, Clan is working to produce the solution in its medical image business for 2025 to continue offering state -of -the -art services to its customers to obtain better quality data and successful clinical trials.

Conclusion

The collaboration between CLARIO and AWS demonstrated the potential of AWS ai and Machine Learning (ai/ml) and generative ai models, such as the claude of Anthrope, to rationalize the processes of generation of documents in the life science industry and, specifically, for complicated clinical trial processes. By using these technologies, Clio could improve and optimize the generation process BRS significantly, improving precision and scalability. As CLARIO continues to adopt ai/ML in its operations, the company is well positioned to boost innovation and offer better results for its partners and patients.

About the authors

Kim Nguyen It serves as MR Data Director in Clio, where a team of data scientists in the development of innovative IA/ML solutions for the medical and clinical care trials industry. With more than a decade of experience in clinical data management and analysis, Kim has established itself as an expert in the transformation of complex data of life sciences into processable ideas that promote commercial results. His professional trip includes leadership roles in Clio and Gilead Sciences, where he constantly pioneered data automation and standardization initiatives in multiple functional equipment. Kim has a master's degree in data science and engineering of UC San Diego and a degree from the University of California, Berkeley, which provides the technical basis to excel in the development of predictive models and data -based strategies. Based in San Diego, California, takes advantage of his experience to boost the approaches with a future vision of data science in the clinical research space.

Shyam Banuprakash It serves as senior vice president of data science and delivery in Clio, where it directs complex analysis programs and develops innovative data solutions for the medical image sector. With almost 12 years of progressive experience in Clio, he has demonstrated exceptional leadership in decision -based on data making and the improvement of the commercial process. His experience extends beyond his main role, since he contributes his knowledge as a member of the Advisory Board for the Modal Client Experience and UC Irvine program. Shyam has a master's degree in an advanced study in data science and engineering of UC San Diego, complemented by specialized MIT training in data science and Big Data analysis. His career exemplifies the powerful intersection of health, technology and data science, positioning it as a thought leader to take advantage of the analysis to transform clinical research and medical images.

John O'Donnell It is a main architect of solutions at amazon Web Services (AWS), where it provides participation and design at the CIO level for complex solutions based on the cloud in the health and sciences industry (HCLS). With more than 20 years of practical experience, he has a proven history of delivering value and innovation to HCLs clients worldwide. As a trusted technical leader, he has associated with AWS teams to immerse yourself deeply in customer challenges, propose results and guarantee transformations of high value, predictable and successful clouds. John is passionate to help HCL customers achieve their goals and accelerate their native cloud modernization efforts.

Praveen Haranahali He is an architect of senior solutions at amazon Web Services (AWS), where he offers expert guidance and architects safe and scalable cloud solutions for various business clients. With almost two decades of experience in IT, including more than ten years specialized in cloud computing, it has a proven history of delivering transformative cloud implementations in multiple industries. As a trusted technical advisor, Praveen has successfully associated with customers to implement Robust Devsecops pipes, establish integral safety railings and develop innovative IA/ML solutions. Praveen is passionate to solve complex commercial challenges through avant -garde cloud architectures and help organizations achieve successful digital transformations promoted by artificial intelligence and automatic learning technologies.