DeepThink Health Strengthens Security and Achieves Unmatched Precision and Scalability in Data Curation

DeepThink is responsible for sensitive health information, making security its most pressing priority. Implementing information security, safeguarding personal health information (PHI), and complying with HIPAA are all major priorities.

To address privacy and security concerns, DeepThink must provide a Precision De-ID solution to automatically de-identify unstructured health data, with an accuracy rate of 99.99 percent. But this process requires powerful server farms with graphics processing units (GPUs).

DeepThink leaders were also concerned with staying cost-effective while trying to scale the AI-based Precision Curation solution to achieve the highest possible accuracy. With millions of patient records to sift through, DeepThink needed a better solution to make the data curated, accessible, auditable, and actionable. The company turned to AWS for help.

To help DeepThink meet these challenges and grow, AWS does the heavy lifting, managing security, including the physical access to any infrastructure. That means DeepThink can focus on implementing security at the system, application, and data levels by using Amazon Redshift. After mapping the technical safeguards of the HIPAA Security Rules and HITRUST domain controls onto AWS features and design choices, DeepThink used the AWS Security Hub to automate security checks against industry standards and best practices, including HIPAA compliance. Additionally, all DeepThink employees were trained in prevention and vulnerability reduction, with a focus on people, culture, process, and technology.

Since building on AWS, DeepThink has consistently exceeded its security objectives and improved efficiency. The company has declared zero data breaches and zero security incidents causing significant brand damage. DeepThink has also achieved monthly uptime of more than 99.9 percent, along with HIPAA compliance and HITRUST CSF certification—the highest standards for health data protection and information security. By using AWS security infrastructure and tools like AWS Web Application Firewall (AWS WAF) and AWS FireWall Manager, the company has also reduced more than 30 percent of the effort (or hours) expended in safeguarding sensitive data.

By building its platform on AWS, DeepThink can scale up and down in minutes and leverage AWS infrastructure, instead of using its own money and resources to manage finite infrastructure. This flexibility helps the company to be more dynamic and fosters more opportunities for growth. Increasing server capacity previously took three months, from procuring the hardware and software and setting up the systems to deploying the applications—with third-party components. Now, DeepThink can complete the process in minutes by spinning up Amazon Elastic Compute Cloud (Amazon EC2).

“We don’t have to reinvent solutions with AWS; we can leverage AWS infrastructure to build what we need much faster. Kind of like electricity. Instead of having your own generator, you can call the electricity company and have them turn it on for you,” says DeepThink president Ping Zhang. 

A unified infrastructure governed by AWS Identity and Access Management (IAM) makes it easier for people to access what they need, anywhere and anytime. Managing a geographically dispersed workforce is no longer a challenge, meaning the team can hire the talent they need from anywhere in the United States.

DeepThink is able to de-identify unstructured data (data from clinical notes, genomic reports, and clinical reports) with 99.99 percent accuracy and curate notes and reports with more than 96 percent accuracy. Already benefitting from the accuracy of the data in an optimal and usable structure, DeepThink can run complex AI models and scale to its needs more cost effectively using AWS.

Data is no longer unstructured, siloed, and inaccessible. Self-service analytics allow non-data scientists to analyze patterns easily, to drive actionable insights and facilitate evidence-driven decision making using Amazon Redshift. For example, DeepThink can find patients with certain gene mutations and clinical conditions, identify the right patients for a clinical trial, or find the right set of trials for a particular patient. Providers and life science researchers have this data at their fingertips, so they can find what they need with a single search and spend more time focusing on their patients or discovering novel treatments.