Cognoa Chief AI Officer: How Immuta Helped Us Overcome Data Challenges of AI-Based Medical Diagnostics

Here at Cognoa, we train algorithms – to diagnose behavioral health conditions, including autism and ADHD – with highly sensitive data from a production database that lives in a Health Insurance Portability and Accountability Act (HIPAA) environment containing patient identifiers and electronic protected health information (ePHI). Data privacy and security are therefore paramount to Cognoa, and we needed a platform that could help enforce data access roles, permissions, and policies beyond the standard resource or table-based control levels.

Cognoa has a group of incredibly talented data scientists who build our run-time engine for diagnostics software. Our legacy practice of providing them with all of the data they required to build models, while removing the ePHI and HIPAA sensitive information, was extremely time and labor intensive. It was essential that we expedite this process, and also continue to anonymize sensitive information for reporting.

Enter Immuta.

With the Immuta platform’s applied, purpose-based restrictions to data – available natively through its data management for AI platform – we can dynamically and in real-time enforce data access and policy restrictions based on the data scientist’s needs. The platform also applies masking techniques that create a view of our data anywhere sensitive information is located, and anonymizes identifiers presenting compliance issues such as names and birthdates.

Previously, we’d create custom scripts where we’d do data dumps through our production database, then cleanse and import to another database where our data science team would run queries against that data. The problem was, every time we did this, we had to update the cleansing strip – format, change columns – before our data scientists could get access. With Immuta in place, we have been able to steamline both data science and engineering teamwork, dynamically adapt in real-time, and accelerate overall productivity. We can also now define and enforce detailed data access policies that guarantee the security and anonymity of sensitive data to meet or surpass industry regulations.

Immuta has also helped us solve latency challenges. Prior to Immuta, our data scientists were constantly looking at historical snapshots of data – of which the cleansed script could be up to a month old. Since implementing Immuta, Cognoa can dynamically adapt in real time –  and when data scientists make a query, it hits the system live.

This is only the beginning. We’ll keep you posted journey together progresses. In the meantime, click here to read the case study that highlights our Immuta implementation to date.