Skip to content
Nubent

/Industry · Healthtech

AWS for healthtech, with patient data under control

You know where every piece of clinical data lives, who read it and when. And if you add AI, the data never leaves your account.

Problems we solve in Healthtech

  • Clinical data copied everywhere

    Macie finds sensitive data and policies keep it out of non-production environments.

  • Nobody knows who accessed what

    KMS keys per data domain and dedicated accounts for sensitive data, reached only over private endpoints.

  • HIPAA, GDPR and local rules at once

    CloudTrail data events and application logs tagged with each user's identity.

  • AI on patient data

    Bedrock with guardrails, anonymized data and everything inside your network.

Regulations:HIPAAGDPRISO 27001ENSSOC 2HL7 FHIR

More about Healthtech on AWS

HIPAA and GDPR on AWS: what is yours and what is AWS's

AWS offers HIPAA-eligible services and a business associate addendum, but that does not make your platform compliant on its own. Configuring access, encryption, logging, retention and backups is your responsibility. Our job is to design that layer so the controls exist, live in code and can be demonstrated.

For GDPR we add data minimization, explicit region selection, processes to handle access and erasure requests, and records that show the legal basis for each processing activity. The same technical controls serve several regulations, so we build one solid foundation instead of separate patches for each audit.

Clinical data traceability from ingestion onward

Traceability starts with identity. We use IAM Identity Center for people, temporary roles for services and Amazon Cognito or your identity provider for end users. Every access to sensitive data is recorded with who, what, when and from where, both in CloudTrail and in application logs.

Those records are centralized in a log account with S3 retention and Object Lock, so nobody can delete them. With Athena or CloudWatch Logs Insights the team can quickly answer an auditor's question or a patient request, without depending on the one person who knows where to look.

Generative AI for healthcare, with safeguards

AI has clear use cases in healthcare: summarizing notes, classifying documents, supporting administrative teams. We do it with Amazon Bedrock inside your account, with private access, guardrails that filter content and personal data, and invocation logs to audit what was asked and what the model answered.

Before connecting clinical data we define which information can be used, anonymize whatever does not need to be identifiable, and evaluate response quality against test sets. AI assists clinicians, it does not decide for them, and the architecture makes clear where each responsibility ends.

/FAQ

Frequently asked questions

Another question? Write to hello@nubent.com and an engineer will answer.

Is AWS HIPAA compliant?

AWS signs a business associate addendum and offers a list of HIPAA-eligible services. Compliance, however, depends on how you configure those services: encryption, access, logging, backups and retention. We design that configuration in Terraform so controls are consistent across every environment and easy to demonstrate in an audit.

How do you protect health data in the cloud?

With several layers: encryption at rest and in transit with AWS KMS, separate accounts for sensitive data, private access through VPC endpoints, least privilege in IAM, sensitive data discovery with Amazon Macie and an immutable record of every access. No single layer is enough; protection comes from combining them and verifying them.

Can you use generative AI with patient data on AWS?

Yes, with safeguards. Amazon Bedrock processes data in the region you choose or, with geographic inference profiles, in regions within that same geography, and does not use your data to train base models. Even so, you should anonymize what is not needed, apply guardrails, log every invocation and have clinicians validate responses before taking a use case to production.

What does traceability mean for a healthtech platform?

It means being able to show, with evidence, who accessed each piece of sensitive data, when, from where and for what purpose. On AWS you get there by combining individual identities, CloudTrail data events, application logs with user context and immutable storage of those records for as long as regulation requires.

Next step

What should we build next on AWS?

30 minutes with an engineer. Concrete next steps, whether or not you work with us.