What is Pathanova?

Pathanova performs analyses of the variations in care pathways for a given patient population.

Pathanova identifies common pathways that define de facto standards of care and quantifies variations from that standard.

  • Providers can compare a new patient with a population of similar patients to inform patient-specific treatment decisions (case-based decision support).
  • Medical societies can identify patterns and use them to define clinical practice guidelines. Subsequently, they can perform studies to assess adherence to these guidelines.
  • Payers can implement and monitor population health programs to drive efficient resource allocation and improved outcomes.
  • Industry can use patterns of care to inform upstream marketing activities, including market sizing and opportunity modeling.


Imagine if there were a fast and intuitive way to understand how patients present in our Healthcare system with adult onset diabetes. What happened before they were diagnosed? What interventions took place after their diagnosis? How how are they doing today? More importantly, how can these patterns of care help us treat or even prevent a specific patient from developing type-2 diabetes.


Our team members work with healthcare claims data sets, and have experienced firsthand the significant amount of bespoke work required to build a representation of a typical patient journey for a disease of interest.

The field of genomics has a rich set of methodologies and tools for studying genes. The gnomAD project from the Broad Institute is a great example demonstrating how you can systematically explore individual genes to understand which regions are conserved and the type and significance of variants.


Our idea for this hackathon was to see if we could represent a patient’s journey through time as a coded sequence, much like a gene, and then adapt approaches from genomic analysis to the analysis of patient journeys.

How we built it

We represented each patient’s journey as a gene-like sequence, where instead of genetic base pairs, we coded diagnostic events and strung them together in time. We developed a full-stack React application using an open-source reference serverless framework that we published earlier this summer. This allowed us to focus on the specific value-added functionality for this hackathon and embed it in a scalable and secure environment on Amazon Web Services.

Technologies used:

  • sst: Making it easy to build full-stack serverless apps
  • aws-vault: securely store and access AWS credentials in a development environment
  • rush: a scalable monorepo manager for the web
  • pnpm: Fast, disk space efficient package manager
  • kysely: A type-safe and autocompletion-friendly typescript SQL query builder
  • react: A JavaScript library for building user interfaces
  • mui: Comprehensive suite of UI tools for React
  • vite: Next generation front end tooling

Next steps

Next steps include:

  • evaluate the solution on more extensive claims data
  • extending the approach to include procedures and prescriptions to build a more complete representation of the patient journey
  • improving similarity metrics to cluster groups of patients based on user-provided weighting of comorbidities, procedures, and prescriptions
  • refining the user experience based on feedback from end-users

Built With

Share this project: