Full Case Study
How Bluelight Helped Vasc-Alert Revolutionize Kidney Disease Patient Care with Cutting-Edge AI.
Vasc-Alert is an innovative medical surveillance company focused on vascular access monitoring. Their patented technology automates the collection of patient metrics during dialysis, eliminating manual data handling and delivering critical insights to healthcare professionals.
Challenge: Maximizes Efficiency in Software Development and Deployment
In the US alone, more than 35 million people have kidney disease, and 97% of those diagnosed will undergo dialysis due to the low availability of kidney transplants. Unfortunately, some patients will have complications in the months following the procedure, which in some cases can result in death. Early detection can save lives, but it’s difficult to predict who will have complications.
Vasc-Alert’s goal was to develop a predictive AI model to detect complications early. They had vast amounts of data from millions of dialysis sessions, but data inconsistencies and manual entry errors hindered analysis. Existing processes could identify up to 30% of adverse cases, which increased the need for a more robust and accurate solution. Furthermore, their file-based data silos made integration and processing cumbersome.
Bluelight Solution: Innovation at Every Turn
To unlock the potential of Vasc-Alert’s data, we took the following approach:
1. Data Pipeline Implementation: We built DataFlow pipelines to process and clean data from both automated and manual sources, storing the results in BigQuery. The solution supported both batch and streaming processing, allowing for near real-time data ingestion.
2. Advanced Analytics and Reporting: We centralized data in BigQuery, providing fast, real-time analytics and seamless integration with business intelligence tools like PowerBI and Tableau. We also created custom visualizations and automated reporting to give deeper insights into patient health.
3. AI Model Development: Using boosted tree models, we accurately predicted adverse incidents with a 50% improvement in early detection. The models relied on a small set of key parameters, making results easily interpretable for medical professionals.
4. Productionizing the Model: We built a robust training pipeline using KubeFlow Pipelines, enabling containerized, automated retraining and monitoring. The models were deployed on Vertex AI, providing accessible and reliable API endpoints for real-time scoring.
Tech Stack
Vertex AI
Outcome: Revolutionizing Dialysis Patient Monitoring
- Increased early detection by 50%, allowing for prompt intervention and improved patient outcomes.
- Scalable, automated data pipeline for consistent data quality and integration.
- Intuitive visualizations and reports for deeper understanding and proactive care.
- Automation across all processes, including disaster recovery, updates, patching, and IaC implementation.
Achievement: Empowering Medical Surveillance with Cutting-edge AI
Bluelight’s solution transformed Vasc-Alert’s data challenges into opportunities by combining pioneering Data Engineering and Machine Learning to equip medical professionals with the tools they need to make proactive, data-driven decisions. By advancing dialysis complication prediction with a game-changing AI model, we are helping shape a healthier future for kidney disease patients across the US.
.webp){kind=small}
