The Buck Institute for Research on Aging is a global leader dedicated to extending human healthspan and advancing research into age-related diseases such as Alzheimer’s and Parkinson’s.
To accelerate discovery, the Buck Institute partnered with Compass UOL and AWS to modernize its data-driven research approach through a Proof of Concept using Amazon SageMaker Canvas—a no-code machine learning environment that enables scientists to build models and generate insights without programming expertise.
Analyzing complex Fox Insight datasets—including participant metadata and microbiome data—required technical skills that most researchers did not possess. This limited the speed at which hypotheses could be tested and insights generated.
The Buck Institute’s Parkinson’s research team sought to uncover correlations between exercise habits and gut microbiome composition. However, manual data preparation and analysis made it difficult to scale experimentation, visualize patterns, or validate findings efficiently.
The goal: create an accessible, unified platform that allows researchers to explore Fox Insight data, build predictive models, and accelerate hypothesis validation—without writing code.
Compass UOL designed and implemented an end-to-end AI and analytics environment on AWS, enabling secure, scalable, and collaborative research workflows using:
- Amazon SageMaker Canvas
- Amazon SageMaker Studio
- Amazon S3
- AWS Glue
- Amazon Athena
Solution Highlights
Unified Data Environment: Integrated Fox Insight datasets with microbiome data, using FoxDen as the data exploration platform, with streamlined ingestion and preprocessing pipelines.
No-Code Modeling: Researchers built multi-label classification models in Amazon SageMaker Canvas to predict physical activity levels among Parkinson’s study participants.
Explainability with SHAP: Model interpretability features surfaced key microbial species associated with varying exercise intensity levels.
Collaboration & Accessibility: Curated datasets and model outputs were stored in Amazon S3 and queried using Amazon Athena, enabling continued analysis and collaboration.
Enablement & Training: Hands-on training empowered Buck Institute researchers to independently build, interpret, and iterate on machine learning models.
This foundation also supports future phases, including expanded workflows in Amazon SageMaker Studio and the planned development of a generative AI-powered research assistant leveraging Amazon Bedrock and Amazon Q.
Operational Success
A production-ready SageMaker environment was deployed and integrated with existing research workflows.
Scientific Insight
Early findings suggest that exercise may reduce harmful bacteria while increasing beneficial microbial populations in Parkinson’s patients.
Predictive Modeling
More than 200 features were analyzed, validating existing hypotheses and uncovering new correlations.
Research Acceleration
Insights generated through the platform are informing a new NIH grant proposal and future personalized medicine initiatives.
Researcher Empowerment
Non-technical scientists can now conduct advanced machine learning analysis autonomously.
Read other stories
PagueMenos
Brazil’s second largest pharmacy chain uses AI & ML to personalize product ranking and experiences
Compre & Alugue Agora
Brazil’s first and only social media for Real Estate reduces chatbot costs by 80% and improves responses with AI