Leading hospital improves patient and practitioner outcomes with predictive analytics
At a glance
Key services
- Data, AI & ML
- Transformation strategy
- Discovery & Design
- Advisory
Our client provides the most comprehensive range of healthcare services in Victoria, through its multiple hospital campuses, as well as an extensive network of community programs and statewide services.
The data and analytics team faced a challenge in developing and evaluating predictive models. Their existing manual processes for gathering and incorporating end-user feedback into model refinement were time-consuming and inefficient. Additionally, the team recognised the need for a more scalable and user-friendly approach to MLOps, to support growing business needs and future use cases.
Challenges
Manual processes
The existing workflow for extracting predictions and obtaining feedback from clinical staff was time-consuming and inefficient, making the rapid iteration and improvement of models hard.
Limited user interface
The lack of a user-friendly interface for collecting feedback made it difficult to engage end users and incorporate insights from hospital staff into the model development process.
Scalability concerns
Our client’s existing approach to MLOps was not designed to support the growing demands of the organisation, limiting the potential for future model development and deployment.
Mantel worked closely with our client to design and implement a comprehensive, tailored solution that addressed their specific challenges.
Discovery workshops
The project commenced with in-depth workshops to explore the available data, technical landscape, and aspirations for utilising predictive analytics. This ensured a clear understanding of the requirements and objectives.
Interactive web front end
A user-centric web interface was designed to display and navigate predictive models, capture end-user feedback, and track model performance over time. This provided a seamless and intuitive way for end users to interact with the models and contribute their expertise.
Technical architecture and roadmap
To integrate the newly designed web front end with Databricks, a robust technical architecture and a clear roadmap were created. This integration streamlined the previously manual process of data and feedback gathering, enabling greater efficiency and scalability.
MLOps uplift
Mantel recommended and implemented a more scalable approach to MLOps, ensuring that our client could efficiently develop, deploy, and manage machine learning models going forward.
Achieving outcomes together
The results of this partnership were transformative. The hospital’s data and analytics services team now has a clear roadmap for implementing the new web frontend, completing ML model evaluation and feedback loops, and enhancing their MLOps capabilities. Additionally, they received invaluable insights into MLOps best practices, fostering a culture of efficiency, repeatability, and productivity in future model development endeavours.
Finally, the partnership culminated in design recommendations for a target state Databricks architecture that aligns with the Databricks well-architected framework, ensuring scalability and performance as the hospital’s data and analytics needs grow.
By partnering with Mantel, our client successfully transformed not only its approach to data gathering, and using predictive analytics and feedback loops, but their MLOps journey as a whole.
Implementing a user-friendly web interface, automation of ML feedback loops, and adoption of best practice MLOps have empowered our client to use data-driven insights for improved patient care, and laid a strong foundation for future growth and innovation in machine learning throughout their organisation.
Ready to get started?
Fill out the form to let us know how can we transform your business.
