Care Continuity AI Framework
The Care Continuity AI Framework
In this presentation, I want to introduce you to the Care Continuity A.I. framework and explain how we leverage natural language processing and machine learning to create value for health systems. While we won't dive into all the technical details, this short walkthrough will highlight how the Care Continuity data science team applies proven technology and takes a very practical, transparent approach to A.I. tools.
Machine Learning + Natural Language Processing
Care Continuity leverages two core technologies in our A.I. framework, natural language processing and machine learning. Our solutions rely on parsing clinical records to find provider recommendations, referral opportunities and instructions for patients to follow-up with a specialist appointment. Using natural language processing, or NLP, we can search through unstructured discharge notes and after visit summary text to find recommendations that were not captured as discrete referral orders. NLP technology has been around for several decades and we use this approach in a very practical way to find valuable referral opportunities in daily ED and urgent care discharge notes. Care Continuity also utilizes machine learning models to build predictive scores and stratify referrals for follow-up navigation. We ingest structured data from the EMR and evaluate more than 50 statistical factors to understand the correlation with downstream outcomes and specialist appointment behavior. These models are adjusted and weighted for each specialty, and they are fine-tuned as the model is continually updated with new health system outcomes data. For each customer, we build a unique model using historical and real-time data only from that organization. Let's look at how these technologies are applied in real life.
NLP Example - ED Discharge Notes
This is an illustrative example of an after-visits summary for a patient recently discharged from the emergency department. Using natural language processing, Care Continuity will parse through these clinical notes and uncover follow-up recommendations and opportunities for specialist appointment scheduling. In the ED in particular, these recommendations are often buried in unstructured text, no referral order has been created, and the patient is on their own to figure out the next steps. NLP converts this text into data, enabling the Care Continuity platform to surface referral opportunities and stratify these patients for appropriate follow-up.
ML Example - Predicting Referral Outcomes
Here's another example, showcasing the Care Continuity machine learning models. We apply more than 50 statistical factors, ranging from clinical severity to patient demographics to social determinants of health. These factors are weighted across four dimensions, producing a master score for each patient referral. And importantly, this scoring isn’t one-size-fits-all. Your cardiology service line might prioritize certain factors differently than orthopedics or GI. Our models can flex to reflect those priorities — ensuring each specialty stratifies their referrals to drive their desired outcomes.
ML Development Process
When you look at the full machine learning development process, here’s how it works: Data is ingested and analyzed. Models are developed and deployed. Smarter workflows go into production giving us valuable insights. But, at every stage, our clients remain directly involved —validating the data, shaping the scoring model, and ensuring that everything reflects their unique needs before anything goes live. And because this is a living system, models can be adjusted in real time as priorities shift and the model learns.
Responsible AI Framework
Finally, everything we build at Care Continuity operates under our Responsible AI framework. This means transparency into how the models work, governance in how data is used, and flexibility in how decisions are applied. It ensures our solutions are both powerful and trustworthy — already proven in many large health systems across the country.
A Practical Application of Transparent AI Tools
In summary, Care Continuity uses AI technology in very reliable and transparent ways - NLP and ML have been used in healthcare for many years. The process is proven and the inputs and outputs are clear to our health system customers. When we talk about AI-powered network navigation, we're not referencing futuristic industry developments like generative or agentic AI tools. The fact is, Care Continuity takes a more practical and proven approach to data science and analytics.
