Smile Digital Health is making positive strides to improve health care through innovative technologies.
Artificial intelligence (AI) remains an exciting area of development in healthcare today, but it is often hyped. As a relatively new technology in the field, some people have inflated expectations of what AI can do. Real and meaningful transformation in healthcare requires managing the risks of AI, with its many benefits.
At its core, AI refers to self-learning probability-based algorithms that adapt over time based on feedback. The term is sometimes applied too broadly, to include rules-based approaches, or too narrowly to focus only on generative AI. Conflating the understanding of AI leads to inflated expectations and risky implementations.
Healthcare has long struggled with patient data trapped in proprietary IT systems, which don’t talk to each other. Like most technologies, the output of AI is only as good as the data you give it. This means that the data needed by AI modules is disconnected and inaccessible. AI can be used to automate many of the operational processes that help achieve high quality data. When it can process high quality data—clinical data that is readily available, standardized and complete—it provides a more meaningful transformation of healthcare. This includes reducing burdens on physicians, and improving timely, high-quality care to populations.
Healthcare Data Standards and AI
Over 30% of the world’s data comes from healthcare. Though it is growing faster than other industries, most of it is unusable and unshareable because it isn’t standardized, and is locked in proprietary IT systems. Healthcare is only just starting to adopt a common shared framework of data exchange called FHIR® (Fast Healthcare Interoperability Resources) comes in. FHIR is an open-standards framework for securely sharing health data across different platforms and care settings, easily and efficiently. Smile Digital Health (Smile) pioneers the paradigm shift in healthcare, using FHIR to build healthcare systems that break down silos and make structured data more accessible, when it’s needed most.
There’s a temptation to jump straight to using advanced AI technologies without first organizing the data properly. Instead of using AI on disorganized data, we leverage AI as a tool to get the data right—standardizing health data across different systems into FHIR, so it’s accurate, understandable and shareable. This is accomplished by training AI modules to complete high-volume, administrative and repetitive data mapping and conversion tasks efficiently.
Healthcare is a knowledge-rich industry. It is important to convert that knowledge into clinical decision tools that help physicians diagnose and make treatment plans based on latest evidence-based research. Creating standardized clinical decision support rules from lengthy clinical practice guidelines documents is a complex, tedious and manual process. At Smile, we use AI to mine these documents, extract knowledge and turn it into decision tools that physicians can access at their fingertips. Our aim is to significantly reduce the 17 years it takes to get cutting-edge research into clinical practice.
Transforming Care through AI
Once the data challenges are solved, AI is poised to transform healthcare in many ways, from delivering personalized care to helping manage costs and administrative burdens.
One of the challenges Canadians face is continuity of care—keeping their medical records up-to-date and accessible when they move between cities, provinces, clinics, or long-term care facilities. Generative AI can address this by creating a ‘Patient Summary’, pulling data from multiple sources (clinics, hospitals, labs, pharmacies) to generate an updated summary of a patient’s health, maintaining clinical context and relevance. This helps new healthcare providers understand a patient’s medical history at a glance.
Another transformational application of AI is predicting future healthcare outcomes before they happen. Healthcare is a journey of events—doctors visits, symptom diagnosis and treatment plans. AI can be trained to discover patterns within these sequential events to make weighted predictions about what is likely to happen next. This helps physicians identify patients at risk and guide them towards better health outcomes. It is also becoming more important to build trust in Health AI applications. To this end, Explainable AI—a system that makes the AI decision-making processes transparent and understandable to humans—is an upcoming development to reduce biases and align to ethical norms.
