Determining which doctors are in a given health plan's network is a notoriously difficult data problem with no single source of truth. Even the insurance carriers themselves only have about 50-55% accuracy. Solving this requires integrating multiple data sources and running machine learning models to weigh their reliability.
Related Insights
The majority of what payers identify as 'care gaps' are actually 'data gaps'—a lack of information leads to an assumption of missing care. By solving the data acquisition problem first, organizations can distinguish between the two. This dramatically shrinks the problem set, focusing expensive outreach efforts only on patients with true care needs.
Many pharma companies chase advanced AI without solving the foundational challenge of data integration. With only 10% of firms having unified data, true personalization is impossible until a central data platform is established to break down the typical 100+ data silos.
A patient's self-reported data can be incomplete or biased, as they may only report the "good measures." To get the full picture, companies must gather input from multiple sources, like caregivers and clinicians. Each perspective helps correct the others, creating a more accurate and holistic view of the patient's journey.
Electronic Health Record (EHR) companies have historically used proprietary formats to lock in customers. AI's ability to read and translate unstructured data from any source effectively breaks these data silos, finally making patient data truly portable.
The effectiveness of AI and machine learning models for predicting patient behavior hinges entirely on the quality of the underlying real-world data. Walgreens emphasizes its investment in data synthesis and validation as the non-negotiable prerequisite for generating actionable insights.
Despite the hype, Datycs' CEO finds that even fine-tuned healthcare LLMs struggle with the real-world complexity and messiness of clinical notes. This reality check highlights the ongoing need for specialized NLP and domain-specific tools to achieve accuracy in healthcare.
A key risk for AI in healthcare is its tendency to present information with unwarranted certainty, like an "overconfident intern who doesn't know what they don't know." To be safe, these systems must display "calibrated uncertainty," show their sources, and have clear accountability frameworks for when they are inevitably wrong.
Chronic disease patients face a cascade of interconnected problems: pre-authorizations, pharmacy stockouts, and incomprehensible insurance rules. AI's potential lies in acting as an intelligent agent to navigate this complex, fragmented system on behalf of the patient, reducing waste and improving outcomes.
The CEO of Datycs applies lessons from standardizing wireless networks in the 90s to today's healthcare challenges. He compares siloed EHRs to old proprietary cell towers, highlighting how open standards like FHIR can solve a problem that the telecom industry conquered decades ago.
OpenAI's move into healthcare is not just about applying LLMs to medicine. By acquiring Torch, it is tackling the core problem of fragmented health data. Torch was built as a "context engine" to unify scattered records, creating the comprehensive dataset needed for AI to provide meaningful health insights.
