JMIR Publications

ABSTRACT

Humanity stands at the brink of a revolution in biological understanding, disease treatment, and overall wellness, driven by advancements in artificial intelligence (AI) and evolving consumer engagement in healthcare. This paper explores the convergence of consumer behavior, health data, policy, and AI to maximize the potential benefits of large language models (LLMs). With data generation expected to reach 180 zettabytes by 2025, the healthcare sector—responsible for a third of this volume—faces an unprecedented opportunity to harness insights from this vast resource. Yet, AI can only unlock its full potential when trained and fed with comprehensive proprietary health data sets. The foundation of AI’s capabilities lies in its data input. Open data repositories have fueled LLMs like GPT, Claude, and Llama, but privacy concerns and proprietary interests limit the data available to AI. To achieve transformative healthcare insights, AI must have access to exhaustive, longitudinal health records (LHRs) that span clinical, genomic, non-clinical, and wearable data. Unfortunately, current systems like Epic’s EMR, despite their extensive coverage and interoperability efforts, cannot maintain complete LHRs due to limitations in policy and technology, such as HIPAA regulations. Key advancements like the 21st Century Cures Act and the introduction of Fast Healthcare Interoperability Resources (FHIR) have improved data accessibility and interoperability. However, significant gaps remain, as evidenced by the growing need for comprehensive data integration and the ONC’s mandates for standardized APIs and information exchange. The challenge is ensuring that consumers, as the primary custodians of their health data, aggregate and provide complete LHRs to facilitate AI's effective use. Rare disease communities exemplify the critical role of consumer-driven data aggregation. With over 7,000 recognized rare diseases affecting more than 10% of the global population, these communities have shown extraordinary willingness to share their data for better outcomes. Registries like the Cystic Fibrosis Foundation Patient Registry and the Diamond Blackfan Anemia Registry demonstrate how comprehensive data collection accelerates disease understanding and treatment. Diamond Blackfan Anemia (DBA), an ultra-rare congenital genetic disorder, serves as a prime example of a suitable use case for pioneering LHR aggregation. The DBA Registry’s collaboration with patient advocacy groups fosters robust participation, making it an ideal environment to test and perfect a scalable data aggregation model. This approach will not only enhance DBA research but also pave the way for replicating the system across all rare diseases, significantly reducing the $1 trillion economic burden of rare diseases in the U.S. By aggregating LHRs and utilizing AI, healthcare providers can optimize decision-making, overcome time constraints, and achieve faster, more accurate diagnoses. Ultimately, this paradigm shift emphasizes that no one cares about health outcomes more than patients themselves. As consumer participation becomes universal, the integration of AI and LHRs will transform the future of medicine, advancing our collective understanding and treatment of diseases at an unprecedented pace.