Navigating the Role of AI in Healthcare: Addressing the Data Dilemma

In the realm of healthcare, the transformative potential of Healthcare Information Systems (HIS) with Artificial Intelligence (AI) is unmistakable, extending its impact to the practice of medicine, patient health, and the overall productivity of the healthcare sector. However, the efficacy of any AI application hinges on the quality of data fueling its models, emphasizing the need for smart data analytics and advanced technologies. The U.S. healthcare system, a prime candidate for AI integration, faces a critical deficiency in comprehensive, standardized, and readily accessible real-world health data. 

The digitialization in healthcare

Embracing the positive momentum, the NEJM Group has recently introduced NEJM AI, a groundbreaking journal dedicated to identifying and evaluating state-of-the-art applications of AI in clinical medicine. The New England Journal of Medicine (NEJM) has also delved into the realm of "AI in Medicine," recognizing both the tremendous potential and substantial challenges inherent in the field. One significant challenge highlighted is the mismatch between the data set used to develop an AI system and the real-world data on which it is deployed, emphasizing the need for a more inclusive approach. 

Drawbacks of AI in healthcare

Unfortunately, progress in addressing this challenge has been slow, with persistent data silos and a lack of a national infrastructure for open health data in the U.S. The annual report from the Office of the National Coordinator for Health IT (ONC) highlights barriers such as insufficient progress in electronic health information sharing, fragmented state/regional health information exchanges (HIEs), and limited incentives for health IT and data exchange adoption in certain healthcare segments. 

The absence of incentives for data interoperability, particularly in securely sharing digitized records among healthcare systems, is a widespread issue across the entire U.S. healthcare continuum. Recognizing this, Paul Howard, Senior Director of Public Policy at Amicus Therapeutics, emphasizes the need for a forcing function for standardization and incentives. The solution, he suggests, lies in focusing on reimbursements as a means to re-align incentives and encourage the creation of high-quality, interoperable machine-readable data sets crucial for AI algorithm development and validation. 

Overcome the challenges

Addressing the challenges specific to rare disease research, initiatives proposed by Howard and fellow researchers include non-proprietary patient registries, improved data standardization, global regulatory harmonization, and new business models promoting data sharing and research collaboration. 

Dr. John Halamka, President of Mayo Clinic Platform, identifies technology policy and collaboration barriers to data sharing, advocating for a data-centric approach to AI development. The shift from a "model-centric" approach to a data-centric one, derived from a national open data infrastructure, holds the promise of successful AI deployments across all healthcare settings. 

While acknowledging progress, Halamka emphasizes the ongoing need to break down data silos and upgrade the development of healthcare AI. Collaborative efforts, spurred by the recent pandemic-induced real-world evidence gathering, highlight the growing realization of the necessity to overcome data challenges. 

Industry-wide associations and federal government initiatives are emerging to address these challenges, integrating smart data analytics and advanced technologies. The Coalition for Health AI (CHAI), co-founded by Halamka, is developing guidelines for credible, fair, and transparent health AI systems. The Alliance for Artificial Intelligence in Healthcare (AAIH) is bringing together technology developers, pharmaceutical companies, and research organizations to establish responsible standards for AI in healthcare. 

Government initiatives, such as the 21st Century Cures Act, are slowly moving towards fulfilling the promise of health data interoperability. The Trusted Exchange Framework and Common Agreement (TEFCA), a new health information exchange framework, are steps in the right direction, fostering a universal on-ramp for interoperability. 

These collective efforts, infused with smart data analytics and leveraging advanced technologies, aim to contribute to the creation of widely shared open data standards. This supplementation of government-mandated data exchange practices will drive new incentives for data sharing through reimbursement requirements, unlocking the full potential of AI in healthcare and ushering in a future of innovation and improved patient care. 

--------------------------------

At VinBrain, we create DrAid™ Enterprise Data Solution where we accompany you in defining, analyzing, and implementing the strategy for your medical work. “Transforming raw data into valuable knowledge is the core value of comprehensive digital transformation” Steven Truong, CEO of VinBrain, says. In DrAid™ Enterprise Data Solution, we offer you 4 solutions including: Smart Automatic Report Generation, EMR Smart Analytics, Hospital Dashboard and Predictive Analytics, and Data Lake, Data Management & Knowledge System.  

Transform your business with our smart solution – DrAid™ Enterprise Data Solution now, to enable a better result effortlessly!