The recent scandal involving Facebook and Cambridge Analytica with its harvest of over 50 million personal profiles is yet another sombre reminder for us that we need to be more proactive and vigilant with issues that surround our data, especially privacy.
The recent Healthcare Information and Management Systems Society (HIMSS) meeting was mostly an artificial intelligence (AI)-inspired technology fest with frequent allusions to AI, but without sufficient focus on data and all its challenges such as access and security.
The current imbroglio in health care data creates a conundrum in health care:
- First, there is an escalating volume of healthcare data with upcoming data “tsunamis” in the forms of genomic information and also wearable devices.
- In addition, data access, and paradoxically data security, remain opposing but similarly daunting challenges for both the clinicians as well as the patients.
- Third, health care data is often segregated in either data warehouses or data lakes without a common repository that all stakeholders can work on.
- Fourth, health care data is in a myriad of locations, structures, formats, and complexities imbued with unclear definitions of even basic medical terms.
- Lastly, the relational database structure in health care is out of date and difficult to extract information from, and it is in need of innovation to better suit the array of AI tools.
We need final solutions to this health care data conundrum:
- First, we need an AI-focused solution to the volume of data by applying this technology at the point of data acquisition and collection.
- Second, a disruptive technology such as blockchain or zero knowledge protocol can perhaps provide the public of healthcare data access while concomitantly assure patients of data security.
- Third, we need large common repositories (data reservoirs) that share storage and agility characteristics so that every stakeholder can work on the data. Fourth, we need much better data infrastructures that is organized, complete, and therefore AI-friendly.
- Lastly, a graph database format with its elements that are more flexible for health care data would be a good forward-thinking solution to the present-day relational database format. All of these aforementioned changes can lead to a biomedical data nirvana: a federated or virtual software-defined meta-database architecture for all of medicine in a dynamic hybrid cloud with real-time analytic processing (or RTAP) to provide an AI-in-medicine-as-a-service (“AIMedaas”).
The latest issue of AIMed Magazine focuses on wearable technology in the context of artificial intelligence. Just this morning I received a text from a patient’s mother who is concerned about her daughter’s heart rate being low in the middle of the night and queried if we can set up a monitor “real-time” so she can be reassured. This level of data transparency with accompanying anticipatory guidance from clinicians or surrogates will be the rising expectation of our patients and families as more accurate and intelligent wearable devices become available.
Earlier this month, FDA approved Medtronic’s Guardian Connect continuous glucose monitoring tool that uses AI with a predictive algorithm to better modulate glucose levels under different physiologic conditions. While this is an exciting era of improved patient monitoring with AI-embedded devices (towards IoE, or internet of everything), it behooves us to think proactively about how the data will be intelligently used to improve patient care without generating more data exhaust.
In short, we need to be both prudent and patient as not to overhype AI and to focus more on all aspects of data stewardship in healthcare: its acquisition, access, storage, security, and structure. AI is, after all, mostly about the data: AI and its ostensible knowledge dividends will be the valuable reward if we all do a much better job with data itself as it is the foundation of AI.
Dr Anthony Chang, MD, MBA, MPH, MS
Dr Anthony Chang is Chief Intelligence and Innovation Officer
Medical Director, The Sharon Disney Lund
Medical Intelligence and Innovation Institute (MI3)
Children’s Hospital of Orange County