A century ago, the British novelist E.M. Forster distilled his vision of a moral life into two simple words he used as the epigraph for his masterful novel Howards End: “Only connect.”
Graphics & Animation: Dave Hänggi
those two words describe the greatest challenge that healthcare faces. Over the
past two decades, the sector has undergone a rapid and far-reaching digital
transformation. But digitalization has generated a new challenge: information
overload. According to one estimate, the volume of healthcare-related data
being generated digitally doubles every 73 days. Much of it is stored in
discrete silos – such as DICOM images, ECGs, laboratory data – that make
cross-system access difficult. Meanwhile, powerful diagnostic tools often lack
interoperability. The result: instead of supporting informed and actionable
decision-making, the digital revolution too often hinders more efficient
diagnosis or improved patient care.
To realize the true potential of the digital revolution, we must “only connect” the disparate parts of healthcare – facilitating seamless interoperability, open, secure data exchange, and universal functionality across relevant areas of healthcare. The goal: to ensure that all relevant data is at hand when it is needed by patients, healthcare providers, and medical researchers alike.
Healthcare today: Gaps, bottlenecks, silos
The costs of and consequences of the current fragmented state of healthcare data are far-reaching, from operational inefficiencies and unnecessary duplication to treatment errors and missed opportunities for basic research. Recent medical literature is filled with examples of missed opportunities and patients put at risk because of a lack of data sharing.
More than four million Medicare patients are discharged to skilled nursing facilities (SNFs) every year, for example, most of them elderly patients with complex conditions. According to a 2019 study published in the American Journal of Managed Care, one of the main reasons patients fare poorly during the transition is a lack of health data sharing – including missing, delayed, or difficult-to-use information -- between hospitals and SNFs.[3,4] “Weak transitional care practices between hospitals and SNFs compromise quality and safety outcomes for this population,” researchers noted.
Weak transitional care practices between hospitals and SNFs [skilled nursing facilities] compromise quality and safety outcomes for this population.
Health IT Infrastructure, “Hospitals and nursing facilities lack health data sharing infrastructure.”
Even within hospitals, sharing data remains a major problem. In a 2019 study published in the journal Healthcare, researchers analyzed interoperability functions that are part of the Promoting Interoperability Stage 3 requirements, a protocol that has been adopted by US hospitals. Among 2781 non-federal, acute-care hospitals responding to an American Hospital Association survey, only 16.7% had adopted all six core functionalities required to meet the objectives.
Silo-ing of data and incompatible datasets remain another roadblock. In a 2019 article in the journal JCO Clinical Cancer Informatics, researchers analyzed data from the Cancer Imaging Archive (TCIA), looking specifically at nine lung and brain research data sets containing 659 data fields in order to understand what would be required to harmonize data for cross-study access. The effort took more than 329 hours over six months simply to identify 41 overlapping data fields in 3 or more files and harmonize 31 of them.
As researchers wrote in an August 2019 article in Nature Digital Medicine, “[i]n the 21st Century, the age of big data and artificial intelligence (AI), each healthcare organization has built its own data infrastructure to support its own needs, typically involving on-premises computing and storage. Data is balkanized along organizational boundaries, severely constraining the ability to provide services to patients across a care continuum within one organization or across organizations.”
Data is balkanized along organizational boundaries, severely constraining the ability to provide services to patients across a care continuum within one organization or across organizations.
Panch, Mattie, Celi, The ‘inconvenient truth” about AI in healthcare.
What can be done to bridge these gaps? Technology innovators and healthcare IT experts have already taken up the challenge. Important progress is being made on many fronts. Along the way, several key “best practices” have emerged.
Patient-centric data access. Historically, digital imaging and other data have been siloed within a particular department – radiology, cardiology, orthopedics or oncology, for example. In the future, a patient’s data will follow that patient across all healthcare encounters and all specialties, using an open patient data model. The principle of “many into one” will mean that all data required to support improved quality of diagnosis and treatment for that patient will be functionally “many into one” in one place, and always associated with the patient.
Central core software modules. To address the continuing challenge of interoperability and ensure that all the parts of the system use the same syntax and speak the same language, central core software modules will be used to “translate” data coming in from a variety of sources, including third party vendor offerings. As these software modules expand, they will be able to make more and more connections, incorporating more and more data and functionality across medical specialties.
Strategic use of AI. Too often clinicians have to shift through reams of irrelevant data in order to find the key information they need to support actionable decisions. AI is proving to be a powerful tool to identify and highlight key diagnostic findings, delivering the data physicians need at the push of a button. AI can even be used by physicians to identify the most appropriate workflow for a given case.
AI at the local level. AI is usually associated with mining large datasets. But this powerful tool can also execute learning on-site – analyzing a medical system’s data about patients with a particular form of lymphoma, for instance, and then comparing its findings to insights gained by other groups. Increasingly, on-site algorithms will conduct science in real time as clinicians do their work.
All the tools in one place. The design of user interfaces will be critical. The goal is to make the data physicians need to require an actionable decision and the tools they need to analyze that data available together in one comprehensive and easy-to-use interface. The principle of “many into one” will mean that an interface provides information relevant to a patient across all specialties, breaking down the walls between “ologies” that have hampered information exchange in the past.
across the care spectrum. The same principle will position everyone, from
every specialty, as equal players in a patient’s care. Using a single
interface, all the members of the care team will be able to participate
throughout the course of care delivery.
Knowing is not enough
As we “only connect” the powerful tools of digitalization, the benefits will be far-reaching. For physicians, a single interface will save time, reduce the risk of errors, allow physicians to work at the top of their license, improve outcomes, and interact more with patients.For members of a care team, centralizing all the aspects of that care in one place will encourage a sense of teamwork and enable all the members of the team to provide insights in how to improve care. By way of just one example, radiologists with expertise in AI will become a multi-disciplinary resource.
For healthcare administrators, streamlining functions and reducing unnecessary duplication will cut costs while improving value. Interconnected systems will improve the efficiency of operations at every level.
For researchers, expanding access to data now stranded in discrete silos – and making that data easy to analyze – will provide insights that could lead to new diagnostic tools, new treatments, and even cures.
The most important benefits will accrue to patients. The more knowledge physicians have, the better the care they can provide. Streamlined workflows and optimized procedures will improve the care experience and reinforce a patient’s confidence that everyone on the team is on the same page. Through making connections, we can improve preventive care and ensure better outcomes when patients need care.The digital revolution has provided powerful new tools and unprecedented insights. Now it’s time to truly put those resources to work. As another towering writer, Johann Wolfgang von Goethe, declared more than two centuries ago: “Knowing is not enough; we must apply. Willing is not enough; we must do.”
About the Author
Peter Jaret is a frequent contributor to the New York Times and other publications. He is the author of several books, including Nurse: A World of Care (Emory Press) and Impact: From the Frontlines of Global Health (National Geographic).
Last accessed Nov. 9th, 2020
 Healthcare Dive, “Google Cloud’s head of healthcare on data silos, cybersecurity and APIs in the age of the cloud,” February 20, 2019. https://www.healthcaredive.com/news/google-clouds-head-of-healthcare-on-data-silos-cybersecurity-and-apis-in/548730/
 American Hospital Association, “Sharing Data, Saving Lives,” 2019. https://www.aha.org/system/files/2019-01/Report01_18_19-Sharing-Data-Saving-Lives_FINAL.pdf
 American Journal of Managed Care, DA Cross and JS McCulloch, Jan 7, 2019, Drivers of Health Information Exchange Use During Postacute Care Transitions https://www.ajmc.com/view/drivers-of-health-information-exchange-use-during-postacute-care-transitions
 Health IT Infrastructure, “Hospitals and nursing facilities lack health data sharing infrastructure,” https://hitinfrastructure.com/news/hospitals-nursing-facilities-lack-health-data-sharing-infrastructure
 Healthcare, Thompson, MP, Hospital adoption of interoperability function, 2019 Sep;7(3):100347. https://pubmed.ncbi.nlm.nih.gov/30595558/
 Basu, A et al, JCO Clinical Cancer Informatics, Call for Data Standardization: Lessons Learned and Recommendations in an Imaging Study, 2019 Nov; 3:1. https://pubmed.ncbi.nlm.nih.gov/31834820/
 Panch, Mattie, Celi, The “inconvenient truth” about AI in healthcare, NPJ Digital Medicine (Nature Online), Aug 16, 2019 https://www.nature.com/articles/s41746-019-0155-4
Syngo Carbon consists of several products which are (medical) devices in their own right. Some products are under development and not commercially available. Future availability cannot be ensured.