By Jill McKeon
Shidonna Raven, Chef Editor
Copyright. All Rights Reserved. Please contact us for republishing permission and citation formatting.
September 15, 2022
Source: Tech Native
Photo Source: Unsplash
Often the following restrictions are put in place to protect patients from their information and data being used without their authorization and permission. Thus, to appease both sides the data is give an amity. Nonetheless, the data is still released, used and available to people within the medical industry without knowledge nor authorization of the patient. As detailed below much of this information is poorly secured. It is available electronically and online. Where is your medical data? Is it secure? And how is it being used? This data is also often being used to market to patients directly online and via other media outlets.
We have become experts at using data for convenience. We rely on location data to hail an Uber, use banking data to pay our friends in seconds, and Google auto-completes our questions as we type.
Sharing data in healthcare, however, is not as easy or convenient. Patients are forced to recite their medical history every time they encounter a new doctor. And a lack of data sharing in healthcare means a longer time-to-medicine for all patients.
Innovation that benefits humanity is put on pause when researchers cannot find and access the data they need. But these problems are not caused by a lack of data – in fact, it’s quite the opposite.
The healthcare industry is one of the biggest generators of data worldwide, accounting for an astounding 30% of all global data. The amount of medical data has increased by nearly 1000% percent in recent years, reaching an incredible 2,000 exabytes in 2020.
This wealth of data could potentially solve some of humanity’s most pressing issues, yet 80% remains completely untapped and unutilized.
Sharing medical data can save lives. Healthcare data is essential for longitudinal care, ensures that caregivers can access complete patient information and is vital for drug research and development.
So, what is stopping us from translating data into priceless insights and meaningful action? Let’s look at the top five obstacles slowing down data sharing in healthcare and explain how to expedite data partnerships.
Fragmentation Collaboration is significantly easier when you share a common language. But when it comes to data, health organizations seem to all speak a different dialect, making it extremely difficult to share information effectively.
In the United States’ healthcare system alone, thousands of different IT platforms are used to collect and store medical information – each with different standards and privacy controls. One way to organize this data is with FHIR standardization, but many health systems still do not have the tools to adapt to this standard.
Less than 40% of health systems can successfully share data with other organizations, and one in four U.S. patients report that their medical records were not transferred between providers in time for their appointment. This holds true both within and across organizations. This paints a troubling picture. Medical information is siloed across many different databases, forcing clinicians and researchers to work with incomplete information, compromising quality of care. It also severely limits the scope of global medical breakthroughs that rely on access to rich and diverse datasets for verified findings.
Privacy and Compliance The healthcare industry is subject to over 600 regulations regarding data privacy and security, Consequently, the fear of noncompliance tempts organizations towards inaction.
Sharing medical data compliantly is a tricky and time-consuming business. In an attempt to keep up with regulation and protect patient identity, most health systems have tokenized their data, meaning that they replace the patient’s name with a different identifier such as a medical reference number or a financial ID. While this is a fairly simple solution, it is also superficial and easy to breach.
The next possible compliance method is de-identification, meaning that data is anonymized. This, however, is not a straightforward process. It requires data aggregation or removal of multiple personal identifiers. The statistical level of the anonymity is commonly evaluated by K-anonymity (a coefficient to mark the level of the anonymity. Higher values of K imply a lower probability of re-identification). Often performed manually, this task not only takes considerable time, but can pose a compliance risk as it exposes analysts to private information. HIPAA suggests two acceptable ways to share data: The first, Limited Dataset, meaning that some identifiers are removed or coded.
This is not considered as De-Identified Data Sets, rather a level of anonymization. This method must have a Data Use Agreement in place (DUA). The second, De-identified Datasets, that also have two paths: Safe Harbor or Expert Determination. Safe Harbor requires the complete removal of 18 specific identifiers of PHI, which usually result in barely usable data. A more complex path is the path of Expert Determination, which is usually taken for a specific use case and requires a formal determination by a qualified subject matter expert. In that scenario, the statistical risk of re-identification is very low.
Yet, even both of those methods don’t eliminate the risk of identifying the patient. A professional and determined adversary could still, by running different combinations of queries, uncover the patient identity. This is not the only problem, usually there is a tradeoff between anonymity and data quality, so anonymization may result in data of degraded quality that isn’t good enough for use in medical research.
Security Data breaches are incredibly common and damaging, wreaking havoc on company finances, operations and reputation, no matter the industry. The danger is especially acute given the sensitive nature of medical data. According to IBM, the healthcare industry pays the highest cost for these breaches, with an average of $9.21 million per incident.
In 2018, an average of 17,000 patient records were exposed daily. These records often contain highly personal information, such as medical history, addresses and passwords. To make matters worse, it takes an average of 212 days until a breach is identified and yet another 75 days to contain.
Healthcare data is typically transferred to a central location for storage. The larger the data sharing lake, the more valuable it is to clinicians and researchers and the more attractive it becomes to cyber criminals. One in three healthcare cyber incidents occur on third-party servers, which is exactly where most existing data sharing solutions store data.
Low Quality Data Patient information must be anonymized and aggregated to meet compliance requirements, which results in low quality records. Researchers must be able to access reliable patient-level information for research, for creating new treatments and for obtaining FDA approval.
Creating a longitudinal record of the patient journey is nearly impossible. One in five patient records are matched incorrectly within the same healthcare organization and up to half of the records transferred between health systems are mismatched or lead to duplicate records. Low-quality data skews results and is not reliable or accurate enough to support research, drive value-based care, or offer personalized medicine.
Outdated Practices Finally, and perhaps the most subtle challenge, pertains to existing practices. Healthcare organizations and researchers have grown accustomed to slow and tedious data sharing processes.
It is normal for data to take over a year to acquire. In cancer research, for example, many patients will die long before researchers get even a first glimpse of the datasets needed to find a cure.
As things stand today, sharing medical information is a long and complex process. In research that uses AI models, 50% of time is spent merely on data preparation and deployment to procure the data.
The worst part is that at the end of an expensive and arduous process, organizations end up with datasets that may be largely irrelevant and already out of date. These hard-earned datasets also cannot be applied for different research purposes.
How can such practices impact one's health? Life? Why?
Share the wealth of health with your family and friends by sharing this article with 3 people today.
If this article was helpful to you, donate to the Shidonna Raven Garden and Cook E-Magazine Today. Thank you in advance.
Comments