By Bill Siwicki, Healthcare IT | July 23, 2018
Experts say emerging tools could become as important to precision medicine as HTML is to the web.
Precision medicine is something of a Holy Grail in healthcare: Being able to deliver personalized treatments to individual patients to best cure specific ailments is the ultimate in healthcare.
While precision medicine is still fairly nascent today, one can look forward and see what’s coming down the line to change the way personalized health can be delivered. And though precision medicine is a tricky arena to predict, experts have their ideas on where the complex healthcare field is heading, and what the next generation of precision medicine will look like.
The term “next-generation technology” has different connotations for different healthcare organizations, depending on where they are on the innovation continuum; but machine learning-enabled medical image analysis software should be at the top of the list, said Paul Cerrato, an independent healthcare writer who has collaborated on three books with Beth Israel Deaconess System CIO John Halamka.
“To date, machine learning algorithms are now capable of delivering more accurate interpretations of radiological images than human ophthalmologists, and interpretation of dermatological lesions that is just as accurate as that provided by dermatologists,” Cerrato said. “For instance, with the use of deep neural networks, it is now possible for computers loaded with the appropriate software to diagnose skin cancer as well as experienced dermatologists.”
Similarly, Google researchers have demonstrated that a deep learning algorithm is more effective at diagnosing diabetic retinopathy than experienced eye doctors and residents. That feat was accomplished by using the software to scan more than 11,000 retinal images.
Additionally, hospitals need to find a way to integrate genomic data into their EHR systems so doctors can gain quick access to this data at the point of care and advise patients on how it should impact their treatment, Cerrato said.
“But raw genomic data can’t just be dumped into the EHR,” he said. “Provider organizations need an add-on that turns the data into actionable insights that doctors can use.”
“Most doctors are ill-prepared for [patients with genomics data] now, but they must prepare for this to remain viable and competitive in a market where con”
Joel Diamond, MD, 2bPrecise
Beyond cancer: Pharmacogenomics
While most of the actionable data today is in the field of cancer care, there is another area that is probably more important for primary care physicians and will eventually have a larger impact in clinical outcomes: Pharmacogenomic testing.
“The list of drugs that are affected by an individual’s genetic variants is very long,” Cerrato explained. “Certain mutations can increase the effects of specific drugs, making them more toxic. Other mutations can cause a faster breakdown of drugs, decreasing their effectiveness. The FDA has approved pharmacogenomic testing for several of these drugs. The problem to date is that third-party payers have refused to reimburse for most tests.”
But the scientific evidence to support the value of these tests is growing rapidly – to be ready for this future, providers should have the genomic testing in place, he added.
Joel Diamond, MD, chief medical officer of Allscripts subsidiary 2bPrecise, said there will be a continued “greater-than-exponential” rise in the new types of -omics information.
“We haven’t yet conquered the genomics data challenge and soon we will see the influx of other data types – proteomics, metabolomics, transcriptomics, the microbiome, personal device data, etc. – and we will have the similar challenges of making sense of the information within a specific patient encounter,” Diamond said. “There are no standards in vocabularies and terminologies. It is not binary data, and will all rely on interpretations. There will be an increasing need for the merging of this data with clinical information, and the marrying with the equally as rapidly evolving evidence-based science.”
Elsewhere, healthcare will see the rapid rise of consumerism, forcing more transparency and competition in the provider market, Diamond added.
“This will be the case with genomics and precision medicine, with most people having rich data on their genome and expecting their providers to know what to do with it,” he said. “Most doctors are ill-prepared for that now, but they must prepare for this to remain viable and competitive in a market where consumer demand will be like nothing we have seen previously in healthcare.”
And gene therapy is another promising next-generation functionality that will change the way care is delivered, he added.
“CRISPR is first to bat and with it comes a myriad of ethical, financial and IT challenges,” he said. “Health systems are still worrying about interoperability and things that technology has been available to address years ago. They will need a solid foundation in place if they are going to be ready to apply this clinically, outside research labs.”
“As [SMART applications and FHIR interfaces] evolve, the ability to ‘write once, run anywhere’ could be as significant to medicine as HTML has been to general applications.”
Don Rule, Translational Software
Operationalizing precision medicine
A key function of operationalizing precision medicine is the ability to access genetic test results from the clinical context, within the existing workflow, whether the results are stored in the EHR or an ancillary system like a PACS or medication management system.
This will require interoperable IT tools and application programming interfaces that are able to integrate genomic data for use with existing systems without significant IT development or impact to existing system performance, said Don Rule, CEO of Translational Software, a genomics clinical decision support and precision medicine company.
“APIs developed using the Fast Healthcare Interoperability Resources specification, an open-sourced standard based on HL-7 for exchanging health information to ensure interoperability and security, can facilitate integration of genomics data and test results seamlessly and cost-effectively to deliver on this ability at the point of care,” Rule said.
The ability to plug in new forms of clinical decision support and other healthcare apps that make data useful within the clinical context is another next-generation precision medicine necessity, Rule said.
“The FHIR standard helps to normalize the format of data sent ‘over the wire’ between systems, and layered on top of this is the need for a Substitutable Medical Applications, Reusable Technologies (SMART) health data layer that builds on FHIR to facilitate the creation of apps for healthcare,” he said. “Using an EHR that supports the SMART standard, clinicians can access SMART apps like genomic decision support within their existing workflow to enable precision medicine.”
SMART provides a common vehicle for authentication and authorization with the host system that allows a conformant application to function with any compliant EHR without specialized knowledge of the system.
“SMART applications are in their infancy now because many FHIR interfaces are still read-only, and most do not support the CDS-Hooks standard for launching applications based upon events that occur within the EHR,” Rule explained. “However, as these evolve, the ability to ‘write once, run anywhere’ could be as significant to medicine as HTML has been to general applications.”
Twitter: @SiwickiHealthIT
Email the writer: [email protected]