Articles

Matthew A. Michela, CEO & Board Director at Flywheel.io 

At Flywheel, we understand both the power imaging data holds and the challenges it presents to researchers. Because 80% of researchers’ time is spent managing data rather than analyzing it, increasing data efficiency is key to developing new life-changing treatments in a timely fashion.  

Our CEO, Matthew A. Michela, has spent many years addressing the complex needs of providers in the health, health tech and imaging spaces. As a founder, CEO, investor and board member for multiple healthcare ventures, he’s seen how the fundamental differences in managing imaging versus structured data slows down important research and, more importantly, delays the delivery of health care innovations that address cost and quality. We sat down with Matt to discuss the top issues impacting researchers today and how Flywheel can help. 

1. Accessibility and scalability 

The scale of imaging data required to power cutting-edge medical research is immense. Because of the file size of imaging data (e.g., MRI and CT scans require more data to store than text-based files), large datasets are difficult to create but essential for making data-driven discoveries in medicines, therapeutic approaches, AI, and for conducting population health studies.  

However, accessing imaging data at almost any useful scale isn’t typically straightforward. Researchers encounter obstacles a) assembling data because of location or ownership, b) sharing imaging because of lack of interoperability, and c) storing imaging because of cost. Having a scalable system to access, store, and compute vast amounts of imaging data across clinical trials and research studies would enable researchers to more efficiently focus on developing lifesaving treatments. Otherwise, they are left to assemble multiple complex solutions and wrangle much of the data themselves, generating significant delay and wasting resources. 

"What we know is that all researchers are dealing with the same structural issues concerning imaging data management,” Michela says. “How do they ingest data at scale, within their time frame and budget?  How do they assemble imaging on a patient cohort large enough to perform their research? How do they enable peer review and other collaborators to see and verify the research?” 

This is where Flywheel’s ability to aggregate and store massive amounts of data, from different imaging modalities, comes into play. “It might take a typical researcher many months to figure out how to collect all the data they need,” Michela says. “We reduce lengthy administrative cycles cutting down that time dramatically, so analysis can begin much more quickly.” 

2. Curation 

Imaging data comes in a wide variety of formats, each with its own specific use case and requirements. This makes it difficult to organize imaging data into relevant cohorts so it can be queried for analysis and AI development. 

Valuable data exists in the metadata associated with an image set, and that metadata may include hundreds of data fields organized differently by modality and equipment manufacturer. So being able to extract and curate this data on individual images and from multiple types of imaging manufacturers can be quite challenging. 

“Proper curation of imaging data,” notes Michela, “remains the most costly and complex tasks facing researchers. It is influenced by the clinical modality, the type of imaging manufacturer, unique practices of technicians and radiologists in handling the imaging, and many other factors. 

“Curation is essential, however, for medical and technology experts to make meaningful insights and even to create AI. Imaging data is unstructured and non-standard in many ways, and when you try to analyze it over many gigabytes and terabytes of data, it’s incredibly challenging.” 

That’s why Flywheel has implemented the ability to interrogate metadata down to the image acquisition parameter level. This enables researchers to standardize and query datasets with a previously unattainable level of granularity. 

3. Interoperability and shareability 

One of the basic principles of research in all fields is to make data available to collaborators and other researchers so they can verify the experiment and methods, participate in multisite studies, and rerun and build upon previous data experiments. This creates the opportunity for the larger research community to advance breakthroughs more quickly.   

Given the importance of imaging diagnostics in medicine, it’s especially important to continue to improve imaging interoperability and shareability. Data must be easily shareable across enterprises, between institutions and even across countries.  

Unfortunately, inconsistency in imaging data standards and formats can create substantial challenges in both interoperability and shareability. 

“Even within a single medical center, pharma or device creator, you will have different parts of the organization conducting related research using the same data redundantly,” Michela explains. “Independent approaches in managing the data means that each of those areas, in many cases, is operating like an independent island, with its own staff, complementary budgets, different databases and different software.”   

This generates data management driven errors and makes it geometrically more costly to conduct research, slows progress, and makes collaboration and testing verification incredibly difficult. 

“Imaging is incredibly diverse,” Michela notes, “and you need a comprehensive solution like Flywheel to facilitate the efficient management and sharing of imaging data; otherwise the burden of managing multiple solutions or vendors on all dimensions including security and usability is carried by the researcher.” 

To address this challenge, Flywheel has made the ability to automatically standardize and harmonize data central to the platform. This ensures researchers can achieve the FAIR principles of data—that is, ensuring data is findable, accessible, interoperable and reusable. 

4. Security and compliance 

With efforts to increase data sharing and storage and create all forms of advanced imaging AI, protecting sensitive patient information has become more complex. Security is not only about preventing unauthorized access but also about ensuring that data remains compliant with regulations such as HIPAA, GDPR and standards promoted by the government organizations such as the FDA, the NIH and the Department of Defense. For researchers working with personal health data, security protocols must be rigorous to protect against cyber threats and to uphold patient privacy. 

“Medical data access was dramatically different years ago, when the only solutions were to fax documents or produce CDs to share data,” Michela says. “Thank goodness much of this is unnecessary now, with the advent of cloud and advanced computing, but by making data more accessible and attainable in large patient cohorts, the cost of protecting it has dramatically increased. Where once we could make sure data was locked in cabinets and alarms were on at night, now the industry has to continually invest in people and advancing technology to protect electronic data that is accessible to bad actors globally.” 

In addition to ensuring data is able to meet compliance requirements, Flywheel can also be deployed on a validated instance, so researchers can establish provenance, traceability and an audit trail for 21 CFR Part 11 compliance. This accelerates the process of FDA submittal and helps organizations safeguard themselves in the event of an audit. 

“A benefit of Flywheel is managing both data security and shareability, with incredible de-identification functionality and verification protocols,” Michela says. 

5. Balancing innovation with standardization 

Medicine and technology are both evolving rapidly, with new modalities and diagnostic tools emerging constantly, presenting unique data formats and analytical challenges. While these innovations are invaluable for advancing diagnostics, they can pose obstacles to data standardization, which is critical for enabling collaboration and long-term usability of imaging data. 

“People are creating these unbelievably fantastic ways to see inside the body so clinicians can understand and diagnose what’s happening to patients,” Michela says. “While innovation in imaging has evolved, it also introduces new interoperability challenges. We’re in this race—we want innovation, but we must keep pace with the technology involved so we can solve even more complex medical challenges.” 

Helping researchers solve their greatest challenges around medical imaging 

Michela says that “at the end of the day, coming up with a way to solve the mechanics of these imaging challenges with a comprehensive and integrated software solution is essential so researchers can focus on the clinical aspects of their training and effectively allocate scarce resources to the most impactful parts of research.”   

“The current imaging management model is akin to asking researchers to assemble a way to manage and transport their data before even starting their work,” Michela says. “Think of it this way: We have researchers who have spent decades learning their craft and treating patients, and they know they want to get somewhere — to create a new drug that saves lives, for example. But before they can get to work, they have to build an automobile in their driveway from scratch to take them there. They have to be experts in rubber and tires and steel to construct a proper axel. They have to figure out how to construct a working engine and forge a steel frame. They need to create an energy source to power it all.   

“Alternatively, they can become project managers and find the dozens of vendors who do these things and try to design their pieces into a working car.  It’s incredibly complicated, expensive, time consuming, and an inefficient use of their training and expertise.” 

“At Flywheel,” Michela notes, “we’ve already figured out how to build and operate all the essential components of the car and make sure they work together so you can get on the road and start innovating much more quickly.” 

The goal with Flywheel is to accelerate researchers’ progress by reducing time wasted on mundane, but essential, data management tasks so they have more time to focus on important, potentially life-changing studies.  

“Flywheel is not in the business of developing the next life-saving drug or device or therapy,” Michela says. “But if we can help the industry shorten the decades-long process of accessing, curating, investigating and validating the hidden data of imaging, we can shorten innovation cycle-time and support a better use of our scare healthcare resources. That’s incredibly important.”