Data Drives Value

Data is King

Data - it drives everything we do.?Data allows us to analyze the world more efficiently and make informed decisions to drive change or improve outcomes.?Without data, we’re effectively flying blind; second guessing or using trial and error to reach desired results, and how can you improve your results and outcomes if you don’t understand what you need to change to improve??Data provides leaders with the power and insight to make more informed decisions.?Understanding data allows you to evaluate processes and determine where improvements need to be made or change workflows that may be more effective.?With insight from data, you can ask new questions and look at old problems in new ways.?And if we consider all the data locked inside our imaging archives, what would you do if you could tap into more insight from that data??Ask better questions??Improve processes you thought were efficient (but now see room for improvement)??Solve challenges you previously thought were unsolvable??The options may be endless but regardless, data must be used correctly to be effective.

Nowhere is data – and understanding that data - more important than in healthcare.?For example, consider all the data points one must analyze as new drugs are being evaluated during human clinical trials for safety and effectiveness.?Maybe consider all the data points you would need to measure when evaluating patient throughput and the workflow efficiency of your staff and processes??How about analyzing patient conditions and procedure mix with coding and billing information??Gaining a better understanding of every process, every workflow, and every outcome allows us to make better decisions, and when analyzed correctly, dramatically improves patient care, reduces costs, and increases revenue.

Executives are presented with data every day that must be interpreted correctly to make decisions that affect the outcomes of their organizations.?If the data they are presented with is incomplete, their decisions may not have the desired outcome they are expecting.?More importantly, if the data does not align to their organizational goals, it really doesn’t help them.?The right data allows us to make impactful changes, yet without it, the value of that data provides zero value in your decision-making process.

Data Driven Healthcare

To have a conversation about data driven healthcare, we must first define the key areas data is typically captured from for analysis:

1. Clinical data – data coming from individual patients - such as blood pressure, pulse, data collected from wearables, etc.
2. Service data – such as operational improvements (wait times, process time, etc), clinical outcomes, and patient experience data (post visit surveys, for example)
3. Organizational data – operational performance (financial metrics, growth rates, geographic expanse, etc.) and employee performance/work experience (turn-over rates, competitive salary rates, etc.)?
4. Clinical Research – data gathered during clinical trial phases to ensure the safety, efficacy, and viability of new drugs, medical equipment, or other solutions used to improve health

As healthcare organizations look to use data to drive their decisions, a key factor that must be considered is what will be analyzed and what outcome are they trying to achieve.?Take Artificial Intelligence (AI) solutions for an example.?Today, there are around 350 AI algorithms approved by the FDA and that number will continue to grow as the AI market continues to evolve.?With that many solutions available, how does an organization choose the best solution(s) to help them solve the clinical challenges affecting their organization??

To narrow the field of options, you need to set proper organizational and business goals, as well as define what you want to achieve as an outcome.?Once you understand your goals, you should quickly be able to narrow the list to a handful of solutions that would provide you the most value and match your needs. ?Most of today’s AI solutions are considered point solutions, meaning their scope is very narrow (they typically do one thing - like identify a lesion in the chest or liver) and many are not broad enough to return a strong ROI except in specific clinical situations (meaning scale is a challenge).?When you factor in data quality issues, the ROI becomes even less.?Deploying multiple point solutions brings further challenges, such as 1) how will you manage multiple solutions in your workflow; 2) what data was used to train the models; 3) how much data was used in the model training and what was its source (focus on model bias); and 4) how will you continue to train the models, to name a few. ?Most solutions use limited data sets of test data and when presented with real-world data, often find lower significance in their outcomes.

A typical AI algorithm can take upwards of 12 months to implement and may require as many as 10 FTEs, so adding multiple AI algorithms can significantly increase resource requirements, increasing the costs to manage these new solutions.?Yet even with these challenges, AI is proving its benefits in many industries and healthcare is no exception, as new studies appear weekly documenting AI's benefits.? For your organization, if the solution matches your original goals, you will see the value. ?AI can also drive significant improvements (and greater ROIs) in areas of workflow bottlenecks and challenges, providing even greater operational gains.?But to get optimal results, everything circles back to the most basic point of this article - data. ?The better the data your AI has to work with, the more accurate the outcome.

The Path Towards Data Driven Healthcare

By 2025, the world will be producing around 175 zettabytes of data (that’s 1 billion terabytes!).?If you’ve never heard of that much data, on DVDs, it would circle the earth 222 times, or take you 1.8 billion years to download it to your PC!?The number is staggering, which is why it’s important to understand what data to use and how to create a culture that uses the data you capture to make better decisions.?To drive a data driven organization, the following are suggested steps you should take:

1. Eliminate disparate silos of information – Virtually every healthcare organization is a patchwork of somewhat connected (or not connected) IT solutions that lack the capabilities of strong interoperability.?Gather all your data in a single repository, or if you plan to keep your imaging data and clinical data separate, find a solution that can aggregate the two when pulling data for analysis.
2. Figure out where your organization is in terms of analytics maturity – Need some help??Look to the HIMSS Analytics Adoption Model for Analytics Maturity (AMAM).
3. Set and align business goals with healthcare analytics – Design your use cases to achieve business goals.?Start small with a targeted approach before growing large.?You will learn a lot in the process and understand what you need to change before going large scale.
4. Governance – Data driven culture needs to be at every level in your organization when driving data-based decisions for both patients and employees (clinical and operational).?You MUST standardize your data to ensure clinical, IT, and operations are all aligned.
5. Single source of truth – If you want to achieve value-based care, you need to collate your data into a single set of reliable, standardized healthcare metrics.?Ensuring you have tools to collect, clean, and standardize your data will help you get there effectively.
6. Create stakeholder driven applications – Many times, the “tools” are chosen over the resources needed to maintain the tool or alignment with the business outcomes desired.?One must not put the cart before the horse so to speak.?Understand what you are trying to achieve, then select the appropriate tool to get you there, so the tool produces the results you are hoping for.
7. Democratize your data – make sure you empower your employees to use the available tools by making sure they are flexible, role-based, and self-served.

It’s worth repeating that poor data can fail your attempts to have a data driven organization.?Clean, standardized data is the critical first step to achieve success, yet data standardization in healthcare continues to lag, which answers the question of why better ROIs aren’t achieved.

Data in Medical Imaging

In the US alone, over 7.6 billion images are inaccessible due to non-standardized data.?To impact greater change in healthcare outcomes, we need to tap into the valuable information locked inside the vast amount of imaging data that exists today (both historical and present).?More data provides us with more insight and the real power of AI can be in detecting and understanding the nuances and patterns this data would provide while creating real-world insights to clinicians at the point of care.?In fact, almost all real-world evidence solutions today do not include imaging data, meaning we are missing out on a lot of valuable insight.

But how do we get there??DICOM, although the standard used in medical imaging, is a loose standard.?For instance, data populated in DICOM tags from two CT scanners can be in completely different places – or labeled differently.?This wreaks havoc on today’s PACS systems (and PACS Administrators) as rules are built to address as many permutations as possible to identify a CT Abdomen with and without contrast correctly from different scanners or locations.?Miss one and surely your PACS display protocols will fail – and this adds frustration and time to radiologists who are under pressure to read more every day!?Even worse, DICOM fields may have missing data (or the data is populated in a different tag location than expected).?Considering these challenges (and they occur in virtually every facility in the world today), how can we tap into our imaging data with any real consistency??How would you search through terabytes of data to find “every female patient between the age of 40-45 that has a pulmonary embolism from 1.5-3mm in size, that had a CT Chest, and lives in zip codes 323xx or 324xx”??Basically, it’s virtually impossible, as you will surely miss studies without a standardized ontology.

We must begin to standardize the data we capture by populating DICOM fields in a consistent manner to achieve actionable insight into our imaging archives.?Then, we must pull relevant data from the clinical reports and EHR to provide well rounded insight and information about the data we have. ??Once we get the data standardization right, we can begin to build a real-world database that includes valuable information from medical images.?But it all starts with data standardization, and once we get there, we’ll see a big impact on clinical insight and research, which can truly begin to change the outcomes in healthcare we all desire to achieve.

Solving Data Standardization

Standardizing data is a big step and more complex than many believe – hence that's why it’s not been tackled in the past (early on, motivation and cost may have prohibited it).?Some solutions may have limited capabilities to solve some of the problems, but they remain departmentalized and don’t necessarily address the entire problem.?What’s worse is that healthcare continues to lag other industries in terms of data standardization, simply due to the siloed approach that has evolved over time.?To achieve greater insight in the data that exists, organizations must begin to focus on data standardization and interoperability.?Just like building a house, the foundation is critical for long-term success, and that goes for data too – you must invest and start at the beginning as data standardization will help your organization decrease costs, streamline workflows, and improve processes, and that leads to better insight and outcomes.

Recognizing the large gap that exists in data standardization, Enlitic developed an expandable enterprise platform that standardizes data and solves the challenges all healthcare organizations face: efficiency, capacity, and scale. Our Curie|Endex solution uses computer vision and natural language processing, along with sophisticated AI to create a standardized ontology of clinical data that ensures consistency is applied across all your medical images.?The value of this can be seen in better workflows, improved data exchange, consistent display protocols, better communication among caregivers, reduced billing errors, and greater insight to your imaging data.?Data standardization is critical to healthcare organizations that are focused on data driven healthcare, whether in clinical research, improved data searches, better operational workflows, reduced clinician burnout, or time saved.?If you would like to learn more, head over to www.enlitic.com or reach out to me directly for a conversation!