Big Data Analytics in Healthcare

The term "big data" was coined in the early 2000s by industry analyst Doug Laney. In Laney's definition, the term referred to information existing in large data sets, with high update rates and in a variety of formats: from numeric data in traditional databases to structured and unstructured text. Big data has now evolved from 3 initial V’s to 5:

• Volume: The quantity of data produced across all sources.
• ​Velocity: The speed at which new data is created, collected and analyzed.
• Variety: The different types of data being created, including structured data sets, semi-structured data and unstructured text.
• Veracity: The unpredictability of the data collected. Is it of good quality? Did it come from a trusted source?
• Value: The worth of the data being collected.

​While the volume and variety of big data represents a major technical challenge to any organisation, the payoffs are also substantial: enabling patterns and trends to be identified that can be used to inform decision making at all levels. 

Emerging Trends in Big Data

Beginning with the adoption of Electronic Health Records (EHR’s), the role of big data in healthcare has continued to increase in importance at an exponential rate. According to a new market intelligence report by BIS Research, titled "Global Big Data in Healthcare Market- Analysis and Forecast, 2017-2025", the big data in healthcare was estimated $14.25 billion in 2017, and is estimated to grow over $68.75 billion by the end of 2025. In the US, the ongoing evolution from fee-to-service to value-based care models in healthcare has also provided a major impetus. 

The Role of Healthcare Payment Models

The shift in the US payment models put the focus back into increasing quality of care, and improving patient outcomes, as well as into efforts to prevent or decrease progression of healthcare conditions. Value-based care is a positive move back to the original patient-centric essence of healthcare. Fortunately, this also provides a model that incentivizes both provider and payer to work together for better patient outcomes.

This pay for performance (P4P) methodology coincides well with the Department of Health and Human Service initiative of Healthy People 2020, and will continue to gain momentum in Healthy People 2030. Unstructured data is a vital element in the big data analytics in healthcare initiative, allowing for a full 360-degree view of the patient.

Artificial Intelligence and Big Data Analytics

The role of Artificial Intelligence (AI) and big data analytics within healthcare is now recognized around the world, and AI was emphasized as a means of ‘potentially reducing the burden on overloaded health staff’ by the Director-General of the World Health Organization. Natural Language Processing (NLP) is a fundamental component of AI and is part of this exciting opportunity to transform healthcare.

Patient wellness is impacted by many factors, not just clinical care.

Healthcare versus Business

Big data in healthcare differs from business in one important way: the data in healthcare is used to treat people. A business by its very nature needs to make a profit to continue its existence. Healthcare organizations are in the business of caring for people. It is a business that by all ethical principals should not be performing unnecessary procedures to stay out of the red and yet this is still done by some organizations. 

The biggest challenge that big data poses in the healthcare industry is apparent in most areas of medicine. The sheer volume of information makes it difficult for humans to keep up with and it is common for important health related clues to slip by undetected. Timely diagnosis, and treatment is often delayed, due to this challenge. It is therefore important for healthcare payers and providers to make the most of efficient ways to capture and manage big data analytics in order to maintain a healthy business. Read our blog to learn more about the importance of balance of profit and quality care for healthcare organizations.

Priorities for Big Data in Healthcare

Variety, veracity and value are three areas that are especially important in healthcare. Clinicians documenting similar information, will almost always differ in one way or form, and that is why it is imperative to be able to explore your source documents and add certain terms for your specific data questions.

In addition, we should not forget the velocity of information big data produces in this day and age. Although the population rate has been slowing since around the late 1980’s, the population is still growing steadily year on year. It is forecast that the population will rise over 8 billion by 2025. Access our webinar to find out more about big data and population health.

It is in the best interest of payers and providers to find a suitable, evolving solution in order to keep up with the big data  revolution in healthcare, which will only continue to increase further as the population does. Find out how big data analytics is key to competitive advantage for payers in our blog. 

Technologies for Handling Unstructured Data

Structured data is valuable, but it is often focused on supporting billing rather than clinical care. For example, heart failure disease code may be a structured data field, but important information related to conditions remains trapped in unstructured data. The percentage associated with a patient’s ejection fraction will help tell the real story of the patient’s heart failure condition and if they need an adjustment to current care. To simply have a diagnosis of heart failure in structured data is not enough.

Roughly 80% of clinical data is estimated to be in an unstructured form. If the goal is to improve clinical care with insights from clinical care experts, Artificial Intelligence (AI) techniques like Natural Language Processing (NLP) must be adopted as part of big data analytics in healthcare.

Natural Language Processing of Unstructured Data

NLP-based text analytics platforms like Linguamatics NLP enable unstructured text to be translated into discrete data fields by identifying the key concepts and their relationships in healthcare documentation. It enables concepts such as TNM cancer stage, patient ambulatory status, and ejection fraction to be identified through NLP text mining and provided as structured fields. Life style factors such as smoking, exercise, and alcohol and drug use are also important and while structured fields in the EHR they may not be completed and must be mined from clinical notes with NLP. Interest in social determinants of health such as social isolation and food insecurity are also trapped in notes and have been shown to be strong predictors of risk.

Learn more about the role of unstructured data within big data analytics in healthcare. You can also download our whitepaper on mining unstructured patient data for successful population health.

Machine Learning and NLP

In order to enable patterns and trends of big data analytics that can be used to inform decision making, healthcare organizations must utilize AI Machine Learning (ML) technologies available to them. Here are just a few of the areas where Natural Language Processing (NLP) has addressed some of these challenges: 

• Radiology and pathology incidental findings may be missed. 

  • Using NLP text mining solutions radiologists can extract clinical attributes such as cancer stage, tumor size, histology, and biomarker values to quickly identify high complexity cancer cases to better coordinate care. Find out more in our blog about how NLP can help uncover incidental findings.
• Cancer diagnosis/treatment can be delayed due to coding issues.
  • NLP technology can support real-time monitoring for early signs of cancer such as pulmonary nodules in radiology reports. Read more about the cancer conundrum in our blog. You can also download our whitepaper on 9 ways to improve cancer insights with NLP.
• Complex diseases (such as Diabetes) need both unstructured and structured data to ensure proper patient care.
  • Clinicians can take advantage of miningMine scientific literature for insights into genetic mutation and disease association with NLP. Find out how utilizing NLP for HEDIS Metrics can help healthcare and patients.

Big Data Analytics

The value of utilizing big data analytics for improving health is immeasurable. The conclusions drawn from population data may result in substantial changes to clinical care, not only to the population being observed, but also setting standards of future care for others. For this reason, it is important to have a human element connected to many of these decisions, combining the best of technology and human know-how in what is becoming known as Intelligence Augmentation (IA). This approach enables the power of NLP to help identify high risk individuals for example, while ensuring there is human review.

One example of a big data application in healthcare is Atrius Health. Through their use of Linguamatics NLP, Atrius Health proves that once surfaced unstructured data can make a difference on patient care, hospital efficiency and increasing funds. Below are just some examples of this. 

• In 2017 Atrius Health identified 92 otherwise undocumented CHF and COPD patients. Identifying these patients has helped them gain between $50,000 and $100,000 in additional annual risk-adjusted revenue per disease area.

• Before Linguamatics NLP, nurses had to manually review approximately 1,000 charts to find one applicable case. After applying the NLP solution, they were able to decrease the manual review case to now finding one in reviewing six.That’s a 99.4% decrease in time spent on this task.  
• They also report a 33% increase in composite metric performance, ensuring heart failure patients are on beta blockers and ACE inhibitors.

You can find out more by watching the webinar on operationalizing NLP to support value-based care or downloading the case study on supporting value-based care at Atrius Health.

Another prime example of a big data analytics use case in healthcare is where utilizing Linguamatics NLP can help save and improve lives:

A large payer needed to extract member-related data to improve their analysis of Congestive Heart Failure (CHF) populations from a mixture of unstructured formats held in a data lake. Extracted data needed to be integrated with conventional data warehousing and analytics approaches to support improved patient stratification.

Linguamatics implemented their NLP platform in combination with an automation workflow to ingest data from Hadoop, process it, and load it into a data warehouse for analysis. Linguamatics demonstrated that Linguamatics NLP could be integrated into existing Hadoop and Netezza systems to enable insights from unstructured data to be used as part of risk stratification analytics for CHF. The NLP infrastructure is easily extended to support other diseases areas and risk factors, for example diabetes and obesity. Download the full case study to learn how this large payer improved patient stratification using unstructured big data.

Technology available today is already helping us make sense of big data in healthcare but as technology continues to evolve, so does the impact of the information we can extract and utilize from big data. These population scale insights are vital to understand the best treatment options, which result in the best patient outcomes. These large scale insights can then be applied at the individual patient level to see how similar patients have fared and what were their best option. This inevitably leads into the precision medicine area where we can look at the individual in detail.

Think back to 2015, when President Obama announced the Precision Medicine Initiative (PMI): its purpose was to take an individual's genes, environments, and lifestyles, as part of the equation in a person’s healthcare, and ultimately tailor their healthcare treatment to each individual accordingly, using a 360 degree view of patient care approach. Find out more about the use of cancer NLP to support precision medicine, clinical research and population health in this webinar.