Solving the Data Conundrum in Healthcare - Part 2

Innovations and Technology as Solutions: The second of a three part series. Read Part 1: Barriers to Innovation here. 

In technology — specifically in the area of data analytics — the myriad of available tools can be overwhelming for healthcare administrators or program champions to sift through. Combine that with the integration of said tools into a solution model, and the challenge becomes ominous. Tools (technology innovations) must first be defined and then incorporated into a real-world solution model. Upon understanding and defining these tools, it becomes easier to understand the services and partnerships needed to achieve an institution’s goal. Specifically, understanding what is available allows an administrator or program champion to identify a strategic partner in order to co-create a solution. Simply put, resources are available; healthcare organizations just need help using them.

Natural Language Processing

Of the many innovative technologies available to an organization, natural language processing (or NLP) can be of tremendous value. NLP, essentially, allows computers to understand human language as it is naturally spoken. So how does NLP fit into an institution’s strategy? It is highly relevant, and has become a necessity as we become more data-centric. NLP is already being used extensively throughout the business arena. For example, a business can better determine product placement or target audience through deep analysis of social media posts and tags. Understanding the content and context of posts allows a company to determine where its resources or strategies should be directed. NLP in medicine, helps with the cleaning and sorting of data to make it more useful and insightful. Clinical application of NLP is well-exemplified by the following statement:

The bladder contained stones in a patient with right upper quadrant pain.

A clinician would be able to infer that bladder signifies gallbladder, stones signify gallstones, and that quadrant refers to the location of the pain in the abdomen. But how well do traditional computer algorithms process these pieces of information in context? Without leveraging NLP libraries, it could lead to many mistakes. Bladder might be misinterpreted to mean urinary bladder. Stones could be interpreted as kidney or ureteral stones. Quadrant could be interpreted as location of an organ, not necessarily the body region. With NLP, context is applied and a computer algorithm would interpret the aforementioned statement as referring to gallstones within the gallbladder which account for pain in the patient’s abdomen. This tool or technology can be applied to numerous situations in healthcare, including radiology reports, clinic notes, operative reports, and so on. An institution can achieve highly efficient and successful data normalization and structure through the use of NLP.

Machine Learning

Under the umbrella of artificial intelligence, machine learning (ML) is an application that provides systems the ability to automatically learn and improve from experience without being explicitly programmed. Data is piped into an ML training algorithm. That algorithm is evaluated, and in the clinical space, is validated by a human thus creating a model which is then applied to the actual ML algorithm to which actual real world data is applied. A prediction is made by the algorithm. Accuracy can be very high, but more importantly, accuracy continuously improves with time and increased data sets since the evaluation/validation process is ongoing, further training and refining the ML algorithm model. We see this used today in many industries, including facial recognition in photo library applications. An individual’s face is defined (by the human user), and through continuous data (i.e. more photos in the library) processing, the program can start to accurately “guess” who that person is. Human validation continues to enhance and improve the ML program so that prediction accuracy continuously improves.

This very powerful technology can be applied to healthcare. An example in medicine is the use of ML algorithms to perform image analysis through deep learning models. An ML algorithm is trained to identify lung nodules or lung masses found within the chest upon imaging (e.g. Computed Tomography). With human validation and continued comparative analysis from prior successful predictions, a lung nodule/mass ML algorithm can (and does) achieve great accuracy. This results in improvement of patient care and ultimately impacts clinical outcomes positively by adding another layer of diagnostic assessment.

Cloud-based Platforms

Once thought to represent an ambiguous and ominous term, “cloud” architecture has become an essential platform to perform computer operations, communications, transaction, and data storage/management. Healthcare has been slow to adopt and integrate cloud-based architecture into the IT space due to misconceptions with data security. It was commonly believed that data living behind hospital firewalls was the safest scenario to protect PHI (Protected Health Information). The reality is that cloud-based data storage and management offers even greater security through leveraging industry-leading security standards, large encrypted server environments, and greater access control within the cloud so as to minimize information leaks. In addition to security benefits, data redundancy and data integrity are exponentially improved when leveraging a cloud platform. This is especially important when considering the nature of data in the healthcare space. Hospitals continue to spend exorbitant amounts of money creating redundancy and enhancing integrity for their data when cloud solutions, which provide even greater security standards while offering better levels of redundancy and integrity, are available at significantly reduced costs. In addition, cloud solutions allow for increased scalability, which is a huge benefit as many organizations work to prepare for today, while not accounting for unexpected growth or data management needs of tomorrow. They are then  faced with both financial and logistical dilemmas on strategizing for scalability and expansion.  

There are numerous, relevant examples of cloud-based technology in our daily lives. Look at office productivity applications. Traditionally, these software applications were required to be installed physically on a local machine These services have now largely migrated to the cloud. Benefits include improved security through the use of more strict login requirements and authentication (many now requiring two-factor authentication) as well as increased file security and redundancy through cloud auto-saving functionality. Gone are the days of forgetting to save your Word document to the local drive and having to recreate an entire report. With cloud auto-saving and remote access from virtually anywhere, the legacy method has become nearly obsolete. In healthcare IT, cloud-based platforms are a necessity and institutions should also benefit from the technological advances available on these platforms.

Blockchain

Most recognizable for being the technology behind cryptocurrency, blockchain may eventually serve an appropriate function in the area of data integrity and a world without Electronic Health Records (EHR). Simply defined, blockchain refers to a type of data structure that enables identifying and tracking of transactions digitally and sharing this information across a distributed network of computers. The platform functions as a database on which numerous computers reside and function as “nodes” which, based on the number of nodes, determine the level of security. What is unique about the blockchain platform is how it functions as a ledger for all transactions. Any time there is an update or transaction that updates the database, that update is made and tracked on every computer (or “node”) on the network. Because of this ledger redundancy, it makes it extremely secure and very difficult for network compromise — as a violator would potentially have to hack a large number of nodes on the blockchain. As such, blockchain platforms with more nodes are going to offer even more security. This high level of security potential is where blockchain may eventually have a place in healthcare IT.

Financial systems use blockchain platform to enhance security and data integrity of transactions occurring across the system. This ultimately adds high levels of financial security through protection against hacking and minimizes financial losses. A great level of transparency is also established allowing for more complete transaction analysis.

A potential application for blockchain in healthcare may be as a potential EHR platform. Leveraging blockchain could allow for a very secure and personalized digital health chart that monitors all “transactions”occurring within a patient’s medical history or timeline with the benefit of being completely portable. An emergency room visit in California with various diagnostic imaging and lab testing is easily and securely retrievable by a primary care physician in Kentucky without the need for medical record requests or having to reinvent the wheel for the attending physician when assessing the patient’s medical history. There is potential for this EHR to be a more living, breathing EHR platform that actually becomes a non-EHR technology solution. The future of healthcare may indeed be an “EHR-less” world, and blockchain may very well be at the center.

Automation/AI

Artificial intelligence (commonly referred to as “AI”) is probably the most audible technology innovation in our lives right now. We see AI being used or touted in everything ranging from movies, to mega e-tailers, to finance industry, to law enforcement, to travel, and to logistics. Simply defined, automation is the use of technology and computers to reduce or minimize the amount of human resource requirement to perform a job or highly-repetitive task.

In healthcare, there are many repetitive tasks that occur daily, often requiring human input, and thus adding much time to workflows. A nurse navigator is constantly entering data into spreadsheets for purposes of a cancer screening program, but is having to re-enter many of those data components with each new patient event or update. Automation can reduce the human resource requirement by automatically populating required database fields, thus allowing the navigator to spend less time with data entry, and more time focused on patient care. This can allow for exponential program growth in both volume and quality metrics. Applying artificial intelligence offers even greater functionality and accuracy. A software application can rapidly comb through hundreds of thousands of patient charts and records, abstracting relevant data, analyzing the data, and then automatically populating that information into registry databases. This frees up the nurse navigator thus allowing for more time to be spent validating patient data and other components of the screening program, including community outreach, physician education, and brand exposure, ultimately improving the programs overall success. 

We see AI and automation throughout our daily lives, shipping and logistics industry. Shipping warehouses not only track and sort packages, but AI allows for the detection of more streamlined processes with minimization of errors associated with package routing. It is the reason the online ordering/shipping process has been reduced to hours and days when previously ordering/shipping could take weeks. This efficiency and error reduction should be applied to healthcare as well. Reducing costs associated with human resource requirements is imperative. Human resources should be utilized for program growth and quality initiatives. This ultimately improves patient care, quality, and exposure to resources that will save human lives.  

Stay tuned as we discuss how to tie it all together through co-creation in Solving the Data Conundrum Part 3 of 3.  Missed Part 1: Barriers to Innovation? Read it here.

 

About Thynk Health
The Thynk Health platform optimizes data-driven workflows and provides operational and clinical analytics for lung cancer screening programs and other quality initiatives.

1 thought on “Solving the Data Conundrum in Healthcare – Part 2”

  1. Pingback: Solving the Data Conundrum in Healthcare - Part 1 - Thynk Health

Leave a Comment

Your email address will not be published. Required fields are marked *