In 2009, I managed an electronic health record (EHR) implementation project for an Independent Practice Association (IPA) and was instrumental in transitioning them from a paper-based system to a paperless system. If you are familiar with the complexity of the transition to an electronic system and its qualification for Meaningful Use 1 incentives, then you know the painful journey of health record abstraction. It involved outsourcing, adding resources, squeezing time from busy schedules to make room for abstracting individual charts and saving it as a discrete data for quality reporting, research or any secondary use. Scanning modules within the EHR only attributed to storage cost and unstructured data that could not be easily interpreted for clinical decision making. The manual process was limited in its value proposition.
Eleven years later, the healthcare challenges remain the same and have grown multifold in complexity due to data explosion. However, what has changed exponentially is technology. With the advent and evolution of advanced healthcare technology, handling these challenges is much easier. Health record abstraction which was once only limited to extracting, categorizing, searching and basic reporting in the EHR, is now helping healthcare professionals to predict clinical outcomes, personalize treatments, and maximize revenue with the help of latest technologies.
The latest technologies powered by computer vision can now extract information from a wide variety of unstructured, semi-structured and structured patient records such as documents, images, and videos. Natural language processing (NLP) techniques help in speech analysis, extracting information from transcription notes. Our Artificial Intelligence/Machine Learning algorithms help our clients abstract intelligent actionable insights from data accurately and rapidly. Visualization tools make data analysis effective in identifying and understanding trends, patterns, or correlations.
It’s amazing to see the endless possibilities that these hidden insights provide. For one of our oncology healthcare providers, our team built a disease recurrence prediction model using the latest technological advancements. They wanted an efficient solution to identify cancer recurrence early rather than waiting for more than a year. We developed a solution that would predict cancer recurrence at the time of encounter and more accurately thirty days out. NLP based transformation models allowed for inclusion of unstructured data. This enabled our client to take critical decisions on time and save lives.
Providers can also maximize revenue with the help of abstraction technology. Speech Recognition and NLP techniques abstract information from a broader set of data (clinical notes, discharge notes, transcription notes) for increased coding accuracy and transparency. The segmentation models help auditors review prioritized high value charts to facilitate compliance with payer rule books and quality reporting.
With the wide variety and volume of data available from multiple sources, extracting insights from the right datasets becomes pertinent. One of our clients, a global biopharmaceutical company, had a centralized repository that stored documents related to chemical processes, and drug production generated by its R&D team but did not have a system that could effectively derive insights or answer queries using these documents. We designed a search tool powered by Google Knowledge Graph/Base technology that allowed the users to efficiently search and confidently fetch information in the chemistry domain. At Quantiphi, we constantly work on innovative solution offerings so that our clients can succeed in their goals. Embrace the era of Big Data and take a meaningful step towards digital transformation!
Contact us to learn more about how intelligent insights from health records can improve patient outcomes, lower cost, reduce clinician and administrative staff burnout and enhance patient experience.