Artificial Intelligence (AI) success stories are becoming more common within the healthcare industry, as healthcare leaders are faced with finding new ways to control rising costs, identifying new revenue streams, and improving patient care.
Many are exploring innovative AI technology to augment existing workflows, automate manual processes, and help staff members be more effective and efficient with their time. In the recent webinar “Leveraging AI to Solve Common Healthcare Challenges Hear from the Experts” (watch now on demand), hosted by Healthcare Innovation, three experts from Wolters Kluwer discussed how AI can address these challenges. The experts addressed how AI-powered software solutions, including natural language processing (NLP) and machine learning (ML), can help optimize the tedious process of reviewing patient medical records to more accurately inform high-value organizational initiatives and drive improved patient care.
A foundation of quality data is key for artificial intelligence in healthcare
Sarah Bryan, Director of Product Management, Health Language at Wolters Kluwer, explained that enriched healthcare data is key to high-value organizational initiatives. But while there is no shortage of data within the healthcare industry, it is often disparate and “messy.” Being able to leverage claims data, clinical data, and even emerging data types, is powerful for any healthcare organization and enables a more complete 360-degree view of an individual’s health. To do this, data must first be consolidated from across the various data sources incorporating all data types. Equally important, data must be normalized to interoperability standards such as SNOMED CT®, RxNorm, LOINC®, and HL7. Finally, data must be enriched by extracting and normalizing clinical concepts locked in unstructured notes.
Once healthcare organizations establish a foundation of high-quality, semantically interoperable data, they can leverage the value of their own data assets using emerging technologies such as AI.
How are healthcare organizations using natural language processing and machine learning?
Dr. Chris Funk, Senior Medical Informaticist, explained that AI is a spectrum of applied intelligence. Starting with assisted intelligence AI can be used for tasks such as lane correction in an automobile; using cognitive intelligence AI can mimic human behavior. For the most part, healthcare is applying augmented intelligence AI to support human decisions and augment existing workflow.
While AI technologies like machine learning and natural language processing are being applied in a variety of areas within the healthcare space, Chris reminded us that AI’s success relies on accurate and complete data, as Sarah had explained earlier in the webinar.
Healthcare organizations may have access to clinical data, but much of the most valuable information is captured in unstructured text: 80% of patient data may be locked in unstructured text such as free text fields within electronic health records as notated by clinicians at the point of care, and therefore unusable. To access unstructured data, healthcare professionals must manually review each patient’s medical record, often an inefficient, time-consuming, expensive, and error-prone task.
The key to improving this process is utilizing a more specialized type of AI technology called clinical natural language processing (cNLP). Different from the more traditional NLP models, cNLP is tailored to understand the nuances of the healthcare domain.
To illustrate how cNLP can be applied, Chris walked us through three use cases where cNLP can extract clinically relevant patient data to inform and accelerate high-value initiatives including reviewing patient risk and quality measure reporting. Watch the webinar on demand to learn the details around these use cases.
Chris showed that in addition to physician notes, cNLP can offer value in areas such as journal articles, pathology reports, and consumer literature and educational pamphlets.
No matter the use case, the value cNLP brings is clear: automation of labor-intensive processes, optimization of existing workflows, increased staff efficiency, reduced manual labor and cost associated with administrative review, and increased accuracy of the data used to inform initiatives that impact reimbursement and influence patient care.
To further elaborate on the value of AI in healthcare, John Langton, Director of Applied Data Science at Wolters Kluwer, provided real-world examples of advanced AI projects that he has helped develop for WK clients, such as building intelligent content libraries, reducing readmission risk, enabling differential diagnosis, providing clinical decision support, predicting onset of disease, and targeting population health initiatives. John explains that once you unlock unstructured data, it opens a universe of possibilities where AI can be applied within your organization to bring value.
Moving forward with natural language processing in healthcare
As the healthcare industry evolves and places more focus on patient outcomes, technology will play a more important role in our ability to leverage new data sources and structures. Data generated from telehealth, chatbots, genomics, social determinants of health, digital assistants and smart devices, and wearables will all need to be leveraged to gather a 360-degree view of an individual’s health.
Now is the time to get started. Sarah outlined a three-step action plan for healthcare organizations looking to adopt advanced technology:
- Identify high-value initiatives that could benefit from improved data quality.
- Leverage cNLP, AI, and even non-AI technologies to accelerate data extraction.
- Empower machine learning models and AI investments with high-quality data.
Sarah informed the audience that Wolters Kluwer empowers healthcare organizations with its Health Language solutions, including Reference Data Management to establish a single source of truth for streamlined data governance. Wolters Kluwer also provides Interoperability and Data Normalization to map data to standard terminologies for semantic interoperability, and cNLP to extract and codify data locked in unstructured text for improved data accuracy and optimized workflows. Speak to an expert to learn how Health Language solutions can help your organization improve the quality of your data.
SNOMED CT® is a registered trademark of the International Health Terminology Standards Development Organisation (IHTSDO).
LOINC® is a registered trademark of Regenstrief Institute, Inc.