Natural Language Processing poses some exciting opportunities in the healthcare space to swim through the vast amount of data currently untouched and leverage it to improve outcomes, optimize costs, and deliver a better quality of care.

In the first part of this two-part series, we discovered the Drivers of NLP in Healthcare. The branch of AI seems to be critical for navigating through the growing volume of data already in silos and generated daily. The article outlines the factors that are driving the growth and implementation of Natural Language Processing in healthcare, the plausible benefits of the implementation and the future of Artificial Intelligence and Machine Learning in healthcare.

Let’s explore further how mature machine learning and AI are in the healthcare domain, the various ongoing and under-scrutiny use cases of natural language processing in healthcare, and a few real-life examples where these technologies are improving care delivery.

Natural Language Processing can be stated in layman terms as the automatic processing of the natural human language by a machine. It is a specialized branch of Artificial Intelligence which primarily focuses on interpretation as well as human generated data – text or speech based. The technology has various sub-disciplines, including Natural Language Query, Natural Language Generation, and Natural Language Understanding.

At the outset, when it comes to healthcare, the technology has two use cases:

• Comprehending human speech and extracting its meaning.
• Unlocking unstructured data in databases and documents by mapping out essential concepts and values and allowing physicians to use this information for decision making and analytics.

A majority of all additional use cases of machine learning and NLP in healthcare will sprout out of these two primary functions of the technology.

Use cases of Natural Language Processing in Healthcare

In a report by Chillmark Research, the company has outlined 12 use cases across three stages of maturity when it comes to use cases:

Mainstay use cases of Natural Language Processing in healthcare that have a proven ROI –

• Speech Recognition – NLP has matured its use case in speech recognition over the years by allowing clinicians to transcribe notes for useful EHR data entry. Front-end speech recognition eliminates the task of physicians to dictate notes instead of having to sit at a point of care, while back-end technology works to detect and correct any errors in the transcription before passing it on for human proofing. The market is almost saturated with speech recognition technologies, but a few startups are disrupting the space with deep learning algorithms in mining applications, uncovering more extensive possibilities.

• Improvement in Clinical Documentation – Machine Learning in healthcare has touched clinical documentation, freeing up physicians from the manual and complex structure of EHRs, allowing them to focus more on care delivery. This has been possible because of speech-to-text dictation and formulated data entry that capture structures data at the point of care. As machine learning in healthcare advances, we will be able to pull pertinent data from other emerging sources and improve analytics used to drive PHM and VBC efforts.

• Data Mining Research – The integration of data mining in healthcare systems allows organizations to reduce the levels of subjectivity in decision-making and provide useful medical know-how. Once started, data mining can become a cyclic technology for knowledge discovery, which can help any HCO create a good business strategy to deliver better care to patients.

• Computer-assisted Coding – NLP-driven CAC promises to improve coder accuracy. Computer-assisted coding extracts information about procedures and therapies to capture every code and maximize claims. Research has led us to discover that the current vendors in the market building CAC solutions will have to shift their solutions to meet the challenges of a value-based paradigm and ensure they are working as expected.

• Automated Registry Reporting – An NLP use case is to extract values as needed by each use case. Many health IT systems are burdened by regulatory reporting when measures such as ejection fraction are not stored as discrete values. For automated reporting, health systems will have to identify when an ejection fraction is documented as part of a note, and also save each value in a form that can be utilized by the organization’s analytics platform for automated registry reporting.

Emerging use cases of NLP and Machine Learning in Healthcare that will have an immediate impact –

• Clinical Trial Matching – Using NLP and machine learning in healthcare to identify patients for a clinical trial is an exciting and moreover an essential use case. A few companies are now trying to resolve the challenges in this area using NLP engines for trial matching. With current advancements, it looks like NLP has the potential to automate trial matching and make it a seamless process.

• Prior Authorization – A survey has revealed that payer prior authorization requirements on physicians are increasingly on the rise. These requests increase practice overhead and delay care delivery. The issue of whether payers will agree and authorize reimbursement might not be around after some time, thanks to natural language processing. IBM Watson and Anthem are already working on an NLP module used by the payer’s network to determine prior authorization quickly.

• Clinical Decision Support – Natural language processing and machine learning in healthcare can help physicians make better decisions. Certain areas in healthcare need better methods of surveillance, such as medical errors. NLP is also being used to aid clinicians in checking symptoms and diagnosis.

• Risk Adjustment and Hierarchical Condition Categories – Hierarchical Condition Category coding, a risk adjustment model, was initially designed to predict the future care costs for patients. In value-based payment models, HCC coding will become increasingly prevalent. HCC relies on ICD-10 coding to assign risk scores to each patient. Natural language processing can help assign patients a risk factor and use their score to predict the costs of healthcare.

Next-gen use cases that are on the horizon –

• Ambient Virtual Scribe – Clinical documentation needs speech recognition software that would completely do away with human scribes. When this happens, clinical documentation will become a game completely changed by NLP and Artificial Intelligence in healthcare.

• Computational Phenotyping and Biomarker Discovery – NLP can also potentially help physicians with the complexities of phenotyping patients for analysis. With NLP, phenotypes will be defined by the patient’s current conditions as opposed to the knowledge of the physician. NLP could also be used to analyze speech patterns and detect neurocognitive injuries such as Alzheimer’s, dementia, and other psychological conditions.

• Population Surveillance – An application of NLP to EMRs can be identifying a subset of an ethnic or racial group for eventually documenting and mapping health disparities. The existing administrative databases lack the granularity to determine critical socio-cultural differences and execute population surveillance. However, NLP presents a vital use case in the area for further research and advancement.

How Can Healthcare Organizations Leverage NLP?

Healthcare organizations can use NLP to transform the way they deliver care and manage solutions. Organizations can use machine learning in healthcare to improve provider workflows and patient outcomes.

Here is a wrap up of the use of Natural Language Processing in Healthcare:

• Improve patient interactions with the provider and the EHR – For their part, natural language processing solutions can help bridge the gap between complex medical terms and patients’ understanding of their health. NLP can be an excellent way to combat the EHR distress. Many clinicians utilize NLP as an alternative method of typing and handwriting notes.

• Increasing patient health awareness – Even when patients can access their health data through an EHR system, a majority of them have trouble comprehending the information. Because of this, only a fraction of patients are able to use their medical information to make health decisions. This can change with the application of machine learning in healthcare.

• Improve care quality – NLP tools can offer a better provision to evaluate and improve care quality. Value-based reimbursement would need healthcare organizations to measure physician performance and identify gaps in delivered care. NLP algorithms can help HCOs do that and also assist in identifying potential errors in care delivery.

• Identify patients with critical care needs – NLP algorithms can extract vital information from large datasets and provide physicians with the right tools to treat patients with complex issues.

Implementing Predictive Analytics in Healthcare

Identification of high-risk patients as well as improvement of the diagnosis process can be done by deploying Predictive Analytics along with Natural Language Processing in Healthcare along with predictive analytics.

It is vital for emergency departments to have complete data quickly, at hand. For e.g., the delay in diagnosis of Kawasaki diseases leads to critical complications in case if it is omitted or mistreated in any way. As proved by scientific results, an NLP based algorithm identified at-risk patients of Kawasaki disease with a sensitivity of 93.6% and specificity of 77.5% compared to the manual review of clinician’s notes.

A set of researchers from France worked on developing another NLP based algorithm that would monitor, detect and prevent hospital acquired infections (HAI) among patients. NLP helped in rendering unstructured data which was then used to identify early signs and intimate clinicians accordingly.

Similarly, another experiment was carried out in order to automate the identification as well as risk prediction for heart failure patients that were already hospitalized. Natural Language Processing was implemented in order to analyze free text reports from the last 24 hours, and predict the patient’s risk of hospital readmission and mortality over the time period of 30 days. At the end of the successful experiment, the algorithm performed better than expected and the model’s overall positive predictive value stood at 97.45%.

The benefits of deploying NLP can definitely be applied to other areas of interest and a myriad of algorithms can be deployed in order to pick out and predict specified conditions amongst patients.

Even though, the healthcare industry at large still needs to refine its data capabilities prior to deploying NLP tools, it still has a massive potential to significantly improve care delivery as well as streamline workflows. Down the line, Natural Language Processing and other ML tools will be the key to superior clinical decision support & patient health outcomes.

Final Word

We are already witnessing a vast amount of critical applications of conversational AI in healthcare, it is imperative that NLP is well and truly placed to improve healthcare delivery when it comes to better clinical decision making and improved patient outcomes. The various use cases of Natural Language Processing discussed here present an opportunity for the healthcare industry to break down antique silos and plug gaps in the care delivery system to make progress for the patient segment. Get in touch or write to us on hello@marutitech.com us to learn how Maruti Techlabs is enabling leading hospitals and healthcare providers across an extensive range of use cases with NLP and AI solutions.

Source: https://marutitech.com/use-cases-of-natural-language-processing-in-healthcare/