NLP in Healthcare

04.09.2020

So you already know about NLP (thanks to this article) but as a biomedical engineer you might be interested on how to use NLP for healthcare issues.

Well, that is what we are talking about on the following article.

1. The context: about the Healthcare industry

As any other industry, the healthcare ibe relizes more and more the importance of data, as all the information is being digitalize, and how the fact of collecting that data becames essential in improving the life of patients. As part of the life sciences professionals, biomedicals engineer are also a product of that modernization and digitalization. Some of them might not use data at all, but others, those who work in the artificial intelligence field, will do, a lot. 

However, there are far more data than time to analyze it, and the manual process to identify that data would take an eternity. 

In order to not disturb the clinical workflows, and to have an accurate extraction of unstructred data, NLP is necessary. 

In order to make effective decision in healthcare through analyticis, there is this need to leverage this unstructured mixed data so we can shift to a value-based care system. 

There is where NLP intervenes, and its capacity to mimic human behavior, as for exemple summarization or analysis ability, so it can be to parse information and extract critical strings of data.

2. Some examples in the use of NLP focused on Healthcare

Here I present some of the most interesting applications of NLP to healthcare, but I invite you to go further and check more applications. At the end of this article, you will find a few extra links. 

  • Mapping data elements present in unstructured text to structured fields in an electronic health record in order to improve clinical data integrity.
  • Answering unique free-text queries that require the synthesis of multiple data sources.
  • Conducting speech recognition to allow users to dictate clinical notes or other information that can then be turned into text.
  • Summarizing lengthy blocks of narrative text, such as a clinical note or academic journal by identifying key concepts or phrases present in the source material.
  • Engaging in optical character recognition to turn images, like PDF documents or scans of care summaries and imaging reports, into text files that can then be parsed and analyzed.
  • Converting data in the other direction from machine-readable formats into natural language for reporting and educational purposes.

3. Data mining sources

Patient health records, order entries, and physician notes aren't the only sources of data in healthcare.

More than 20 millon  people  people have already added their genetic information to commercial databases through take-home kits.

And wearable devices have opened new floodgates of consumer health data. 

If we look at the data sources , we can etablish the following list. 

1. The Internet of Things (think FitBit data)

2. Electronic Medical Records/Electronic Health Records (classic)

3. Insurance Providers (claims from private and government payers)

5. Opt-In Genome and Research Registries

4. Other Clinical Data (including computerized physician order entries, physician notes, medical imaging records, and more)

6. Social Media (tweets, Facebook comments, etc.)

7. Web Knowledge (emergency care data, news feeds, and medical journals) 

4. Personal experience

For now, i have used NLP processes in order to extract relevant information in order to answer specific questions in a healthcare domain. Due to the world situation in 2020, the subject of my work has been Covid-19. 

Therefore, I have been focusing in Automatic Summarizations. This is interesting as the summaries always specifically answer questions about coronovirus, which provides a helpful hand to researchers that otherwise might be lost in the high amount of information. 

You can find more information about this project in the portfolio secion (still w.i.p)

5. Links of interest