Menu
Back

Relative Health: Exploring the language of medicine

Medical language is one of the most difficult to process using machine learning.

Because of the highly specialized vocabulary, oftentimes conventional topic analysis is not suitable for the healthcare industry. For example, words, like administer or apply, may be tagged as work or office-related topics, rather than medical.

This posed a significant challenge when analyzing healthcare-related text data, so our dedicated team of data scientists explored different ways of processing language that would be tailored to medical English, with a specific focus on patient conversation, which aligns with the usage of our customers.

Introducing Relative Health

Through the hard work of our diligent data scientists, Relative Health has been born. Aimed specifically at the pharmaceutical, healthcare, biotech and medical device companies and agencies, Relative Health helps them analyze and visualize their text data better than ever before.

Rather than repurposing a broader text analysis solution for the medical world, we have created a specialized healthcare processing pipeline. This allows us to categorize complex, industry-specific language into topics, phrases, words, grammar, and emotions in a meaningful and contextually relevant way, resonating with those working in the healthcare industries without the need to rely on work-around solutions and interpretation. 

33 unique medical tags, available only in Relative Health, will have a small blue Rod of Asclepius symbol (⚕) appear alongside them to set them apart from our general topical analysis. 

These include topics such as:

  • Administration
  • Medical abbreviations
  • Disease (including infectious, anatomical, genetic etc.)
  • Syndromes
  • Symptoms (including respiratory, reproductive, cardiovascular, and more)
  • Alternative medicine
  • Over the counter
  • Physical therapy
  • Procedure
  • Psychotherapy
  • Dietary supplements
  • Surgery
  • Treatment 

Medical lexicon in action

To illustrate this, we have run a comparison to explore how Type 1 and Type 2 diabetes are discussed on medical forums and social media. Type 1 diabetes is a genetic condition that often shows up early in life, and Type 2 is mainly lifestyle related and develops over time.

What we found is that in discussions around Type 2 diabetes, people are more likely to talk about lifestyle, with nutrition discussed 2.5x more frequently and food appearing 2.2x more. 

However, in Relative Health, we also see specific medical topics such as treatment, dietary supplements, and exercise appear. These complex medical terms are recognized and classified accordingly to provide meaningful and relevant context for health and pharmaceutical companies and agencies. 

We debuted Relative Health at the IIEX Health 2022 in Philadelphia, demonstrating its potential to healthcare specialists looking to do more with their text data.

Relative Health will be rolled out to our customers in the near future – ensure you’re one of the first to observe the increased capabilities offered by our new health lexicon by requesting a demo now.