Challenges:
  • Very large textual content library – no metadata or categorization
  • Each article corresponds to multiple specialized medical conditions only experts can assess
  • Some conditions have very few articles (limited or no training data)
Solution:
  • Fokal Data Services collected other related content to increase input for modelling
  • Fokal NLP and ML Toolkit used to develop very accurate (more accurate than panel of doctors) Classification Model to match content to conditions and medications

Lorem ipsum dolor sit amet consectetur adipisicing elit?

Challenges
  • Very large textual content library – no metadata or categorization
  • Each article corresponds to multiple specialized medical conditions only experts can assess
  • Some conditions have very few articles (limited or no training data)
Fokal Data Services
  • Web Data Integration
  • Data Cultivation
  • Massively Parallel Processing
Fokal ML Services
  • Natural Language Processing
  • ML Toolkit
OUTCOMES
  • Text Classification Models
  • Articles classified into High / Medium / Low for the two POC conditions
  • High – The article is primarily about the given topic
  • Medium – The article is about one or more topics that are closely related to the given topic. OR The article is about multiple topics of which the given topic is one of them.
  • Low - The article indirectly references the topic or has some of the keywords associated with the topic, but is primarily about something else