Texts in healthcare and pharmacology

A lot of data in healthcare and pharmacology is represented in the form of free texts. There is high potential hidden in these texts. We have been working, e.g., with discharge summaries and summaries of product characteristics. We are able to extract meaningful information and provide it to the end users.

Medical records

Medical records contain complete unstructured data on patients. We automatically extract information about medication, medical history, symptoms and diagnoses and make them available in a database or a search application. We further analyze these data to discover potential healthcare associated diseases (HAI), to automatically classify the diagnoses into DRG groups etc.

Drug Encyclopedia

Drug Encyclopedia is a modern source of information about drugs for physicians and other professionals in healthcare. It is also suitable for the educated public want information about prescribed drugs. We have used modern technologies such as Linked Data for data integration and NLP for connecting information hidden in unstructured data sources.


  • Easily searchable information and reporting for hospital management.
  • Improved coding and higher reimbursements in DRG-based payment systems.
  • Better preventive measures based on machine learning analysis of risk factors.
  • Liability risk decrease (clinical guidelines adherence)

Process digitalization

Back-office is maybe not as sexy as marketing or sales, but it has a lot to say about business. There are piles of documents in the back-office area which hide a lot of information. This information can be extracted and used in decision making. We have experience with mining information from contracts, RFPs and other legal documents.

Contracts mining

There are piles of contracts in each company. It might be handy to organize the contracts and get the information in a structured form. We have been working on several types of contracts, mostly on leases. We were able to organize electronic documents in folders and then automatically extract information about contracting parties, prices, contract duration, automatic prolongations etc.

RFP processing

Some companies receive a large amount of RFPs (Requests for Proposals). Each one of them can easily contain several hundreds of pages. We help with organizing this information automatically. We are able to pinpoint important pieces of the RFP and also provide tools for transforming the RFP into a meaningful proposal with all possible collaborative features (commenting, messaging, versioning...).


  • Organization of the documents - classification, time information...
  • Automation means less errors in processing
  • Faster processing - less amount of manual work needed
  • Reduced operating costs
  • Extracted information is now easily accessible

Customer relationship management (CRM)

Managing relationships with customers is extremely important in these days. Traditional methods analyze structured data generated from systems. We are mostly concentrating on textual data, e.g., communication with clients through various channels or notes from sales representatives. We have been able to prove that the information contained in these sources improves marketing campaigns and lowers the churn rate when used in predictive models.

Processing clients' filings
and extracting information about clients

Everyone would like to know their clients well. A lot of information about clients is hidden deep in the text notes, e-mails, social network posts etc. We know how to process these sources and enrich the data warehouse with the extracted data. These usually contain social and demographic data, information about lifestyle and lifetime events, clients' relationships to the competition and their products etc. We also use results from psycholinguistics and can segment clients according to the style of their writing.

Each company receives clients' filings - such as complaints, objections, proposals etc. - in the form of a text. Their processing is mostly manual, slow and without the possibility for a fast reaction. We address these shortcomings with our solution. We classify the texts into the required categories and automatically assign them to the solvers. Thanks to the information extraction we know how to monitor topics or key words and how to make quick reviews via reports or allerts.


  • Extracted data combined with data in data warehouse result in better marketing campaings.
  • Decreasing churn rate by enhancing models with extracted data.
  • Clients' segmenting based on their psycholinguistic profile - personalized client care, higher customer satisfaction.
  • Improving processes by automatically classifying texts.