Dôvera, a leading health insurance provider in Slovakia, needed to extract structured data on medication dosage for various patient categories from package leaflets, which are typically available only as free-form text. Traditional parsing methods failed due to the high variability in formatting and language use.
Dosage information in these leaflets is:
Presented in unstructured text (sentences, tables, or notes)
Expressed in varied linguistic forms, making rule-based parsing ineffective
Dependent on multiple factors such as age, weight, and patient condition
The client needed to transform this information into a structured format to use it for:
Internal analytics (e.g., comparing prescriptions to standard dosage guidelines)
Control mechanisms (e.g., detecting dosage deviations by age group)
We designed and implemented an AI solution that:
Automatically retrieves package leaflets for medications in a defined ATC group from public sources
Uses a language model to extract dosage information based on various parameters (e.g., for children, adults, seniors)
Outputs a structured dataset that categorizes dosage by patient group, method, and frequency of administration
This output is ready for further use in the client’s BI systems or control tools.
„We gained access to data that previously didn’t exist in any structured form. The solution from Insynaps opened up new possibilities for working with dosage information and strengthened both our analytical and control processes.“
Dôvera Health Insurance Company
- The client gained a precise, machine-readable overview of dosages for selected medications—data that previously couldn’t be systematically extracted.
- Significant time and labor savings, eliminating the need for manual review and transcription of dozens of documents.
- New opportunities emerged for prescription audits, comparisons with treatment standards, and care optimization.
The solution is scalable to other drug groups or control scenarios.