Real-life studies accelerated thanks to the extraction of structured data in medical databases
Real-life studies play a crucial role in evaluating medical treatments and allow for a better understanding of their efficacy and long-term effects. However, these studies are often costly and time-consuming, which limits their feasibility. Our dedicated solutions and support enable the automation of many data collection and analysis tasks, which significantly reduce the efforts and costs involved, with controlled precision
1. Study objectives taken into account
We adapt the models used to the objectives of your study (in terms of entities to be extracted and expected level of precision). Our language models are retrained on annotated datasets that are most relevant to your study. For certain entities, manual annotation of the datasets will be necessary (for around 10% of the study's data volume)
2. The ability to analyse 100% of CRF and medical CR of DMP
The language analysis technologies we develop enable us to analyse thousands of reports in bulk. The precision achieved varies depending on the entities to be extracted, it is generally in the range of 85% to 95% before specific retraining and 90% to 99% after retraining adapted to your study
3. Investigations can be supported by our PraxyConsultation tool
In the event of patient calls to conduct investigations, our PraxyConsultation solution for interview summary and structured data extraction can also be mobilised to save time and improve precision See PraxyConsultation . These data are, of course, reviewed and validated by the investigator
HDS SaaS or on-premise solution
Pay-per-use result
depending on the volume of medical reports analysed and useful entities for your study that will be extracted
Controlled precision
We evaluate the precision of each entity to be extracted before and after retraining our models. The thresholds are adapted to the challenges of your study
Scalability and speed
The solution is quickly deployed to process thousands of reports or databases