Our client needed to fill in a Case Report Form with 70 variables from 2 pathologies, covering at least 5.000 patients from 8 hospitals.
Accessing and reading clinical records requires the patient’s informed consent.
Manual collection of information bears enormous costs and does not allow to collect the millions of clinical records available.
Only between 15-25% of all information contained in hospitals is structured. The rest, is in plain text.
Data extraction, using Natural Language Processing
Implemented simultaneously in all 10 hospitals participating in the study.
Electronic Health Records (EHR) Text Based
Structred Database through NLP Coded.
Extraction of variables.
A queries was run upon 10 hospitals at the same time, after authorization from hospital.
Complete dataset with anonymized information per patient and variables needed for the study is obtained.
Statistical analysis carried out by using this dataset.
High-quality observational study, with less effort and time.
150 times more data points
Than the ones that would be found with the current methodology.
Data obtained in 10% of time
Saving up valuable time and resources.
Dataset went through three independent QA filters, ensuring data excellence.
Access to the entire dataset
Instead of access to plain reports, avoiding “back box” logics.