Data extracted from Amazon Comprehend Medical, a software that mines patient health records, can now be linked to select medical ontologies, according to an Amazon Web Services blog post published this week.
Amazon launched Comprehend Medical in November 2018. The software uses natural language processing and machine learning to highlight key data points, including medical condition and clinical trial reports, from EHRs and clinical notes. Physicians can then use the data to inform a patient's care.
The ontology-linking application programming interfaces allow clinicians to detect medication and medical conditions in unstructured clinical text. Amazon now supports links with the following ontologies through Comprehend Medical: ICD-10-CM and RxNorm.
ICD-10-CM is a diagnosis code set that can be applied to population health analytics; the code set can identify medical conditions as entities and connect related information such as diagnosis and severity as attributes of the particular entity.
RxNorm allows healthcare providers to create an accurate, up-to-date list of all possible medications a patient is taking by identifying medications as entities and linking attributes including dose and frequency to that entity.