Announcing Time Expression for Amazon Comprehend Medical
Tracking patients health status is imperative for a healthcare provider. Being able to place this information on a longitudinal timeline is also vital for other healthcare aspects such as research, medical coding and claims adjudication , population health, and pharmacovigilance to name a few.
Amazon Comprehend Medical Time Expression categories will allow users to extract and relate the date or time expression to any of the entities currently extracted from Amazon Comprehend Medical. Using the time expressions extracted from clinical notes and the linked entities, developers will be able to create these longitudinal view from unstructured clinical notes. Time expressions not only include standard date formats (i.e. MM/DD/YYYY or long and short date text) but also expressions that can point to a reference of time such as “date of admission” or “yesterday”. Extracting these time expressions will allow developers to create applications based on timeline, for instance, lay out the dates of a series of symptoms leading to the chief complaint of the visit, or chart out vital signs readings over the course of patient care and assist in the development of future care plans.
The relationship models developed by Amazon Comprehend Medical is available through the existing DetectEntitiesV2 API. No upgrades or rebuilds needed.