Skip to main content

Hands-On Lab: Amazon Comprehend Medical & Amazon Transcribe Medical in the AWS Console

Key Takeaways​

Amazon Comprehend Medical and Amazon Transcribe Medical provide interactive evaluation consoles within AWS to demonstrate how specialized healthcare AI processes complex clinical shorthand, anatomical references, pharmacology, and speech dictation without manual feature engineering.

The hands-on walkthrough illustrates:

  • Decoding Medical Shorthand: Comprehend Medical parses compressed clinical jargon (e.g., PT →\rightarrow Patient, 87 YO →\rightarrow Age: 87, PO →\rightarrow Per Os / Oral Route, QHS →\rightarrow Every night at bedtime).
  • Entity-Relationship Binding: Comprehend Medical does not extract terms in isolation; it establishes relational bindings connecting dosages, administration routes, and frequencies to specific drug names.
  • Integrated Console Architecture: Amazon Transcribe Medical is accessed directly within the standard Amazon Transcribe console under its dedicated left navigation menu.

Hands-On Workflow: Testing Comprehend Medical & Transcribe Medical​

  1. Launch Real-Time Analysis in Amazon Comprehend Medical:
    • Open the Amazon Comprehend Medical Console in a supported AWS Region.
    • In the left navigation pane, select Real-time analysis (or click Launch real-time analysis from the service dashboard).
    • Locate the sample clinical input text editor containing unstructured physician notes.
  2. Analyze Unstructured Clinical Shorthand:
    • Review the sample doctor's note containing clinical abbreviations and acronyms:
    • Notice compressed clinical phrases such as:
    • Click Analyze to submit the raw text payload to the clinical NLP engine.
  3. Inspect Clinical Entities, Attributes & Relationship Mappings:
    • Review the extracted structured entities and their visual relationship connectors:
    • Demographics & PHI: Observe 87 classified as AGE, woman classified as GENDER, and high school teacher identified as PROFESSION.
    • Relationship Extraction: Inspect how the dosage (20mg) and the administration route (PO / oral) are visually and relationally linked to the medication name (Atorvastatin).
    • Temporal & Acuity Traits: Notice how symptoms (e.g., chest pressure) are mapped with acuity and linked temporally to today.
      !
  4. Navigate to Amazon Transcribe Medical in the Console:
    • Switch to the Amazon Transcribe Console:
    • In the left navigation pane, scroll down to the Amazon Transcribe Medical section.
    • Select Real-time transcription.
    • Select the transcription mode (Dictation for single-clinician notes or Conversation for doctor-patient discussions).
  5. Execute Live Medical Speech Transcription:
    • Test real-time medical Automatic Speech Recognition (ASR):
    • Click Start streaming and grant browser microphone permissions.
    • Speak a medical sentence into the microphone: "I have abscess near my knee after I fell down yesterday."
    • Observe the transcription output pane rendering medical pathology and pharmacology terms accurately without phonetic misinterpretations.
    • Click Stop streaming.

Shorthand Decoding & Entity Relationship Mapping​

Clinical Shorthand / TermLatin / Clinical MeaningComprehend Medical Extracted TypeAssociated Relationship / Target Entity
PTPatientPROTECTED_HEALTH_INFORMATIONAnchors clinical record context
87 YO87 Years OldAGE (PHI category)Binds to demographic profile
PO (per os)Orally / By mouthROUTE_OR_MODELinked directly to target MEDICATION
QHS (quaque hora somni)Every night at bedtimeFREQUENCYLinked directly to target MEDICATION
PRN (pro re nata)As neededFREQUENCY / TRAITLinked directly to target MEDICATION
BID / TID / QIDTwice / 3 times / 4 times dailyFREQUENCYLinked directly to target MEDICATION

Exam Guide​

Exam Tips​

  • Console Location of Transcribe Medical: Amazon Transcribe Medical is hosted within the Amazon Transcribe console interface under a dedicated navigation header.
  • Relationship Extraction: A primary differentiator of Amazon Comprehend Medical is its ability to perform relationship extraction—linking attributes like dosage, route (PO), frequency (QHS), and duration directly to their respective medication entities rather than treating them as disconnected tokens.
  • Clinical Shorthand Support: Pre-trained medical NLP models natively recognize standardized medical acronyms, Latin prescription notations (PO, QHS, PRN), and clinical shorthand (PT, YO, Dx, Rx).
  • HIPAA Eligibility: Both Transcribe Medical and Comprehend Medical are HIPAA-eligible out of the box for handling Protected Health Information (PHI).

Practice Test​

Question 1​

A clinical software developer tests an electronic medical record ingestion tool using Amazon Comprehend Medical. The input clinical note contains the line: "Rx: Lisinopril 10mg PO QD for hypertension." How does Amazon Comprehend Medical parse and structure this data?

  • A. It outputs an unlinked list of keywords: Lisinopril, 10mg, PO, QD, hypertension
  • B. It identifies Lisinopril as a MEDICATION and establishes directional relationships linking 10mg as Strength, PO as Route, QD as Frequency, and hypertension as the linked MEDICAL_CONDITION
  • C. It translates the Latin acronyms into French using Amazon Translate
  • D. It routes the note to Amazon Mechanical Turk for manual transcription
Correct Answer
  • B. It identifies Lisinopril as a MEDICATION and establishes directional relationships linking 10mg as Strength, PO as Route, QD as Frequency, and hypertension as the linked MEDICAL_CONDITION
    • Explanation: Amazon Comprehend Medical performs advanced entity and relationship extraction, binding medication attributes (strength, route, frequency) and diagnoses directly to the target medication entity within a structured JSON output.

Question 2​

Where in the AWS Management Console can a developer access and configure real-time streaming sessions for Amazon Transcribe Medical?

  • A. Inside the Amazon Rekognition console
  • B. Under the Amazon Transcribe console in the dedicated Amazon Transcribe Medical navigation section
  • C. Exclusively via the AWS Systems Manager console
  • D. Inside the Amazon Comprehend Custom Classifier builder
Correct Answer
  • B. Under the Amazon Transcribe console in the dedicated Amazon Transcribe Medical navigation section
    • Explanation: Amazon Transcribe Medical is integrated directly into the Amazon Transcribe Console under the left navigation section labeled Amazon Transcribe Medical.