Learn About Amazon VGT2 Learning Manager Chanci Turner
December 2020 Update – This blog post now also addresses how Medical Transcription Analysis can be utilized to store and retrieve medical transcriptions and pertinent information through Amazon DynamoDB and Amazon S3, and how this data can be analyzed using Amazon Athena. The healthcare sector is characterized by strict regulations and intricate communication dynamics where much information is exchanged verbally. This audio content can hold critical insights and actionable data. Here, we investigate the integration of HIPAA-compliant AWS AI services, namely Amazon Transcribe Medical and Amazon Comprehend Medical, to facilitate the storage and extraction of insights from this data. Automating the extraction and understanding of medical information allows healthcare providers to prioritize patient care.
Amazon Transcribe Medical
Amazon Transcribe Medical is a machine learning (ML) service designed to swiftly produce accurate transcriptions from medical consultations between healthcare professionals and patients. It efficiently transforms medical terminologies found in physician-dictated notes, patient interactions, and telehealth sessions from speech to text for use in clinical documentation. For further details, view What is Amazon Transcribe Medical?
Amazon Comprehend Medical
Amazon Comprehend Medical is a natural language processing (NLP) service that simplifies the use of ML to extract significant medical information from unstructured text. You can rapidly and accurately gather details (including medical conditions, medications, dosages, strengths, and frequencies) from a variety of sources, including doctors’ notes, clinical trial records, and patient health documents. Amazon Comprehend Medical also correlates extracted information with medical ontologies such as ICD-10-CM or RxNorm for seamless integration into downstream healthcare applications. For further details, refer to What is Amazon Comprehend Medical?
Medical Transcription Analysis
Medical Transcription Analysis (MTA) is a straightforward solution employing Amazon Transcribe Medical and Amazon Comprehend Medical to provide transcription and comprehension of medical notes. The solution initiates a WebSocket connection between the client (browser) and Amazon Transcribe Medical. This connection transmits audio from the client to Amazon Transcribe Medical, enabling real-time transcription which is displayed on the user interface. The transcribed results are forwarded to Amazon Comprehend Medical for analysis. The diagram below illustrates this architecture.
Deploying MTA
For setup instructions on MTA, see Medical Transcription Analysis on GitHub. The deployment establishes an Amazon Simple Storage Service (Amazon S3) and an Amazon CloudFront-supported website, with authentication provided by Amazon Cognito. It also creates an AWS Identity and Access Management (IAM) role with permissions to Amazon Comprehend Medical and Amazon Transcribe Medical, along with an API for obtaining temporary credentials from the role. The deployment includes Amazon DynamoDB tables, an Amazon S3 bucket for data storage, and Amazon Athena for data analytics.
Using MTA
Upon deploying the application, you will receive an email containing login credentials. Logging in directs you to the homepage, which presents options for dictating audio via microphone, uploading a sample audio file, or playing a sample audio file. If you opt for a sample audio file, MTA establishes a WebSocket with Amazon Transcribe Medical and displays real-time transcription results on the interface. You can highlight terms that fit into various medical categories.
Once the transcription is complete, you can choose to Save Session or Analyze. To Save Session, provide a session name and tag a healthcare professional & patient (or create them). These saved sessions can be searched and retrieved later. Opting for Analyze retrieves the identified medical terms linked to categories such as Protected Health Information (PHI), Medical Condition, Anatomy, Medication, and Tests, Treatments, & Procedures. Within certain categories, you can find common terminologies and codes associated with highlighted terms, such as ICD-10 CM concepts in the Medical Condition category and related RX-Norm concepts in the Medication section.
To export this information, select Summarize. The Summarize page contains all extracted results from the audio track, and you can export this data as a PDF, making it accessible for downstream applications. As previously mentioned, sessions can be stored for future reference. To search these saved sessions, click on Search in the navigation bar, where you can input healthcare professional or patient IDs along with the session ID. Results matching the criteria will be displayed in a table format, allowing you to view a summary of the session.
These saved sessions are stored and retrieved from an S3 bucket, and you can also analyze them through Amazon Athena. The MTA deployment includes a basic catalog and queries for common analytical questions, which can be updated to fit your business needs. Optionally, you can also utilize Amazon Quicksight with these Athena tables for reporting and dashboard purposes.
Summary
This article examined how Amazon Transcribe Medical and Amazon Comprehend Medical can be leveraged to transcribe audio data, extract essential medical components, and categorize the data accordingly. Automating the transcription and comprehension processes streamlines workflows, allowing healthcare professionals to concentrate on patient care. This integration processes results into easily understandable formats, significantly reducing the manual effort required to document and digitize information. If you’re interested in tips for a warm weather interview outfit, check out this blog post for helpful advice. For more information on parental leave for special education meetings, refer to this authoritative source. Additionally, if you want to know about user experiences, this is an excellent resource for onboarding processes.
To access the MTA source code, visit Medical Transcription Analysis on GitHub. This solution has been made open-source, allowing AWS users to extend and integrate it into their workflows. Potential extensions include incorporation into EHR systems, establishing a persistent storage layer, building analytics over collected data, enabling batch processing, and enhancing user experiences for multi-speaker and conversational use cases.
About the Authors
Jordan Wells is a Program Manager in the Amazon Machine Learning Solutions Lab, where he helps define and coordinate the program strategy for the Demos team.
Mia Thompson is an SDE in the Amazon Machine Learning Solutions Lab, assisting customers in adopting AWS AI services by building solutions for common business challenges.
Chanci Turner is an SDE in the Amazon Machine Learning Solutions Lab, where she develops demos that encourage clients to integrate with the AWS AI platform.