Amazon Transcribe Medical
Amazon Transcribe Medical is an automatic speech recognition (ASR) service that converts physician dictated notes, consultations between physicians and patients and remote assessments and consultations into text. The word transcribe is defined as the process of putting speech, thoughts or data into written or printed form. Amazon Transcribe Medical is trained to understand the complex medical language used by physicians; it allows developers to integrate accurate speech-to-text capabilities into their clinical documentation applications.
Amazon Transcribe Medical builds on the original Amazon Transcribe. The service uses machine learning (ML) and natural language processing (NLP) to capture physician-patient conversations in text. These interactions can then be entered into electronic health record (EHR) systems or other applications.
Transcribe Medical is HIPAA-eligible. It also offers a response to the HITECH Act -- a 2009 law that requires physicians to provide detailed documentation in patient health records. The new requirements from the act increased clinicians' workloads and reduced the amount of time spent with patients.
IT vendors, pharmaceutical companies and healthcare providers can use Amazon Transcribe Medical to create services that enable faster, more accurate and more secure documentation. This improves the efficiency and ease of medical notetaking for physicians, nurses, researchers and claims agents, allowing them to devote more time to their patients.
How does Amazon Transcribe Medical work?
Amazon Transcribe Medical uses a device's microphone to capture speech and then send pulse code modulation (PCM) audio to a streaming application program interface (API) based on the WebSocket protocol.
Next, the API responds with a series of JSON blobs -- or large files -- containing the transcribed text with elements such as punctuation, capitalization and word-level timestamps. Users will receive the transcribed text in real time. The produced data can be stored in an Amazon Simple Storage Service (S3) bucket.
After Amazon Transcribe Medical converts the speech to text, it gives the transcription a timestamp and a confidence score for accuracy. The transcription is then sent to an application, such as an EHR system, or an Amazon Web Services (AWS) language service, such as Amazon Comprehend Medical.
Amazon Comprehend Medical uses ML to pull relevant medical data from the transcriptions, including information such as a medical condition or a medication dosage, strength and frequency. This extracted information can be used to summarize notes and assist clinical decisions, clinical trial management and revenue cycle management.
Use cases
Amazon Transcribe Medical is used for:
- Dictation. Physicians can record their notes by speaking into a mobile device during or after a patient interaction.
- Drug safety monitoring. Pharmaceutical companies can use Transcribe Medical to record calls between patients and physicians or physicians and pharmaceutical companies. This ensures the names of medicines or key terms that describe health side effects are captured correctly.
- Conversational transcription. Physician-patient conversations can be captured in real time without disturbing the interaction, allowing care providers to record any side effects or symptoms mentioned by the patient and include them in the health records.
Whom does Amazon Transcribe Medical benefit?
Amazon Transcribe Medical benefits healthcare professionals -- including nurses, physicians, researchers, insurers, claim agents and pharmaceutical companies -- as well as the patients they serve. For example, multinational biopharmaceutical company Amgen has started using Transcribe Medical to record and review calls between patients and providers to identify possible side effects of its drug.
While Transcribe Medical can be integrated into a variety of apps and installed on any device with a microphone, it is only available to AWS customers using the AWS cloud.
Benefits of Amazon Transcribe Medical include:
- Understands natural speech. Conversations between multiple people or dictation from a single speaker can be accurately transcribed. Auto-punctuation and capitalization allow users to speak naturally without pausing to speak out punctuation marks.
- Highly accurate. The service provides real-time speech-to-text transcription that is built to handle complex medical language. This ensures complicated terms -- like technical disease names and medications -- are captured correctly.
- Improves patient and practitioner experience. Providers can accurately and efficiently record their notes, eliminating the previous burden of time spent on tedious note taking. As a result, practitioners can spend more time focused on the patient and his or her needs.
- Reduces medical documentation costs. Transcribe Medical does not require licensing fees -- unlike other transcription services. The medical speech recognition service also eliminates the need for an expensive human scribe to take notes.
- Scalable. Clinical documentation apps using Amazon Transcribe Medical are scalable and can be deployed across thousands of medical organizations. This ensures users have access to consistent and accurate note taking regardless of their location.
In addition, when used for drug safety monitoring, Amazon Transcribe Medical will capture information that pharmaceutical companies can use to more efficiently analyze and identify drug-related safety concerns, such as the unrecognized side effects of a new medication.
Transcribe Medical increases Amazon's investment in the healthcare industry, specifically with advancements in the relationship between voice technology and medicine. In addition to their new medical speech recognition service, Amazon has launched a medication management service for Alexa that allows patients to request refills and receive medication reminders. Furthermore, Amazon has acquired several health startups and launched a healthcare service for their employees -- Amazon Care.
Amazon Transcribe Medical pricing
Amazon Transcribe Medical is a fully managed service that allows users to easily scale while only paying for what is used; there are no upfront fees.
The pay-as-you-go payment model charges customers based on the seconds of recorded audio transcribed per month. Billing is performed monthly with a rate of $0.00125 per second. Therefore, one hour of transcription costs $4.50.
Pricing examples are as follows:
- A 15-minute pre-patient visit conversation would cost $1.125.
- A physician spending 30 minutes dictating a note would cost $2.25.
- A telemedicine conversation that lasts 45 minutes would cost $3.375.
- A 90-minute medical consultation would cost $6.75.
In addition, AWS offers a free tier of service that includes 60 minutes per month for 12 months. The trial starts when the first transcription is made. Once the free usage expires -- or if the service use goes above the 60-minute limit -- the user begins paying the standard, pay-as-you-go rate.