Over at AWS re:Invent 2019, Amazon has launched a new machine learning-powered service which can transcribe conversations between doctors and patients, effectively creating a digital copy of those consultations with relative ease.
Amazon Transcribe Medical automatically (and accurately) converts speech-to-text and recognizes medical terms, putting that text straight into the patient’s medical record. It can also pull off tricks like inserting punctuation so the doctor doesn’t have to worry about dictating commas, periods and so forth.
- AWS may finally be breaking up with Oracle
- AWS: Why we need to make machine learning happen faster
- These are the best cloud computing services of 2019
This is essentially a tailored version of Amazon Transcribe, an automatic speech recognition service that Amazon launched back in 2017.
Amazon Transcribe Medical can also be hooked up with Amazon Comprehend Medical, the latter being a natural language processing service which is capable of extracting all kinds of relevant medical information from any given transcription (like the details on medical conditions, medication and doses, frequency of taking, and so forth).
Better patient care
The big idea is that with all this taking place in the background, and being handled automatically, the physician can forget about note-taking, and concentrate fully on the patient and their needs. So even if the doctor only has limited time with the patient – as can often be the case these days – at least they can spend that time in a more productive and effective manner.
Amazon Transcribe Medical is now available to AWS customers, although only in certain parts of the US at the moment, namely in the US East (North Virginia) and US West (Oregon) regions.
Matt Wood, VP of artificial intelligence at AWS, told CNBC that the service is fully HIPAA compliant and its accuracy is ‘very high’, and that in the future, Amazon may publish research showing how it compares favorably to rival offerings.
- Best cloud storage of 2019: free, paid and business options