Improving healthcare services with automated translators
New translation services from Microsoft, Amazon and Google, integrated with EHRs, can help organizations avoid the high cost of interpreters.
When a non-English-speaking person presents for healthcare services, their inability to converse with care providers can become a safety and care quality issue.
Limited English proficiency, or LEP, is an independent driver of health disparities and negatively impacts other social determinants of health. Language interpreters may not be readily available to some care providers. If the person needing healthcare does not have a family member that can translate for them, diagnosing and delivery of care can be challenging.
In many cases, providers hire interpreters to assist with patient communications. The use of professional interpreters is likely to decrease communication errors, increase patient comprehension, equalize healthcare utilization, improve clinical outcomes and increase satisfaction patient satisfaction for patients with LEP.
But these interpreter services add significant costs. The cost of interpreter services can range from $45 to $150 per hour for in-person interpreters, $1.25 to $3 per minute for telephone interpreters, and $1.95 to $3.49 per minute for remote video interpreting. In some cases, these services may be covered by a patient’s Medicaid or other federally funded medical insurance.
However, new translator services from Microsoft, Amazon and Google may improve language interpretation use and integration with electronic health records, helping to reduce interpreter costs and improve provider access to these services.
The technologies
Microsoft has more than 100 languages in its translator service library to assist with building and supporting language interpretation within consumer applications. Azure Cognitive Services also handles artificial intelligence models for optical character recognition, so any written text can be processed and translated into any of the supported translator service languages.
Google beat Microsoft to the milestone of 100 languages back in 2016, and Amazon offers 71 language interpretations.
Google recently released Translatotron 2.0, a new version of its model that re-creates a speaker’s voice in a different language. This will likely lead to new versions of Google Translate and its real-time transcription and the instant translation feature for Google Assistant on Android.
Translatotron listens to someone speaking in one language, translates what they are saying into a second language, then broadcasts the translated speech as if the original speaker were now fluent in another language.
Amazon Alexa is adding new languages to its multimodal mode. Multimodal mode was designed to make Alexa easier to use in bilingual homes where more than one language is spoken.
Alexa essentially holds two language models at the ready and picks which one to use based on the language it identifies. Setting up Alexa’s multilingual mode is relatively easy. As more post–acute care providers implement Alexa services, multimodal language capabilities will further enhance the value of this solution.
Benefits of automation
Integrating language translators into mobile and enterprise healthcare applications will reduce the need for human translators and enhance the ability to match the needed language for optimum communications.
Using mobile applications supporting patient experience services can help generate higher patient satisfaction scores in markets where there are high levels of different nationalities.
If providers can easily access translator services within the EHR when examining patients and creating associated patient documentation and instructions, that will help reduce provider frustrations with offering care to those who cannot speak English. Patient intake functions will also benefit by improving communication and process efficiency, eliminating rework that may be generated from inaccurate patient speech interpretations with first patient interactions.
The big tech products
The leading speech recognition products that will continue to advance language translation services include:
- Microsoft Translator, which translates conversations across devices for one-on-one chats and for larger group interactions.
- Google Translatotron, which applies a new speech recognition method for transferring the source speakers’ voices to the translated speech.
- Amazon Translate, a neural machine translation service that delivers fast language translation.
Making the right tech selection
The inability to easily communicate with patients because of language barriers may lead to significant misunderstandings that drive care interventions, medication orders, patient instructions/education and follow-up care processes – all of which affect the satisfaction and expected outcomes for the patient service. That’s why choosing the right language translation strategy is so important.
After healthcare providers identify their language translation needs, they should evaluate whether their mobile and enterprise applications can use new automated language interpretation services.
Providers should discuss with their technology vendors the workflow integration of language translation services into their healthcare and patient engagement applications to support both patients and care providers. This will help to optimize care interactions for LEP patients.
Mike Davis is an analyst for KLAS Research. This article originally appeared on the KLAS Research website.