Using FHIR to integrate verbal production test results

Language can be defined as biometric data, and that is changing how cognition and mood are assessed in the clinical setting.

A FHIR-based approach aims to better collect information on language issues, moving data off paper and into a format to enable quantifiable analysis.

How language disorders are assessed in the clinical setting will be revolutionized in the coming years, thanks to technology and standards based on the Fast Healthcare Interoperability Resources Standard.

The recent advances in Natural Language Processing (NLP) techniques, Voice Analysis (VA) and Automatic Speech Recognition (ASR) software are changing the way we look at language dysfunctions, a common feature in many neurological disorders such as dementia.

Language requires the simultaneous activation of multiple cognitive systems, making it a non-invasive, easy-to-collect, rich source of amnestic information. Many language tests and metrics still used today are paper-based, but novel digital mobile-based tools to deliver and analyze speech and language data for clinical and research purposes are now available.

New approaches

GATEKEEPER is a European project that seeks, as one of its goals, to build a trust-based and secure platform to foster large-scale deployment of integrated digital solutions for early detection and intervention in different regions across Europe and worldwide, enabling novel business models.

In the context of this project, the HL7 FHIR standard has been used to integrate data produced by the SPEAKapp application into the GATEKEEPER platform. SPEAKapp enables the collection of the user's verbal production through a set of standard tests for the purpose of extracting clinically relevant acoustic and semantic features. This is done by processing raw audio and the text content.

SPEAKapp has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement No. 857223.

As a first step, a set of FHIR profiles, based on the observation resources, have been specified and published in the on-develop project FHIR IG. This set includes a profile used to describe the group of measures associated with the verbal tests available in the app.

This profile uses four other observation-based profiles, each describing the different measures extracted from the user's verbal production. The observations comprise variables related to the number and type of words produced, acoustic variables related to the audio produced, phonation and silence parameters and Natural Language Processing and semantic features.

Transaction bundles have been used to send patient verbal test results, via VPN, to the GATEKEEPER FHIR server. Data collected then are made available to any authorized users of the GATEKEEPER platform for research and clinical use in a secure, standard and interoperable way.

Potential benefits

Integrating novel indexes – based on verbal performance with standardized clinical measures – will lead to novel insights into mental health conditions and the identification of light and reliable indexes of cognitive functioning.

In the long term, identifying a valid marker of treatment efficacy would support clinical research and innovation, facilitating the evaluation of new drug efficacy, the design of targeted approaches and, in the long run, the mental health of the population served.

The defined FHIR observation profile is designed to foster a shared approach in language disorders’ technology-based assessment addressing acoustic and semantic features. The compliance of SPEAKapp to the HL7 FHIR standard, facilitating integration in commercial EHRs, is expected to accelerate the adoption of these novel techniques in clinical practice, assuring by design the conformity to all the mandatory privacy and security regulations while leaving the focus only on the scientific and clinical meaning of the selected variables.

This article originally appeared in HL7's Newsletter.

More for you

Loading data for hdm_tax_topic #better-outcomes...