| Abstract: |
1 INTRODUCTION
Hearing loss (HL) is the fifth leading cause of years lived with disability (GBD 2015 Disease and Injury Incidence and Prevalence Collaborators 2016). Big Data is still underrepresented in hearing-related applications. Internet of Things utilizations slowly make their entrance to the market and open up the possibility for smart applications that involve persons with hearing loss and hearing devices (Clark & Swanepoel 2014). Potential benefits of Big Data applications within hearing healthcare include hearing aid setting adjustments based on environmental factors and public health policy-making support through evidence (Spanoudakis et al. 2017).
Hearing aids (HA) are the main treatment for HL. Data that can relate to HL include, amongst others, physiological measurements, metadata such as personal information anamnesis, reading span score (measure of cognitive abilities), speech recognition abilities and device data, such as HA usage.
This project aims to gather different types of data available regarding research participants visiting a private hearing research centre and, with the help of big data techniques, explore associations to gain new knowledge. This new knowledge is relevant for the future research as well as for clinical care of people with HL.
The proposed poster describes the project methods and progress so far.
2 METHODS
The project is divided into two phases. Phase I, which is concluded, reviewed data types, formats, and locations. Phase II, which is ongoing, is collecting and processing the data.
Unlike classical scientific approach, where a hypothesis is described and then methods are used to confirm or negate the hypothesis, this project aims to collect research data accumulated over multiple projects and clinical care. It uses big data algorithms to make predictions as well as to support profiling of research participants.
2.2 Phase I
The first phase is a scouting phase where the emphasis is on finding critical points in the typical journey of a research participant from recruitment, to clinical care, to participation in hearing research. The participants’ personal identification number is the link between all data sources.
The sample includes 250 active participants with HL. Data sets include: metadata (personal data, HA type, smartphone possession) (n=250), audiogram (n=250), visiting dates (n=250), pupillometry recordings (n=24), reading span scores (n=200), HA usage logs (n=66).
Relevant participant data is located in a patient management system, as well as the HA fitting software NOAH by the Hearing Instrument Manufacturers' Software Association and Genie by Oticon A/S. Furthermore, digital data is stored on local computers as well as server hard drives.
2.2 Phase II
Phase II includes data collection and processing. After collecting relevant data and storing it according to current standards, unsupervised self-learning algorithms will be applied to find associations and to enable different usage including participant profiling, outcome prediction, and research support.
Predictions include the development over time of parameters of a research participant. Those can include the severity of the HL, the reading span score and the ability to participate in research activities.
3. OUTLOOK
Participant profiling is a mean to support research by finding ideal participants for specific experiments. It also has implications for prediction of hearing healthcare outcomes and for better personalisation of services. Mapping participants that are scoring well in e.g. speech understanding tests, or have available EEG recordings can ease the search for participants for a later experiment.
Furthermore, another goal of the project is to describe data handling in the future to ensure good practice with patient and research data.
REFERENCES
Clark, J.L. & Swanepoel, D.W., 2014. Technology for hearing loss-as We Know it, and as We Dream it. Disability and Rehabilitation: Assistive Technology, 9(5), pp.408–413.
GBD 2015 Disease and Injury Incidence and Prevalence Collaborators, 2016. Global, regional, and national incidence, prevalence, and years lived with disability for 310 diseases and injuries, 1990-2015: a systematic analysis for the Global Burden of Disease Study 2015. Lancet (London, England), 388(10053), pp.1545–1602. Available at: http://www.ncbi.nlm.nih.gov/pubmed/27733282.
Spanoudakis, G. et al., 2017. Public health policy for management of hearing impairments based on big data analytics: EVOTION at Genesis. Available at: http://openaccess.city.ac.uk/18205/. |