Basic principles for the development of an AI-based tool for assistive technology decision making

Abstract

Introduction
The impact of assistive technology use on the lives of people with disabilities has long been demonstrated in the literature. Despite the need for assistive technologies, and a wealth of innovative, afford-able, and accessible products, a low rate of assistive technology uptake is globally maintained. One of the reasons for this gap is related to data and knowledge formation and management. Low access to information and a lack of assessment services is evident. Fragmentation of data, inconsistency in assessment methodology and heterogeneity in the competence of assistive technology professionals, has led to a growing interest in the opportunities that data sciences, including AI, hold for the future of the assistive technology sector, as a supportive and constructive mechanism in any decision-making process.

Objectives
In this short paper, we seek to describe some of the principles that such an AI-based recommendation system should be built upon, using the Atvisor platform as a case study. Atvisor.ai is an AI-based digital platform that supports assistive technology assessments and the decision-making process.

Recommendations
Our recommendations represent the aggregated insights from two pilots held in Israel, testing the platform in multiple environments and with different stakeholders. These recommendations include ensuring the continuum of care and providing a full user journey, incorporating shared decision making and self-assessment features, providing data personalisation and a holistic approach, building a market network infrastructure and designing the tool within a wider service delivery model design. Assessment and decision-making processes, crucial to optimal uptake, cab be leveraged by technology to become more accessible and personalised.

•

IMPLICATIONS FOR REHABILITATION
• Provides principles for the development of an AI-based recommendation system for assistive technology decision making.

• Promotes the use of artificial intelligence to support users and professionals in the assistive technology decision making process.

• Personalization of data regarding assistive technology, according to functional, holistic and client centered profiles of users, ensures optimal match and better use of assistive technology.

• Self-assessment and professional assessment components are important for enabling multiple access points to the assistive technology decision making process, based on the preferences and needs of users.

Full text

Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=iidt20
Disability and Rehabilitation: Assistive Technology
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/iidt20
Basic principles for the development of an AI-
based tool for assistive technology decision
making
Moran Ran , David Banes & Marcia J. Scherer
To cite this article: Moran Ran , David Banes & Marcia J. Scherer (2020): Basic principles for
the development of an AI-based tool for assistive technology decision making, Disability and
Rehabilitation: Assistive Technology, DOI: 10.1080/17483107.2020.1817163
To link to this article: https://doi.org/10.1080/17483107.2020.1817163
Published online: 04 Dec 2020.
Submit your article to this journal
View related articles
View Crossmark data

CASE REPORT
Basic principles for the development of an AI-based tool for assistive technology
decision making
Moran Ran
a
, David Banes
b
and Marcia J. Scherer
c
a
Atvisor.ai, Ramat Hasharon, Israel;
b
David Banes Access and Inclusion Services, United Kingdom;
c
Institute for Matching Person & Technology,
Inc, Webster, NY, USA
ABSTRACT
Introduction: The impact of assistive technology use on the lives of people with disabilities has long
been demonstrated in the literature. Despite the need for assistive technologies, and a wealth of innova-
tive, afford-able, and accessible products, a low rate of assistive technology uptake is globally maintained.
One of the reasons for this gap is related to data and knowledge formation and management. Low
access to information and a lack of assessment services is evident. Fragmentation of data, inconsistency
in assessment methodology and heterogeneity in the competence of assistive technology professionals,
has led to a growing interest in the opportunities that data sciences, including AI, hold for the future of
the assistive technology sector, as a supportive and constructive mechanism in any decision-mak-
ing process.
Objectives: In this short paper, we seek to describe some of the principles that such an AI-based recom-
mendation system should be built upon, using the Atvisor platform as a case study. Atvisor.ai is an AI-
based digital platform that supports assistive technology assessments and the decision-making process.
Recommendations: Our recommendations represent the aggregated insights from two pilots held in
Israel, testing the platform in multiple environments and with different stakeholders. These recommenda-
tions include ensuring the continuum of care and providing a full user journey, incorporating shared deci-
sion making and self-assessment features, providing data personalisation and a holistic approach,
building a market network infrastructure and designing the tool within a wider service delivery model
design. Assessment and decision-making processes, crucial to optimal uptake, cab be leveraged by tech-
nology to become more accessible and personalised.
äIMPLICATIONS FOR REHABILITATION
Provides principles for the development of an AI-based recommendation system for assistive technol-
ogy decision making.
Promotes the use of artificial intelligence to support users and professionals in the assistive technol-
ogy decision making process.
Personalization of data regarding assistive technology, according to functional, holistic and client cen-
tered profiles of users, ensures optimal match and better use of assistive technology.
Self-assessment and professional assessment components are important for enabling multiple access
points to the assistive technology decision making process, based on the preferences and needs
of users.
ARTICLE HISTORY
Received 31 May 2020
Accepted 27 August 2020
KEYWORDS
Assistive technology;
artificial intelligence;
assistive technology
decision making;
recommendation system;
market network
The potential outcomes of assistive technology use and the
impact it can have on the lives of people with disabilities have
long been discussed and demonstrated in the literature (for
recent examples see Refs. [1–6]). The COVID-19 pandemic and the
introduction of social distancing and quarantine, led to at-risk
populations, including those with disabilities, to be tied to their
homes and has brought an added urgency to the planning and
provision of assistive technologies. Direct and personal services
must maintain a safe and independent distance, despite a grow-
ing need to find and distribute assistive technologies that func-
tion as effective support.
In a 2019 global consultation, Chapal Khasnabis, the head of
the Global Cooperation for Assistive Technology (GATE) of the
WHO, stated that assistive technology is currently characterised
by “a need, but there is no demand”[7]. Indeed, despite an
evident need for assistive technologies from the user’s perspec-
tive, and a wealth of innovative, affordable and accessible prod-
ucts in an open market of possibilities and mainstream solutions,
this has not translated into common rules of provision, supply
and demand [8]. As a result, the low rate (5–15%) of assistive
technology uptake is globally maintained [9].
One of the reasons for this gap is related to data and know-
ledge formation and management. Low access to independent
information and a lack of assessment services for assistive tech-
nology is evident. Assistive technology, unlike many commodities
or products, requires judgement in identifying appropriate solu-
tions and further customisation and support in training and
usage. This is dependent on the goals and needs of the person
seeking a solution and the complexity of the solution itself.
Comprehensive work in the field of matching person to
CONTACT Moran Ran moran@atvisor.ai Atvisor.ai, Ramat Hasharon, Israel
ß2020 Informa UK Limited, trading as Taylor & Francis Group
DISABILITY AND REHABILITATION: ASSISTIVE TECHNOLOGY
https://doi.org/10.1080/17483107.2020.1817163

technology [10–12] has shown that many elements must be con-
sidered in choosing an assistive technology solution. Assistive
technology professionals, often responsible for facilitating such a
decision-making process together with their clients, may also
struggle with the complexities described (for example Arthanat
et al. [13]).
Fragmentation of knowledge and information, inconsistency in
assessment methodology and heterogeneity in the competence
of assistive technology professionals, has led to a growing interest
in the opportunities that data sciences, including AI, hold for the
future of the assistive technology sector, as a supportive and con-
structive mechanism in any decision-making process [14].
In this short paper, we seek to specify and describe some of
the principles that such an AI-based recommendation system
should be built upon, using the Atvisor platform as a case study
and an example.
Atvisor.ai is an AI-based digital platform that supports assist-
ive technology assessments and the decision-making process.
Used by assistive technology professionals and users, it serves as
a CDSS (clinical decision support system), to help in structuring
the decision-making process, providing relevant ideas for assistive
technology solutions and enables purchasing of the selected
product online. Usage begins with a simple search in the assistive
technology catalogue and may proceed to a more personalised
profile-based search (Figure 1).
The end-user, supported by the assessor, defines his/her
unique profile in keeping with the ICF as a bio-psycho-social
analysis. Following the creation of a personal profile, the user
can then choose the personal goals that are central to him/her
and for which activities assistive technology is required. For
each goal, the platform creates a personal report, containing
recommendations across different assistive technology catego-
ries and subcategories, that are relevant to the profile and the
activity mentioned.
The end-user can review the report, together with the assessor
and independently, consult with family and friends and select the
product he/she would like to buy or have provided. Online
purchasing can be integrated within the platform, alongside infor-
mation about local shops, suppliers and lending programs.
Relevant products can be shared, reviewed or marked as favour-
ites for continuous use, while information from searches, recom-
mendations, purchases and online correspondence is stored both
in the personal archive of the end-user and that of the assessor.
This enables follow up and long-term support, as well as re-initi-
ation of further personalised assistive technology searches for any
additional goals.
On the technological level, Atvisor had developed a new way
of analysing assistive technology products, based on client needs.
The possibility to manage an online matching between the user
and the technology, is enabled through a personal matching
component that contains
an infrastructure of rules sets emulating the thinking process
of a senior professional in the assistive technology field, sup-
porting it and expanding it
a matching algorithm that “connect”between the profile of
the person and the right assistive technology solutions
a functional assessment tool enabling the definition of a per-
sonal profile
The following recommendations represent the aggregated
insights from two pilots held in Israel, testing the platform in mul-
tiple environments and with different stakeholders
1
:
1. Ensure the continuum of care and provide a full user journey
In the assistive technology process, there are many “entry
and exit”points, where the assistive technology user can
deal with questions and concerns, which if not addressed,
may lead to product abandonment. Following the iterative
and repeated cycles of product testing and user feedback, by
assistive technology professionals and users engaged in, the
following stages have been identified:
Discovery (general exploration of possibilities, some-
times with no specific goals declared, and to “open up”
the possibility of using assistive technology),
Saas - digital
assesment plaorm
Marketplace – referral
based
AT Catalogue – local
and internaonal
Selecon
Discovery
Provision and delivery
People with disabilies
Professionals
People with disabilies
Professionals
People with disabilies
Professionals
AT Suppliers
Atvisor – AI based market network for the AT industry
Figure 1. A suggested Market Network model for the Atvisor platform including three platform components –database (catalogue), digital assessment platform
(SaaS) and a marketplace, as well as presenting three stakeholders communities using the platform –assistive technology users, professionals and suppliers.
2 M. RAN ET AL.

Identification (focusing or “zooming in”on assistive
technology categories and subcategories that might be
of relevance to the current search),
Selection (interest that is shown regarding specific or
types of assistive technology products. This may entail
loaning and trial possibilities before reaching the
final decision),
Provision of products from an assistive technol-
ogy seller,
Delivery of the purchased product and beginning
of use.
With the understanding of the value that systems and processes
give to users if support is present throughout [15], three distinct
yet interconnected modules in the Atvisor platform have been
developed –the discovery module, the assessment module and the
provision and follow up support module. By moving forward from
module to module across the platform, the assistive technology
user and professionals can be digitally supported throughout the
identification of needs, the selection of products, the provision of
the desired solution and its optimal use.
As the platform offers many potential solutions to each goal or
challenge presented, the professional can maintain involvement
until completion, without creating an unintentionally biased or
limited recommendation for a specific product.
2. Incorporation of shared decision making and self-assessment
features
Current effective practice assumes that the assistive technol-
ogy user will be involved in the decision-making process and
that a user-centred service, will eventually lead to better
technology uptake by the person [16,17]. Nevertheless, as
learned from pilot results, assistive technology professionals
and users might conceive shared decision-making processes
quite differently. This discrepancy is significant and influen-
tial. The understanding of its mechanisms is recommended
as a matter for further and future research. It appears that
the use of a digital tool that accompanies stakeholders as a
step by step experience can be an effective way to construe
and unify such variance. Based on that understanding, the
next step is a move towards self-assessment, for situations
and products where professional involvement is less critical.
In Atvisor the Self-assessment feature provides a better
opportunity for self-determination and expression of needs
and goals by assistive technology users. It demonstrates in
practice, the move towards participative and affirmative mod-
els of disability while easing pressure on professionals’case-
load and releasing resources for cases where their
involvement most essential.
3. Personalisation
An optimal match between the needs and characteristics of
those with a disability and the most appropriate assistive
technology requires a depth of professional knowledge and a
comprehensive understanding of existing solutions in the
market. Studies show that high levels of product abandon-
ment are due to several factors related to the quality of solu-
tion matching [18–20]. Providing personalised
recommendations that accurately support the counselling
process must be a procedure that takes into consideration
the complex and rich, unique profile of every user, personal-
ity, abilities and needs, preferences, environment, tasks and
goals. By using an AI-based matching algorithm, that calcu-
lates the matrix of every personal profile, the platform identi-
fies and recommends the most suitable assistive technology
solutions for the persons’goals and needs. Thus, from the
wide range of tens of thousands of products available on the
database, only the specific and most relevant products will
appear in any search. The range of products that appear in
the recommendations reflects the platform’s fundamental
principles to open the market and to ensure comparisons of
products from an empowered consumers’position.
4. Holistic approach
When a person with disabilities or an older person seeks to
drink, eat or communicate independently, and has a personal
profile that expresses multiple challenges, capabilities and
preferences, specific recommendations for cups and mugs,
plates or mobile phones, will not be enough, even if they are
a good fit themselves. Many potential challenges might be
encountered, and so the types of assistive technology solu-
tions that should be offered must represent a wide variety of
commercial categories and product segments. An AI-based
matching algorithm can provide this due to the computing
capacity and ability to perform complex needs integration
and interaction. Furthermore, this interaction between limita-
tions and needs, can result in additional recommendations of
assistive technologies and serve as an even more advanced
part of the discovery, education and awareness of both
assistive technology professionals and users.
5. The design of a digital tool should be a part of a wider ser-
vice delivery model design
During product development Atvisor has encountered many
questions regarding the service delivery model, from assess-
ment to provision, that unless addressed, would serve as a
barrier to the adoption of the platform. Creating high-level
collaborations with governmental offices relevant to policy
and regulation, while working with stakeholders on the
ground to learn about their needs and the unique value
propositions that are relevant to each has established the
basis for the development of a unique service combining
human resources and experience, together with digital tools.
This development can eventually support the need for stand-
ardisation and scale in services, in line with global policy rec-
ommendations and systems thinking approach implemented
in the field of assistive technology today [21,22].
6. The importance of building an online network infrastructure
The Atvisor platform offers an ongoing and dynamic inter-
action between assistive technology users, professionals and
suppliers. For example, a user can reach out to their nomi-
nated professional online with questions or requests, and if
necessary, can seek a professional from within the platform.
The professional can share recommendations with the user
and discuss them online, and the supplier can be approached
for a purchase or a request regarding services. This network
of interactions is highly relevant to the availability of assess-
ment services in a period of increasing demand, to the dis-
covery of assistive technology solutions and to the evolution
of a new, much-needed awareness and proactive mindset
that translates itself eventually to demand.
Even before the global pandemic, we recognised that the scale
of unmet need for assistive technologies demanded innovation
and change. Demographics and the aspirations of people with a
disability, within a rights-based approach, meant that the trad-
itional model of delivery could not fulfil demand. The impact of
the pandemic has brought into sharp relief the issue that limited
professional resources, including time, has delayed access to prod-
ucts and services leading to frustration and discrimination for
people with a disability. The approach outlined and implemented
BASIC PRINCIPLES FOR THE DEVELOPMENT OF AN AI-BASED TOOL FOR ASSISTIVE TECHNOLOGY DECISION MAKING 3

addresses those demands and offers an alternative approach
based upon telerehabilitation with artificial intelligence.
Potential access to assistive technologies for the benefit of
people with disabilities and older people could not look more
promising, the diversity of products from social robots to smart
homes to autonomous vehicles offers a world of options at our
fingertips. These can help address many barriers. Our task now
would be to ensure that assessment and decision-making equally
respond to the possibilities, and that those who serve as gate-
keepers to knowledge will have the proper “key”to make it
accessible and personalised.
Today, where remote services become the principal channel
for communication, this has never been more urgent.
Note
1. Pilots results and insight will be shared in upcoming articles.
Disclosure statement
No potential conflict of interest was reported by the author(s).
References
[1] Jamieson M, Cullen B, McGee-Lennon M, et al. The efficacy
of cognitive prosthetic technology for people with memory
impairments: a systematic review and meta-analysis.
Neuropsychol Rehabil. 2014;24(3–4):419–444.
[2] Perelmutter B, McGregor KK, Gordon KR. Assistive technol-
ogy interventions for adolescents and adults with learning
disabilities: an evidence-based systematic review and meta-
analysis. Comput Educ. 2017;114:139–163..
[3] Andreopoulou G, Mercer TH, van der Linden ML. Walking
measures to evaluate assistive technology for foot drop in
multiple sclerosis: a systematic review of psychometric
properties. Gait Posture. 2018;61:55–66.
[4] Brims L, Oliver K. Effectiveness of assistive technology in
improving the safety of people with dementia: a systematic
review and meta-analysis. Aging Mental Health. 2019;23(8):
942–951.
[5] McNicholl A, Casey H, Desmond D, et al. The impact of
assistive technology use for students with disabilities in
higher education: a systematic review. Disabil Rehabil
Assist Technol. 2019;1–14. DOI:10.1080/17483107.2019.
1642395
[6] Perfect E, Jaiswal A, Davies TC. Systematic review: investi-
gating the effectiveness of assistive technology to enable
internet access for individuals with deafblindness. Assist
Technol. 2019;31(5):276–285.
[7] Khasnabis C. 2019. Closing remarks, GReAT consultation,
Geneva, 22 and 23 August 2019.
[8] de Witte L, Steel E, Gupta S, et al. Assistive technology pro-
vision: towards an international framework for assuring
availability and accessibility of affordable high-quality
assistive technology. Disabil Rehabil Assist Technol. 2018;
13(5):467–472.
[9] World Health Organization. World report on disability.
Geneva: World Health Organization; 2011.
[10] Scherer MJ, Sax C, Vanbiervliet A, et al. Predictors of assist-
ive technology use: the importance of personal and psy-
chosocial factors. Disabil Rehabil. 2005;27(21):1321–1331.
[11] Scherer M, Jutai J, Fuhrer M, et al. A framework for model-
ling the selection of assistive technology devices (ATDs).
Disabil Rehabil Assist Technol. 2007;2(1):1–8.
[12] Scherer M. 2017. Matching person and technology. In:
Maheu M, Drude K, Wright S, editors. Career paths in tele-
mental health. Cham: Springer.
[13] Arthanat S, Elsaesser L-J, Bauer S. A survey of assistive tech-
nology service providers in the USA. Disabil Rehabil Assist
Technol. 2017;12(8):789–800.
[14] Banes D. A commentary on influences on selection of
assistive technology for young children in South Africa:
perspectives from rehabilitation professionals. Disabil
Rehabil Assist Technol. 2019;14(8):758–759.
[15] Banes, D. Pervasive and emerging technologies and con-
sumer motivation. In Everyday technologies in healthcare.
Boca Raton (FL): CRC Press, Taylor and Francis Group; 2019.
p. 123–142.
[16] Scherer MJ. The change in emphasis from people to per-
son: introduction to the special issue on Assistive
Technology. Disabil Rehabil. 2002;24(1–3):1–4.
[17] Federici S, Corradi F, Meloni F, et al. A person-centered
assistive technology service delivery model: a framework
for device selection and assignment. Life Span Disabil.
2014;17:175–198.
[18] Federici S, Borsci S. Providing assistive technology in
Italy: the perceived delivery process quality as affecting
abandonment. Disabil Rehabil Assist Technol. 2016;11(1):
22–31.
[19] Cruz D, Emmel M, Manzini M, et al. Assistive technology
accessibility and abandonment: challenges for occupational
therapists. Open J Occup Ther. 2016;4(1):1–10.
[20] Petrie H, Carmien S, Lewis A. 2018. Assistive technology
abandonment: research realities and potentials. In:
Miesenberger K, Kouroupetroglou G, editors. Computers
helping people with special needs. ICCHP 2018. Lecture
notes in computer science, vol 10897. Cham: Springer.
[21] MacLachlan M, Scherer MJ. Systems thinking for assistive
technology: a commentary on the GREAT summit. Disabil
Rehabil Assist Technol. 2018;13(5):492–496.
[22] MacLachlan M, Banes D, Bell D, et al. Assistive technology
policy: a position paper from the first global research,
innovation, and education on assistive technology (GREAT)
summit. Disabil Rehabil Assist Technol. 2018;13(5):454–466.
4 M. RAN ET AL.