=Paper=
{{Paper
|id=Vol-3823/3_mauro_personalized_6
|storemode=property
|title=Personalized Music Recommendation for People with Autism Spectrum Disorder
|pdfUrl=https://ceur-ws.org/Vol-3823/3_mauro_personalized_6.pdf
|volume=Vol-3823
|authors=Liliana Ardissono,Federica Cena,Noemi Mauro
|dblpUrl=https://dblp.org/rec/conf/healthrecsys/ArdissonoCM24
}}
==Personalized Music Recommendation for People with Autism Spectrum Disorder==
https://ceur-ws.org/Vol-3823/3_mauro_personalized_6.pdf
Personalized Music Recommendation for People with
Autism Spectrum Disorder
Liliana Ardissono1 , Federica Cena1 and Noemi Mauro1
1
Computer Science Department, University of Torino, Corso Svizzera 185, Torino, I-10149, Italy
Abstract
The project "ACCESS: Accessibility to Clinical Care for People with ASD through Anxiety Management Using
Personalized Applications and IoT" aims to develop innovative technologies to manage anxiety in individuals
with Autism Spectrum Disorder (ASD) when they undergo dental and otolaryngology treatments. The premises
of this project are the potential of multimedia content to help people with ASD relax when exposed to stressful
conditions in healthcare. The result will be an app that personalizes the selection of music tracks and videos to
be played before and during the treatment to help the patient distract and cope with stressful conditions that
might cause anxiety. For this purpose, the app will receive information about the patient’s arousal state collected
by physical sensors and/or the clinical staff. The app will use this data to personalize the selection of multimedia
content based on the patient’s preferences, arousal level, and level of noise around her or him.
Keywords
Personalized healthcare services, recommender systems, autism
1. Introduction
In this paper, we present the project ACCESS, "Accessibility to Clinical Care for People with ASD
through Anxiety Management Using Personalized Applications and IoT". The project aims to develop
ICT technologies that support the management of anxiety in individuals with Autism Spectrum Disorder
(ASD) before and during clinical treatments, with specific attention to dental care treatments and
otolaryngological examinations. Given the peculiarity of ASD, which induces different sensitivities to
external stimuli, these treatments challenge patients through stimuli concerning hearing and touch, a
social context that includes unknown people, and a possibly high level of surrounding noise. Patients
thus need help in coping with anxiety.
ICT offers a powerful means to assist individuals with autism in everyday life, including healthcare
support [1, 2]. However, for these tools to be effective, efficient, and satisfactory, they should be (i)
accessible and user-friendly for individuals with autism, and (ii) seamlessly integrated into clinical
protocols, which need to be adapted to incorporate and leverage this technology. Recent studies [3, 4]
show the potential to personalize healthcare for people with autism. However, current ICT-based
solutions to healthcare support adopt a one-size-fits-all approach to all patients.
The ACCESS project investigates the benefits of integrating digital personalization techniques to
help patients with ASD undergo treatments adapting to individual needs, sensitivities, and the noise
in the surrounding context. The project builds on findings showing that people’s preferred music is
effective in anxiety treatment [5] and custom multimedia content administration has positive effects in
dental care [3]. However, in these works the selection of the content to be delivered is not automated.
Differently, ACCESS aims to develop an application that plays music and multimedia content to calm
patients with ASD by adapting content in real-time based on their arousal level and the surrounding
noise, and exploiting their preferences for content selection.
Our work in this project concerns the development of the multimedia content selection module. For
this purpose, we will develop a novel interaction model to adapt the elicitation of content preferences
to ASD people with different functioning levels. Moreover, we will develop a recommender system that
HealthRecSys’24: The 6th Workshop on Health Recommender Systems co-located with ACM RecSys 2024
$ liliana.ardissono@unito.it (L. Ardissono); federica.cena@unito.it (F. Cena); noemi.mauro@unito.it (N. Mauro)
0000-0002-1339-4243 (L. Ardissono); 0000-0003-3481-3360 (F. Cena); 0000-0001-8234-3266 (N. Mauro)
© 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
combines context awareness (to consider the patient’s preferences, arousal state, and the level of noise
in the environment) with constraints imposed by the type of visit to be carried out and the patient’s
sensitivities.
2. Related Work
Music is considered a powerful relaxation tool. Both the acoustic features of the music and individual
user preferences play crucial roles in relaxation [6]. In the clinical domain, Lai-Tan et al. report that
the best results in applying music therapy to face depression and anxiety are obtained when people
listen to their preferred songs because this induces enjoyment [5]. For instance, some people find
heavy metal music (that typically has 100-150 Beats Per Minute (BPM)) relaxing while, according to [7],
sedative music used in music therapy is characterized by a slow tempo of 60-80 BPM. Thus, knowing
the contents that the patient likes, and her/his musical preferences, is key to selecting promising tracks
for relaxation.
Personalized technologies for autism are scarce in healthcare contexts [8]. Few works provide
personalized support to people with autism (as well as their caregivers and physicians) in managing
medical care [9]. Nicolaidis et al. [10] developed a tool that allows patients to create a personalized
report for their healthcare provider, improving communication with healthcare operators. Personalized
interventions proved to be very effective, especially in the management of anxiety in people with autism
[8, 11].
Bondioli et al. [3] developed personalized digital tools such as photos, videos, and interactive PDFs to
familiarize ASD children with dental procedures and environments and teach them to perform proper
oral hygiene at home. The results confirmed the potential to personalize the selection of ICT tools
to reduce anxiety in professional settings. However, in that work, the creation and selection of the
multimedia material was carried out by the medical staff in collaboration with the patient and her/his
caregivers, without any automated personalization support.
The ACCESS project differs from the previously cited works because we plan to develop an app
that dynamically selects and plays multimedia content based on the patient’s preferences, sensitivities,
and arousal level, and considering the specific treatment to undergo and the presence of noise in the
surrounding environment.
Notice that, concerning music selection, our work differs from musical biofeedback, where sounds
are exploited as interaction means to convey positive or negative feedback to patients while they carry
out tasks in contexts such as relaxation assistance [12] and stroke rehabilitation [13]. In our case, the
patient is expected to listen to the music passively, and we aim to select music tracks that can make
her/him relax to cope with a clinical treatment.
3. The Project
ACCESS focuses on dental care and otolaryngologist treatments. It is a 2-years project that involves the
University of Pisa (development of the main ACCESS tools), the Polytechnic of Milan (development of
sensors to monitor patients’ anxiety and stress levels), the University of Torino (development of the
multimedia content selection system), and the National Research Council (participatory methodology
for the project’s design, development, and assessment).
As the literature suggests the benefits of music and videos in helping people with ASD cope with
medical treatments, the project will target both media. However, we will first focus on delivering music.
For this purpose, we are developing a web app to acquire the patient’s musical preferences and support
her/him during the medical treatments. The app is used in three separate phases:
1. The patient will be instructed to use the app at home to explore and play the preferred multimedia
content. The app will adapt the level of guidance in preference elicitation and content exploration
to the patient’s autonomy, considering different ASD functioning levels. Given the patient’s
interaction with the app, the system will collect her/his music preferences regarding genres,
authors, and individual preferred tracks. Moreover, it will analyze the acoustic features of the
preferred tracks through the Spotify API.1 As people with ASD have individual tolerance levels
to acoustic features, this analysis is key to building an individual user profile that specifies the
ranges of values compatible with the individual user and can be used to select the tracks to play
before and during medical treatments.
2. Before the medical treatment, e.g., in the waiting room, the app will help the patient listen to
her/his favorite multimedia content or guide her/him in the exploration of new content. The goal
is to keep the user relaxed by suggesting content that reflects the preferences collected in the
previous phase and having features that do not agitate her/him.
The app will sense the surrounding environment to estimate the noise level. Based on this
information and the user profile, it will select the content to be administered (e.g., whether playing
music or showing mute videos to avoid adding further disturbing factors for the patient). For music
selection, we plan to steer the invocation of the Spotify API concerning the recommendations2
to retrieve music tracks compatible with the patient’s musical preferences and sensitivity, and
with the surrounding environment. We will do this by feeding the API with data about authors,
genres, and the minimum and maximum values of the acoustic features stored in the patient’s
user profile.
3. During medical treatment, the app will receive information about the patient’s arousal state (e.g.,
heart and respiration rates collected by sensors, or information provided by the medical staff).
Moreover, it will sense the surrounding environment to estimate the noise level and play the
appropriate multimedia content based on this data and the user profile. In this case, the type of
content selected by the app could change to reflect the patient’s arousal state.
The app will be tested in collaboration with the Audiology and Phoniatrics Clinics (OPC) at AOUP
Pisa to measure the user experience during the interaction with it, and its effectiveness in supporting
patients during medical treatments. So far, we participated in a focus group with 2 caregivers, 1 psychol-
ogist, 1 speech therapist, 1 otolaryngologist, 6 researchers and 2 technologists to gather stakeholders’
requirements and needs. The focus group confirmed that the use of technology can help the user to
prepare before the medical treatment and could also be useful while the treatment is performed. In
addition, a personalized approach could improve the user experience if a strong weight is given to user
preferences.
Acknowledgments
This study is funded by the NATIONAL RECOVERY AND RESILIENCE PLAN (NRRP) – MISSION 4
COMPONENT 2 INVESTMENT 1.1 – “Fund for the National Research Program and for Projects of
National Interest (NRP)” Italian Ministry of Education call PRIN 2022 PNRR D.D. n. 1409 14/09/2022
- Title of the project “ACCESS: Accessibility to clinical care for people with ASD through anxiety
management by using personalized applications and IoT”, project number P2022PBTSK. We thank Luca
Bonamico for his contribution to the project.
References
[1] A. J. Doherty, H. Atherton, P. Boland, R. Hastings, L. Hives, K. Hood, L. James-Jenkinson, R. Leavey,
E. Randell, J. Reed, L. Taggart, N. Wilson, U. Chauhan, Barriers and facilitators to primary health
care for people with intellectual disabilities and/or autism: an integrative review, BJGP Open 4
(2020). doi:10.3399/bjgpopen20X101030.
1
https://developer.spotify.com/documentation/web-api/reference/get-audio-features. This API allows data to be recovered on
the acousticness, danceability, energy, instrumentality, liveness, loudness, speechiness, and tempo of each track.
2
https://developer.spotify.com/documentation/web-api/reference/get-recommendations
[2] E. McGhee Hassrick, L. Graham Holmes, C. Sosnowy, J. Walton, K. Carley, Benefits and risks: A
systematic review of information and communication technology use by autistic people, Autism
in Adulthood 3 (2021). doi:10.1089/aut.2020.0048.
[3] M. Bondioli, S. Pelagatti, M. C. Buzzi, M. Buzzi, C. Senette, Ict to aid dental care of children with
autism, in: Proceedings of the 19th International ACM SIGACCESS Conference on Computers
and Accessibility, ASSETS ’17, Association for Computing Machinery, New York, NY, USA, 2017,
p. 321–322. URL: https://doi.org/10.1145/3132525.3134799. doi:10.1145/3132525.3134799.
[4] S. Petty, M.-L. Bergenheim, G. Mahoney, L. Chamberlain, Adapting services for autism: Recom-
mendations from a specialist multidisciplinary perspective using freelisting, Current Psychology
42 (2021) 1–12. doi:10.1007/s12144-021-02061-3.
[5] N. Lai-Tan, M. G. Philiastides, F. Kawsar, F. Deligianni, Toward personalized music-therapy: A
neurocomputational modeling perspective, IEEE Pervasive Computing 22 (2023) 27–37. doi:10.
1109/MPRV.2023.3285087.
[6] D. Mélo, Investigating the role of personalization when creating relaxing playlists, WebMedia
’20, Association for Computing Machinery, New York, NY, USA, 2020, p. 213–216. URL: https:
//doi.org/10.1145/3428658.3431755. doi:10.1145/3428658.3431755.
[7] J. Yu, B. Benatallah, F. Casati, F. Daniel, Understanding mashup development, IEEE Internet
Computing 12 (2008) 44–52. doi:10.1109/MIC.2008.114.
[8] W. Simm, M. Ferrario, A. Gradinar, M. Smith, S. Forshaw, I. Smith, J. Whittle, Anxiety and autism:
towards personalized digital health, in: C. Lampe (Ed.), Proceedings - The 34th Annual CHI
Conference on Human Factors in Computing Systems, Association for Computing Machinery
(ACM), United States of America, 2015, pp. 1270–1281. doi:10.1145/2858036.2858259.
[9] P. M., Y. C., Food and therapy recommendation system for autistic syndrome using machine
learning techniques, in: 2019 IEEE International Conference on Electrical, Computer and Commu-
nication Technologies (ICECCT), 2019, pp. 1–6. doi:10.1109/ICECCT.2019.8868979.
[10] C. Nicolaidis, D. Raymaker, K. McDonald, S. Kapp, M. Weiner, E. Ashkenazy, M. Gerrity, C. Kripke,
L. Platt, A. Baggs, The development and evaluation of an online healthcare toolkit for autistic
adults and their primary care providers, Journal of general internal medicine 31 (2016) 1180–1189.
doi:10.1007/s11606-016-3763-6.
[11] J. Rodgers, S. Brice, P. Welsh, B. Ingham, C. Wilson, G. Evans, K. Steele, E. Cropper, A. Frcpch,
M. Freeston, J. Parr, A pilot randomised control trial exploring the feasibility and acceptability of
delivering a personalised modular psychological intervention for anxiety experienced by autistic
adults: Personalised anxiety treatment-autism (pat-a), Journal of Autism and Developmental
Disorders (2023) 1–16. doi:10.1007/s10803-023-06112-5.
[12] B. Yu, M. Funk, J. Hu, L. Feijs, Unwind: a musical biofeedback for relaxation assistance, Behaviour
& Information Technology 37 (2018) 800–814. URL: https://doi.org/10.1080/01449290500331131.
doi:10.1080/0144929X.2018.1484515.
[13] P. Kantan, E. G. Spaich, S. Dahl, A technical framework for musical biofeedback in stroke
rehabilitation, IEEE Transactions on Human-Machine Systems 52 (2022) 220–231. doi:10.1109/
THMS.2021.3137013.