Accessible and Usable Websites and Mobile Applications for People with Autism Spectrum Disorders: a Comparative Study (cid:72)

Accessibility , usability and incl usion represen t desir able challeng es of curren t research in the fie d of univ ersal design: in some cases, these features require adaptiv e beha viours and specialised customisa tions, while, in gener al, it is possible to iden tify common and shareable guidelines. We focus our atten tion on chil dren with autism spectrum disorders. Many studies show the positiv e impact of using computer technol ogies for supporting the liv es of these users. Despite that, just a restricted part of the curren t websites and apps is accessible and usable for people with ASD. In this paper , we presen t gener al and shared guidelines and best practices for accessibility and usability for all; and we propose specialised guidelines for designers and dev elopers of websites and mobile applica tions for users with ASD. We then presen t a review of many of the existing websites and applica tions, in order to check which comply with all, or parts of these guidelines.


Introduction
Autism spectrum disorders (ASD) represen t a neurobiological condition tha t significa tl y impacts beha viour , social comm unica tion, social rela tionships, imagination/rigidity of thoughts [1]. Typical interv entions based on the use of Picture Exchang e Comm unica tion Systems (PEC S), aided languag e stim ula tion, visual tools and social stories fin in the Web, in mobile devices, in specialised softw are and, in gener al, in the wide spread of Inf orma tion and Comm unica tion Technol ogies (ICT) a meaningful opportunity to improv e comm unica tion, assist the dev el opmen t of social skills, enhance ability to learn of the people with ASD, and not onl y. In fact , al though these users are di fferen t one from another , they gener all y show good abilities in using computer technol ogies [2], tha t represen t for all, but particular ly for them, user -friendl y, eng aging, mul tifunctional, portable, easil y accessed, and motiv ating

The Autism Spectrum Disorder
ASD is define by the American Psychia tric Associa tion as a neuro-dev el opmen tal disorder with persisten t impairmen ts in social comm unica tion and social inter action, and restricted, repetitiv e pa tterns of beha viour , interests, or activities [3].
The incidence of this disorder is not negligible. In [4], the authors presen t an interesting study on worldwide available da ta tha t estima tes at the da te of 2010 the number of people with ASD as 1 out of 132. The study find no evidence of a chang e in prev alence for ASD betw een 1990 and 2010, al though there are some small chang es depending on regional origins. A more recen t study cond ucted in 2012 among 346,978 chil dren ag ed 8 years in 11 di fferen t cities of the United States, shows a gener al evidence of one chil d in 68 with ASD.
Characteristics of ASD. Each person with ASD is di fferen t, this is where the term "spectrum disorder" comes from. The areas which are most affected are: social inter action, social imagina tion, and social comm unica tion.
Reg arding social interaction, people with ASD typicall y tend to isola te themsel ves showing no interest in other people, do not have a good ey e contact , try to avoid ph ysical contacts, have problems processing their own emotions and the ones of people around them.
Their social imagination is limited: they tend to avoid symbolic games, tend to repea t the same game or ev en mov emen ts over and over (hand flapping spinning or waving objects, etc.), get frustr ated when something chang es in their dail y routine. Finall y, they often show impairmen ts in social communication. These impairmen ts are often rela ted to languag e dela ys or, in some cases, to the complete lack of verbal comm unica tion. The prev alence rate of these languag e impairmen ts is unclear , how ev er the authors of [5] estima ted tha t 25% − 50% of people with ASD nev er dev el op ev en basic comm unica tion skills, which are fundamen tal for any dail y lif e functionality . People with ASD have also problems understanding instructions, gestures and so on.
Finall y, they often show limited attention, i.e., they are able to concen trate on tasks for a limited amoun t of time, and have also Sensory Processing Disorders (SPD), i.e., have problems processing inf orma tion from the fi e senses, from the vestibular system, and/or the positional sense [6].
Possible interventions. People with ASD are very di fferen t one from another , thus there is no unique ther ap y tha t works well for all of them. Wha t is widel y shared among all di fferen t comm unities, is tha t ear ly interv ention is crucial, the ear liest you discov er this disorder , the ear liest the famil y can access specialized interv entions, and can understand their kids beha viour and needs. How ev er, the diagnosis is non-trivial, giv en tha t ASD is a neurol ogical disorder and not a ph ysical one, i.e., there is no medical proced ure to iden tify it (except obviousl y for limited known genetic disorders). Thus, the diagnosis is onl y based on a study of the presence/absence of some specifi beha viours (see, e.g., [7,8]).
Modern ther apies propose very di fferen t approaches (which are out of the scope of this paper), how ev er, it is widel y known tha t people with ASD usuall y presen t good visual abilities, such as visual memory , i.e., are able to represen t concepts by sequence of imag es [9]. Thus, to support these individ uals man y of the proposed ther apies rel y on the use of photogr aphs, imag es, f ow charts, cartoons, checklists, etc. Wha t we will be concen trating on, in this paper , is the use of technol ogy to support all the ther apies, and in particular , the use of imag es as a very pow erful comm unica tion tool. In particular , to support comm unica tion interv entions, often speech ther apies are also sustained by Augmen tativ e and Alterna tiv e Comm unica tion (AAC) techniques, which are based on the use of symbols or imag es as a method for comm unica ting [10 ]. The most common AAC approach is the PEC S: users comm unica te needs and requests by exchanging pictures with their partners; these pictures are lamina ted and stored in a special book tha t has to be carried around [11 ]. Another AAC technique is, e.g., the sign languag e, which can be very effectiv e, but how ev er requires the partners to be trained, and thus restricts the comm unica tion to a limited set of individ uals. An ev olution of AAC techniques relies on the use of di fferen t computer devices such as tablets, smartphones, etc.. These new tools all ow to increase number of stored imag es, have limited ph ysical size, and can thus be carried out ev erywhere. Recen t studies show the effectiv eness of using AAC tools for functional requests by a set us ASD users under the ag e of 16 [10 ].

Accessibility and Usability Guidelines for Websites and Applications
Man y studies show the effectiv e positiv e impact of using computer technol ogies to support the liv es of users with ASD, in order to sim plify inter action with other people, to org anise dail y activities, to improv e rela tions with famil y and friends [12 , 13 ].
Moreov er, users with ASD presen t a positiv e attitude tow ards computer technol ogies due to the 2 EAI Endorsed Transactions on Ambient Systems 12 2016 -05 2017 | Volume 4 | Issue 13 | e5 predictability of the inter action -in contrast to normal day-to-da y inter action with other people -and due to the perf ectibility of the tool, tha t may ind uce repetitiv e beha viours, usuall y pref erred by this set of users. Thus, ICTs are pow erful tools for improving their learning process [2].
The set of available computer technol ogies is very wide, since it rang es from virtual reality , to robotics, mul titouch interf aces, websites, Web apps, affectiv e computing. These tools are often customisable with respect to the di fferen t users' abilities, and thus targ eted to the di fferen t skills.
While dev el oping a website, or a mobile applica tion, it is desir able to incl ude an accessibility and usability specialist in the team so to design and then ev al ua te the resul ts of the design in a proper way [14 ]. We here focus onl y on websites and (mobile) applica tions and discuss possible accessibility and usability guidelines.

Defining Web accessibility and usability
The goals, approaches, and guidelines of Web accessibility and usability overlap significa tl y [15 ]: • Accessibility addresses discrimina tory aspects rela ted to equiv alen t user experience on the Web for people with disabilities.
• Usability is about effectiv e, efficien t, and satisfying design of websites and mobile applications [16 ].
Accessibility does not impl y usability , i.e., a website might be accessible, but not usable. Usable accessibility combines tog ether usability and accessibility and prod uces incl usiv e design (also called univ ersal design, or design for all), which involves designing prod ucts, such as websites, to be usable by ev eryone to the grea test exten t possible, without the need for adapta tion.
Accessibility. Users tha t have ph ysical (visual, auditory, etc.), cognitiv e or neurol ogical disabilities, children or el der ly people are the targ et users of accessibility features.
Foll owing the definitio of [15 ] we can sta te tha t: • Accessibility makes users with a wide rang e of abilities able to perceive, operate, interact and understand a user interf ace.
Accessibility has alw ays been a big concern for websites and app dev el opers; how ev er, it is often neglected during the dev el opmen t phase mainl y due to the lack of knowledg e by the dev el opers, and also to the extr a costs it introd uces. This is in contrast with the sta temen t declared during the 2016 United Nations Convention on the Rights of Persons with Disabilities (CRPD), tha t access to inf orma tion technol ogies has to be considered a basic human right [17 ].
Moving tow ards this direction, in 2012 the W3C [18 , 19 ] has crea ted a new task force group, called Cognitiv e and Learning Disabilities Accessibility Task Force (COGA), whose aim is to propose accessibility guidelines for Web accessibility for people with cognitiv e or neuronal disabilities [20 ]. The COGA group, tog ether with the Protocols and Forma ts Working Group, and the Web Conten t Accessibility Guidelines Working Group, has published in 2016 some interesting gener al guidelines for the dev el opmen t of websites for users with Cognitiv e and Learning Disabilities [21 ], how ev er these guidelines are too generic, and not directl y targ eted to users with ASD, which typicall y show other specifi problems such as limited atten tion, sensory hypersensitiv eness, limited text comprehension, etc.. They can be a good starting poin t, but more specifi targ eted issues shoul d be taken into accoun t, tog ether with further guidelines tha t appl y to more gener al computer applica tions.
In the last years, some work has been dev oted to the definitio of guidelines for people with cognitiv e disabilities [12 , 13 , 21 -25 ]. How ev er, while dealing with users with ASD with cognitiv e disabilities, more specifi features have to be taken into accoun t. WCAG 1.0 and WCAG 2.0 [18 ], which represen t the interna tional standard ref erence model, iden tify three main features: • the websites shoul d be clear and sim ple to use; • the websites shoul d prev ent the users from making mistakes, and, if tha t happens, make sim ple to correct them; • the websites shoul d provide tools for helping the orien tation and restore the context , if atten tion is lost.
In order to satisfy the previous items, they iden tify , in a very recen t documen t [26 ] a larg e set of gener al techniques, grouped in the foll owing 9 main categ ories: 8. Support adaptability and personalisa tion; 9. Minimise the cognitiv e skills required to use the conten t and avoid barriers.
Usability. The definitio of usability in the ISO 9241 standard is: • The exten t to which a prod uct can be used by specifie users to achiev e specifie goals with effectiv eness, efficiency , and satisf action in a specifie context of use.
Usability is a quality attribute, and as sta ted in [29 , 30 ] is rela ted to learnability, i.e., how easy is to navig ate for the firs time in the site; efficiency, i.e., once a user has learned the site, how easy is to accom plish tasks; satisfaction, i.e., the user is happ y about the website design; memorability, the user can remember the site after a period of time; errors, how man y errors are accom plished while navig ating in the site.
Usability is often used as synon ym of ease to use; usability means thinking about how and wh y people use a prod uct; it relies on user -feedback through ev al ua tion; it means a user -cen tred design, where the users' goals, men tal models, tasks and requiremen ts are all met.
In a recen t documen t [31 ], the U.S. Departmen t of Heal th & Human Services define some usability research-based guidelines. It provides a detailed list of 208 best pr actices, grouped in 18 main categ ories (most of which are self -explana tory): For each best pr actice, the documen t iden tifies on a 5-poin t Likert scale, tw o par ameters: the rela tiv e importance and the strength of evidence. As we can notice, while accessibility is rela ted to disabilities, usability is a key issue for all users, and it incl udes accessibility (item 3 in the previous list).
How ev er, during the last years it has been a gener al agreemen t on appl ying a specifi user interf ace (UI) design on the dev el opmen t of webpag es and interf aces. The UI design define the user experience by encour aging a natur al inter action betw een the user and the system. A good user experience reflect on the apprecia tion of the tool, and human factors may impact on this, thus it is crucial in the case of users with cognitiv e disabilities [14 , 32 ].
Since 1996, Jacob Nielsen (the undiscussed guru of usability), and the Nielsen Norman Group have been compiling lists of the top 10 mistakes in Web design. The last documen t [33 ] highlights old and persisten t problems, in spite of modern design pa tterns and aesthetics chang es; the most relev ant are the foll owing: 3. Small pieces of inf orma tion is often sca ttered around the website with little or no connection betw een these pieces; 4. The users are required to use repetitiv e links selecting ag ain and ag ain wha t they want (repetitiv e links); 5. Often prices and fees are hidden or di fficul t to find 6. The use of chil d sites crea te di fficul ties to get back to paren t site; 7. Poor research resul ts is still a major weakness for man y websites; 8. The use of facets and fi ters does not ensure better usability unless it is introd uced to satisfy real needs of users; 9. The pag es are overloaded of inf orma tion; 10. If the links resemble adv ertising, then they are ignored.
It is meaningful to consider the rev olution imposed by the increasing use of mobile applica tions. In fact , in the 2014, mobile applica tions have overtaken over fixe Internet access applica tions. Google itself offers a mobile friendl y test [34 ] to verify for any website if it is responsiv e or not. This epochal chang e has a significa t impact on tw o primary aspects of websites: • Graphical layout and conten t: mobile conten t is twice as di fficul t [35 ]. Websites shoul d become responsiv e and hence adapt their visualisa tions and conten t to the devices, and also to their orien tation (landscape, portr ait).
• Touch-based inter actions: the traditional inter actions with keyw ords and mouses is substituted by touch-based input in the use of mobile devices; this is a firs critical aspect in the crea tion of welladapted user experiences [36 , 37 ].
In March 2016, Google proposed a documen t [38 ] which describes gener al principles of mobile app design: it iden tifie a set of significa t items to consider , as navig ation and expl oration, search, commerce, and conversions. Finall y, it focuses on specifi and apprela ted design decisions tha t can be critical for ensuring a good user experience. The foll owing fi e principles are considered essen tial to design an app tha t caters to users' needs: • Speak the same languag e as your users; • Provide text labels and visual keys to clarify visual inf orma tion; • Be responsiv e with visual feedback after signifi can t actions; • Let the user control the lev el of zoom; • Ask for permissions in-con text.

Accessibility and usability guidelines for people with ASD
In this section, we try to defin appropria te accessibility and usability guidelines for users with ASD. Obviousl y these guidelines also take into consider ation standard accessibility and usability rules. The specifi guidelines are based on curren t liter ature tha t we have studied, compared and analised on the topic. From wha t we have discussed in Section 3, the accessibility and usability features are rela ted to di fferen t areas: • Graphical layout which incl udes the gener al design, i.e., text , imag es, conten t and how responsiv e a website or an applica tion is to a connection and if it is resizable, a feature tha t is fundamen tal while dealing with mobile devices.
• Another area incl udes the structure and navig ation which are fundamen tal while dealing with users with cognitiv e impairmen ts (i.e., for accessibility) and also for users' satisf action (usability).
• A crucial and non-trivial aspect is how to interf ace with a user: eng ag emen t, customisa tion, and adaptivity are all very importan t features to all ow the user to focus on the topic and to be satisfied • Finall y, the languag e is a key issue while dealing with impaired users.
The accur acy of graphical layout is useful to sim plify the inter action: layout and conten t shoul d be predictable. The study of Darejeh et al. [24 ] investig ates gener al usability principles tha t can be succesfull y applied also to users with men tal disorders and autism. For such users, the gener al sug gestion is to red uce the number of features available at any time instan t, and to design interf aces where tools can be found immedia tel y, without any investig ation.
A critical issue is the choice among imag es, photogr aphs, and symbolic pictures. Wha t kind of pictures shoul d be incl uded close to the text? A recen t 5 EAI Endorsed Transactions on Ambient Systems 12 2016 -05 2017 | Volume 4 | Issue 13 | e5 Table 1. Accessibility and usability guidelines for users with ASD

G1
The gener al design and the structure shoul d be sim ple, clear and predictable, secondary conten t tha t distr acts the user shoul d be avoided. The number of features available at any time instan t shoul d be limited. G2 The conten t shoul d be predictable and shoul d provide feedbacks. G3 Pictures shoul d be copiousl y used tog ether with red undan t represen tation of inf orma tion. G4 Pictures can be drawings, photogr aphs, symbolic imag es, shoul d be easy to understand, shoul d not go in the background, shoul d be in a sharp focus. G5 Background sounds, moving text , blinking imag es and horizon tal scrolling shoul d be avoided.

G6
The text shoul d go with pictures. It shoul d be clear , sim ple, and short (at most one sen tence on a line); shoul d be in a big font (14), in plain Sans-serif style (e.g., Verdana), in a mil d color. Headings and titles shoul d be used. G7 The interf ace shoul d be responsiv e.

N1
Navig ation shoul d be consisten t and similar in ev ery pag e/section. N2 The website and ev ery mobile applica tions shoul d have a sim ple and logical structure. Even the firs time, the user shoul d be able to easil y navig ate inside, and shoul d remember the navig ational inf orma tion ev en at successiv e visits or uses. N3 Add navig ation inf orma tion and navig ation buttons at the top and the bottom of the pag e. N4 Efficiency and availability .

U1
All ow customisa tion. U2 Try to eng ag e the user . U3 Make adaptiv e the inter action with users, considering their inter action history , their pref erences, requests, and needs. U4 Decom pose the tasks into sim ple subtasks. U5 The number of errors shoul d be limited.

L1
The languag e shoul d be sim ple and precise. L2 Acron yms and abbrevia tions, non-liter al text , and jarg on shoul d not be used.
paper [39 ] presen ts the firs study to use ey e-tracking technol ogy with a set of ad ul t users with ASD, in order to ev al ua te text documen ts with specifi features, and it provides specifi guidelines for crea ting accessible text for autism. The text was combined with photogr aphs and symbols. The outcome of the study is tha t autistic users pref er texts tha t are paired with imag es, moreov er, both photogr aphs or symbols seem to work well. Note how ev er, tha t the study was done on a set of ad ul t autistic users without dev el opmen tal dela ys. As the authors sug gest , this resul t might be di fferen t with chil dren, since the symbolic understanding users with ASD arriv es la ter in their liv es compared to neurotypical users. The work [23 ] sug gests tha t for chil dren with dev el opmen tal dela ys photogr aphs seem to be more understandable. Other issues are the copious use of pictures and of red undan t represen tation to sim plify the concepts absorption. Moreov er, pictures shoul d not be used when they are non-relev ant or too abstr act to help the text comprehension. Finall y, a gener al and fundamen tal aspect is the adapta tion of the graphical layout to any device, to its orien tation, and to the dimensions of the Web browser windows: the graphical interf ace shoul d be responsiv e.
The structure and the navigation shoul d be sim ple, consisten t and logic. The users shoul d fin all the navig ational buttons and all the necessary tools in a immedia te way. In case of webpag es, the navig ation inside the site shoul d be limited by three clicks, links shoul d not be broken. The site shoul d be responsiv e, i.e. connection shoul d be fast and windows shoul d be resizable.
The user shoul d giv e space to adaptiv e personalisation: curren tl y, customisa tion is applied; eng ag emen t is a very importan t issue. Adaptivity is an open challeng e. In order to eng ag e users, in [23 ] the authors add, to the design of a dedica ted website, some games. These games have resul ted into a deep eng ag emen t betw een the users with ASD and the site. On the other hand, [14 ] introd uces the concept of participa tory design of user interf aces, i.e., users with ASD highl y benefi of personalised interf aces. In this direction we fin interesting resul ts in [23 , 40 ], and also in [41 ], were the authors discuss a participa tory design process experimen ted with four chil dren with autism, to dev el op their own smart object. The aim and the resul ts were to go beyond functional limita tions and to eng ag e the chil dren with ideas, desires and problems. Thus, if the uses are eng ag e, they will be willing to revisit the site o reuse a certain applica tion.
From the usability poin t of view , in the dev el opmen t of webpag es, Tarpin-Bernard et al. in [42 ] propose an environmen t, called the Cognitive User Modelling for Adaptive Presentation Of Hyper-Documents (CUMAPH), 6 EAI Endorsed Transactions on Ambient Systems 12 2016 -05 2017 | Volume 4 | Issue 13 | e5 used to adapt the webpag es to the user' s profile The adapta tion is based on four specifi sectors such as: atten tion, memory , languag e and visuospa tial, and from them some cognitiv e indica tors are extr acted and used to adapt the webpag e. Reg arding softw are tools, in [32 ] the authors propose the use of Adaptiv e User Interf aces (AUI's), in order to adapt the system to the specifi user . To dev el op these interf aces they buil d the inter actions rules of the user with the softw are, based on the Executive Functions. These functions regula te sev eral cognitiv e aspects such as cognitiv e flexibilit , planning, working memory , inhibition, and sustained atten tion, which are strictl y rela ted to the cognitiv e processes used to accom plish tasks. Wha t the experimen tal resul ts show is tha t the way the applica tion decomposes the task seems to mostl y affect the usability since when proper ly done it decreases the cognitiv e load on the user . This decom position all ows the user to easil y accom plish the tasks.
Finall y, the use of the language shoul d be sim ple and precise: it is well known tha t people with ASD liter all y interpret the text conten t, and have problems understanding metaphors and abstr act sen tences.

Websites for Users with ASD: a Systematic Comparison
In this section we presen t a systema tic comparison among accessible and usable websites for people with ASD, and we summarise in tw o separ ate tables which compl y with all, or parts of guidelines presen ted in previous Section 3.
We have done an extensiv e search in the Web for sites whose authors have claimed to foll ow di fferen t accessibility standards (e.g., are complian t with W3C standards for HTML and CSS, can be displa yed correctl y in curren t browsers. etc.). Although this search is not exha ustiv e, we have noticed tha t most of these sites are of autism associa tions and autism conf erences, and they are mainl y directed to researchers, paren ts or ad ul ts with autism.
In the foll owing Tables 2 and 3, we ref er to all the poin ts men tioned in Section 3 reg arding both accessibility and usability features. We ref er to them as G1 up to G7 for the 7 poin ts of the Graphical La yout; N1 to N4 for the 4 poin ts of the Structure and Navig ation; U1 to U5 for the Users, and final y L1 and L2 for the Languag e.
Since none of the websites and apps satisfy U3, we avoid to insert this poin t in our anal ysis. The outcome can be (the guideline has been respected), (no), or (partiall y). While navig ating on the sites we have noticed tha t some of them have similar features, e.g., we have seen tha t in all of them navig ation is consisten t, how ev er not very sim ple. The languag e is sim ple in some sections but others connect to man y links outside and provide too much inf orma tion. There is a lot of secondary conten t inside some pag es, and this conten t is not sim ple, and there are no feedbacks. Tasks are not divided un to subtasks except in few sites. [52 -57 ] contain man y pictures, but not the other sites, the same hol ds for the text tha t is short onl y in [46 -48 , 54 -57 , 63 ]. [49 -51 ] contain some moving text. [43 -45 ] lack of imag es, the gener al design is not very sim ple. [46 -48 ] have few more imag es. [58 ] contains man y pictures and tries to eng ag e the user with pictures and videos. [59 , 60 ] foll ow most of the guidelines, but wha t they reall y lack of is the eng ag emen t of the user and the subdivision in sub-tasks. The sites contain lots of inf orma tion, in some parts the text is too long, and is accessible for users with ASD which are high functioning, able to read and to comm unica te. [61 , 62 ] respect the graphical layout specific tions, how ev er in di fferen t parts the text is too long. Navig ation is coheren t how ev er more buttons and navig ation inf orma tion in the bottom woul d help. There is no user eng ag emen t. Most of the websites are available, there are few exceptions of sites tha t contain some broken links. Some websites are not resizable, thus do not work proper ly on mobile devices. The sites men tioned in [23 ] were designed to foll ow all the guidelines abov e presen ted (except for U 3). They are the firs exam ple of websites explicitl y dedica ted to users with ASD tha t independen tl y want to choose their own touring activity close to a specifi city (in particular , Rieti and Venice). The sites onl y lack of adaptivity , and of dynamical customisa tion of style attributes, on the other hand the users may independen tl y choose di fferen t navig ational pa ths depending on their own interests. To eng ag e the user the authors have added games and videos. The sites were tested on a set of users with ASD tha t have shown their grea t apprecia tion.
Summarising, with the exception of [23 ], all the websites we have anal ysed seem to be directed to users which are ad ul ts and high functioning (i.e., to users with mil d cognitiv e disabilities), and not to chil dren. Moreov er, most of the sites lack of eng ag emen t, responsiv eness, subdivision of tasks, and all of them of adaptivity .

Dedicated Applications for Users with ASD: a Systematic Comparison
Mobile applica tions represen t an importan t opportunity for users with ASD, as they may take adv antag e of the modality of inter action, e.g., the use of touch screen, and the manag eability of the device. Evidence sug gests tha t chil dren with ASD are more eng ag ed and verbal during their use. How ev er, there is a prolif eration of commerciall y available apps, which rang e from free to very expensiv e tools: unf ortuna tel y this lea ves very 7 EAI Endorsed Transactions on Ambient Systems 12 2016 -05 2017 | Volume 4 | Issue 13 | e5  [23 ] little room for quality control and the larg e majority of apps lack any founda tion in theory or research ev al uation [71 ]. Obviousl y, this is a big risk for a vulner able part of the popula tion. Table 3 summarises, for a set of curren t mobile apps, the implemen tation of the previousl y listed accessibility and usability guidelines. The guideline G7 is not incl uded, since all the apps satisfy this poin t.
A set of apps by Touch Autism [64 ] (like Social Stories Creator and Library, Turn Taker, Puzzle Spelling Words, and others) presen t some relev ant limita tions, mainl y located in the areas of graphical layout and navig ation: for exam ple, Puzzle Spelling Words uses an improbable font, starts using a background sound without eviden t control (it may be interrupted onl y by the settings panel), does not offer support at the navig ation (there are neither navig ation buttons, nor exit/pa use buttons). In addition, onl y one set of puzzles is free (Pla yground), while all the others require a pa ymen t. Findme [65 ] has been designed at the Univ ersity of Edinburgh to help chil dren improv e their causal and atten tions skills. It respects the major part of guidelines, but it does not offer navig ation support. The navig ation is more complete in the set of Apps for Autism by EdNinja [66 ]: it is possible open a sim ple visual men u. How ev er, the use of these apps appears to be complex. Niki Apps [67 ] is based on a set of apps based on AAC techniques: the apps presen t di fferen t graphical layouts, navig ation modalities and styles. The navig ation presen ts some limita tions (there are some parts of the app in which it is di fficul t fin the exit); how ev er, it is possible to draw a sketch but it is not clear where is the saved imag e and in which way it coul d be used. Bel onging to the same AAC categ ory are the Prol oquo apps [68 ]: Proloquo4Text and Proloquo2Go. These apps have been crea ted for people who cannot speak, not specifical y for people with ASD; they appear too rich of imag es and conten t, in contrast with an essen tial layout. Autism iHelp Apps [69 ] are vocabulary teaching aids dev el oped by paren ts of a chil d with Autism and a speech-languag e pa thol ogist. There are a set of apps: Same and different; Opposites; Colors; etc.. They are sim ple to use and propose concrete pictures, but they have some limits: the navig ation is linear and is not possible to return back; the concl usion of any activity is not predictable; the activities are not reprod ucible in the same way. Finall y, the prototype of mobile app for ASD people is Volo [70 ]; based on AAC techniques, it uses zz-structures, which are hyperorthog onal, non-hier archical structures for storing, linking and manipula ting da ta. This prototype has been modelled in order to be accessible and usable, the aspects rela ted to the user customisa tion and eng ag emen t coul d be improv ed.
Summarising, we note tha t most of the apps provide tools for editing and adding new and ev entuall y personal da ta, but importan t limita tions involve the process of customisa tion (often di fficul t to realise), the user eng ag emen t and mainl y the lack of user adaptivity .
Some apps use sketched imag es, other real pictures, most of them provide the user with an initial set of pictures and all ow the import of new imag es from a personal computer , a camer a, etc. (see, e.g., [64 , 67 ]). Another feature is the possibility of adding sounds, which can be syn thetic or natur al (see, e.g., [72 ]), or can also be recorded (see, e.g., [67 ]). Some apps all ow the crea tion of calendars: the dail y routine might be org anised in sequence of actions which describe the activities of the day in a fixe tem por al order .
Differen tl y from websites, the apps are conceiv ed for chil dren and they address gener al issues, not alw ays specifical y for people with ASD.

Conclusion and Future Challenges
In this paper , we discussed about possible guidelines for dev el oping accessible and usable websites or mobile applica tions for users with ASD. We have also anal ysed and compared man y of the existing websites and applica tions in order to check which compl y with all or parts of these guidelines.
As future challeng es, we have noticed tha t all the sites and applica tions tha t we have tested lack of a feature tha t represen ts an innov ativ e challeng e: adaptivity tow ards users. Automa ticall y, the systems shoul d be able to adapt their beha viour , considering the history of the users' inter action, their requests, needs and pref erences. Reg arding usability , responsiv eness is often neglected and very few websites take into accoun t the division in subtasks. These features are very importan t and non-trivial. Another issue is rela ted to the presen t syn thesisers available in di fferen t applica tions. We have noticed tha t man y of them prod uce sounds which are not easil y recognisable by users with limited comprehension. The adapta tion of the languag e and also of the voices woul d highl y improv e the quality of these applica tions.