Structured Turkish review of VR, mobile apps, telerehab, and AI for stuttering: VR has the strongest evidence of the four, but research remains preliminary

Background

Technology-based tools for supporting people who stutter have expanded rapidly across multiple domains: virtual reality (VR) for exposure to realistic speaking situations, mobile applications for self-monitoring and home practice, telerehabilitation for remote service delivery, and artificial intelligence (AI) for automated speech analysis. By 2024 no published review had structured the landscape across all four domains in one place, and the Turkish/regional perspective was particularly underrepresented in the English-language literature. Erol and Türker Şener set out to provide that overview, with particular attention to the Turkish patent and thesis landscape alongside the international peer-reviewed evidence.

Stuttering is described in the review as a communication disorder characterized by involuntary disruptions in speech fluency (sounds, syllables, words), with approximately 5% prevalence in early childhood and 1% lifetime prevalence in adults, and with a 4:1 male-to-female ratio in adolescence and adulthood. The review emphasizes the social, psychological, and emotional consequences of stuttering - reduced self-confidence, anxiety in everyday speaking situations, and impact on educational and occupational participation - as the clinical targets for which technological adjuncts are most likely to provide complementary support.

What the researchers did

The authors conducted a structured literature search across six databases - PubMed, Google Scholar, Scopus, ResearchGate, YÖK Thesis (Turkey’s national thesis archive), and Web of Science - covering 2000-2024. Search terms paired ‘stuttering’ with each of: ‘virtual reality’, ‘VR’, ‘mobile applications’, and ‘telerehabilitation’. The initial search identified 55 studies.

Exclusion criteria: studies published in languages other than English or Turkish, full-text inaccessibility, absence of a technological rehabilitation intervention applied by speech and language therapists, presence of additional integrated therapy alongside the technological intervention, and incomplete outcome assessments. After exclusions (other diseases n=14, no open-access full text n=15, non-English/Turkish n=1, unevaluated n=12), 13 studies met inclusion criteria.

The review also surveyed six AI-related patents from Espacenet (international) and the Turkish Patent and Trademark Office, covering mobile-app voice analysis systems, training/therapy devices, biomedical analysis algorithms, speech-correction methods using transformer language models (T5, BERT, VITS), and real-time speech-recognition systems for personalized therapy reporting.

The review is structured around four technology domains (VR, mobile apps, telerehabilitation, AI/patents) and is followed by a synthesis section on effectiveness drawn from the 13 included studies and from cited landmark studies in the field.

What they found

Virtual reality. The review identifies VR as the most empirically supported of the four domains. Findings drawn from the cited literature include:

• VR environments produce communication responses (disfluency patterns, physiological arousal, self-reported anxiety) comparable to real-world speaking, citing Brundage et al. 2006 (challenging vs supportive virtual job interviews), Brundage et al. 2016 (VR audience eliciting subjective distress), Brundage & Hancock 2015 (VR audience eliciting responses comparable to live audiences), Moïse-Richard 2021 (virtual classroom anxiety in school-age children/adolescents), and Almudhi 2022 (adaptation effects in real vs VR environments).
• VR-based exposure therapy for social anxiety in stuttering has been pilot-tested with cautious results - the review cites Chard et al. 2022 (scoping review identifying promising design directions) and Chard et al. 2023 (pilot RCT with limited short-term effects, requiring larger samples and personalized protocols).
• The Walkom 2016 conference paper is cited (reference 24) as one of the foundational works exploring VR exposure therapy for stuttering anxiety.

Mobile applications. Described as digital tools for self-monitoring and structured home practice. Specific apps cited include BroiStu and SpeakUp. The review frames mobile apps as complementary to clinician-led therapy rather than as standalone treatments.

Telerehabilitation. Positioned as an access extender - particularly valuable for individuals in rural or underserved areas. The review notes that telerehabilitation efficacy is supported by emerging research but that comprehensive implementation, particularly in developing countries, is hindered by technological, logistical, and regulatory constraints.

Artificial intelligence and patents. This is one of the more distinctive contributions of the review - a survey of six AI-related stuttering-technology patents:

• Turkish patent 2022/016828: Mobile application system using voice-analysis software and AI to enable self-directed therapy (slow speech, soft speech, voluntary stuttering) without a speech-language therapist present.
• Turkish patent 2008/09165: Training and therapy device for attention deficits, auditory perception disorders, asymmetric and painful hearing in individuals with stuttering.
• US2021315516A1: Biomedical analysis and treatment algorithms for stuttering and rhotacism, using voice-assistant technology and digital applications for data collection and personalized therapy.
• CN118553248A: Speech-correction method using T5 (error correction), VITS (speech synthesis), and BERT (language representation) language models integrated for stuttering correction.
• US2024386813A1: Speech-recognition system using machine learning to analyze voice signals in real time and generate personalized speech-therapy reports (pitch, intensity, phonetic structure).
• One additional unnumbered application focused on speech transcript generation, contextual linkage, and correction-then-synthesis pipelines.

The review notes that none of these patents have associated clinical efficacy data and that AI in stuttering remains theoretical/emerging.

Synthesis section. The review collates findings from the cited studies on stuttering responses to VR job interviews, VR audiences in adults, virtual classrooms in school-age children, real-vs-virtual environment comparisons, virtual reality exposure therapy (VRET) for social anxiety in stuttering, and adaptation effects. The authors note that VR has been shown to elicit physiological reactivity and subjective distress similar to real-world speaking conditions and that VR-supported rehabilitation may help reduce social anxiety associated with stuttering.

Why this matters

This review is useful as a multi-technology landscape orientation for clinicians and researchers asking “what has been studied in stuttering across VR, mobile, telerehabilitation, and AI?” It surfaces both the international peer-reviewed evidence (largely overlapping with the Brundage and Chard programs of work already represented in this Evidence Hub) and the Turkish-language and patent landscape that is less visible in English-language reviews.

For clinical recommendation: the review’s own framing is cautious. The authors explicitly write that “further high-quality, controlled studies are needed to confirm these outcomes and establish clinical efficacy.” Used as a citation, this review is best positioned as evidence that VR is the most empirically grounded of the four domains, not as evidence that any specific VR intervention has been proven effective.

For Therapy withVR specifically: this review does not use, evaluate, or compare against Therapy withVR. Walkom 2016 is cited (reference 24) as a foundational primary source, but Walkom (the withVR founder) is not a contributor to this review. The review’s editorial conclusion that VR is the most promising of the four technology domains is consistent with the Therapy withVR product design philosophy, but cannot be cited as evidence that Therapy withVR specifically has been validated.

Limitations

The review has the following limitations, some explicitly flagged by the authors and some inherent to the methodology:

• Only 13 studies included out of 55 identified, with substantial exclusion (n=42). The methodology section does not provide a formal PRISMA flow diagram or quality assessment of the included studies, so the basis for selecting “best 13” is not standardized.
• No meta-analysis is performed; the synthesis is narrative rather than quantitative.
• Most included primary studies are small. Sample sizes in the cited VR literature for stuttering range from n=3 to n=36 - the review inherits the small-sample limitation of the underlying evidence base.
• Patent-review component is descriptive, not evaluative. Patents reflect intellectual-property activity, not efficacy evidence.
• Publication-bias toward positive findings may be present in the underlying literature.
• Two-language inclusion (English + Turkish) is narrower than international reviews would typically allow, though it does surface Turkish-language thesis and patent material that other reviews miss.
• Journal context. The review is published in Istanbul Gelisim University Journal of Health Sciences, which is indexed in TÜBİTAK ULAKBİM TR-Index but does not have a major international impact factor. This is not a criticism of the review but is relevant context for citation weight.