The published evidence behind Therapoice.

Up to one in five Australian school children is neurodivergent. Their therapy outcomes depend almost entirely on what happens at home, between sessions. The current system gives families a set of exercises and hopes for the best.

It does not work. Not because families do not care, but because they have no time, no confidence, and no structure. The published research on this is consistent across studies, countries, and disciplines.

Up to 20% of Australian school children are neurodivergent. This is the largest and fastest-growing category of disability in the Australian education system. (Square Holes Research, 2025)

Autism affects about 3 to 4% of school-aged children. The Australian Bureau of Statistics recorded 290,900 Australians diagnosed with autism in 2022, a 41.8% increase from 2018. Prevalence is highest under age 25, at 3.1% versus a population average of 1.1%. ADHD affects roughly one in twenty Australian children, with around 800,000 Australians estimated to have ADHD according to Murdoch Children's Research Institute.

739,414 Australians are on the NDIS as of 30 June 2025, an 11.8% increase from 661,267 in June 2024. Autism is the most common primary disability on the NDIS at around 32% of all active participants. Around 170,000 children receive NDIS early intervention supports including speech pathology and occupational therapy. Children under 7 receive an average of $16,700 per year in NDIS funding. (NDIS Quarterly Report 90, June 2025)

The federal and state governments have committed $4 billion over 5 years to the Thriving Kids and Foundational Supports program for children aged 8 and under with developmental delay and autism. The Australian Government also released the National Autism Strategy 2025 to 2031 and a National Roadmap to Improve the Health and Mental Health of Autistic People 2025 to 2035. Sustained policy priority is in place for the next decade.

Wait times tell the same story. One in three NDIS providers report waitlists of more than six months, particularly in regional and rural areas. Only 2.3% of speech pathologists in rural and regional areas report being able to recruit staff. (Speech Pathology Australia, 2024)

This is the centre of the problem. The most cited number in the field comes from a 2024 study by ASHA, the American Speech-Language-Hearing Association.

59% of caregivers reported completing assigned speech home exercise programs less than 75% of the time. 88% of families faced at least one psychosocial barrier.

The barriers are not motivational. The most frequent barriers reported in the same study were lack of time, lack of confidence, caregiver stress, parental anxiety about doing techniques wrong, and managing multiple children with competing activities. (ASHA Perspectives of the Special Interest Groups, 2024)

A 2025 UK qualitative study, "Don't Call It Homework", found that parents frequently felt overwhelmed and underprepared. Labelling activities as "homework" actively discouraged engagement. (SAGE Journals, 2025)

A 2024 scoping review of facilitators and barriers to caregiver adherence to home therapy identified the four strongest predictors of adherence as routine, caregiver education, a positive therapist relationship, and perceived benefit. Each of these is something a guided system can deliver, and a paper handout cannot. (PubMed, 2024)

The clinical recommendation is 2 to 3 therapy sessions per week for children with speech sound disorders. Most children receive one. Home practice is the only realistic way to bridge that gap. (Pearson Assessments, ASHA Evidence Briefs)

A widely cited 2017 systematic review of parent-implemented home programs found that home programs are more effective than no intervention when three things are in place: high dosage, parents directly trained, and the therapist maintains frequent contact with the family. All three are engineering problems Therapoice can solve.

A 2026 peer-reviewed RCT by Raaz et al. compared online parent training plus structured home practice with high-frequency clinic therapy. Both groups achieved statistically significant improvements in speech sound accuracy. Structured home practice was a viable and effective alternative to clinic-only delivery. (SAGE Journals, 2026)

A 2025 scoping review identified 30 studies meeting eligibility criteria from a review of 188 papers on AI-driven technologies for managing paediatric speech and language therapy. The headline finding: computer-based tools are more engaging for children than traditional therapy, offering personalised plans and real-time feedback. (PMC and SAGE Journals, 2025)

A 2025 ScienceDirect review confirmed that AI tools enable therapists to monitor progress and adjust treatment remotely, which is the exact use case Therapoice supports through the therapist dashboard. A 2025 ACM IDC review titled "Is AI Ready to Support Speech Therapy for Children?" found that AI-enabled mobile apps show clinically meaningful outcomes, but it also flagged the gap between research tools and consumer-grade products built for daily family use. That gap is the gap Therapoice fills.

On completion rates, structured digital therapy programs report an average session completion rate of 85.6%, with a range from 68.2 to 100%. Unstructured paper home programs sit below 75% completion for 59% of families. The difference is structure, not effort. (PMC systematic review, 2025)

CognitiveBotics, a US-based AI therapy platform, ran a 12-month observational study on 43 children aged 2 to 18 with autism. The platform was found to be an effective supplement to continuous therapy, improving cognitive, social, and developmental outcomes. (JMIR Neurotechnology, 2025)

Voice-based agents are particularly well suited to autistic children because they offer predictable, consistent, low-stress interaction without the social complexity of human conversation. (Conversational Agent Utilization Patterns, Springer, 2024)

A study on Wysa, a voice-based agent, found autistic users engaged on average for 8.59 days with a median of 97 conversational exchanges. Engagement was nearly daily once the tool was introduced.

A 2026 SpringerLink chapter titled "Exploring the Potential of Conversational AI for Supporting Children with Autism" confirmed that voice AI creates safe, low-stress practice environments that replicate real social scenarios at home, without the anxiety triggers associated with face-to-face interaction.

The NDIS provider sector reached $45 billion in total revenue in 2025 to 2026, growing at 9.3% per year over five years. Assistive technology supports were delivered to 36% of all NDIS participants in 2024 to 2025. The early childhood allied health segment is one of the fastest-growing parts of NDIS spending. (IBISWorld, AIHW)

The broader Australian paediatric healthcare market was valued at USD 279.82 million in 2025 and is projected to reach USD 411.13 million by 2034 at a 4.37% compound growth rate. The Australian home healthcare market, which includes rehabilitation therapy services, was USD 13.78 billion in 2024 and is growing at 11% per year to 2030. Rehabilitation therapy services hold a 60.96% revenue share within that market. (IMARC Group, Grand View Research)

On the public funding side, the Australian Government has invested $30 million in AI health research through the Medical Research Future Fund, including a $57 million round for AI-based health transformation grants. The Industry Growth Program funds innovative products in National Reconstruction Fund priority areas, including Section 5(4) Medical science and Section 5(7)(b) AI technologies. Therapoice qualifies under both.

International signals add to the case. The UK government has funded the Early Language and Support for Every Child (ELSEC) pathfinder project, a government-validated trial of innovative workforce models for speech and language support. A 2025 UK parliamentary briefing confirmed demand for speech therapy far outstrips supply, with 65,114 children waiting in November 2024 and 45.6% waiting more than 12 weeks.

We are not pretending the evidence is complete. Here is what is missing, and what we use instead.

• No Australian RCT on AI voice therapy for children. We rely on global evidence, primarily the CognitiveBotics and Wysa studies, and reference them as proxies.
• No published Australian data on NDIS home program adherence specifically. The ASHA 2024 study covers speech therapy broadly. We cite it explicitly as a proxy.
• No competitor product with public Australian outcomes data. CognitiveBotics, US-based, is the closest comparable. The absence of an Australian competitor is itself part of the opportunity.

Building the evidence base for Australia is part of what Therapoice will do. The pilot with Kind Corner is designed to produce that data.