AI enters the most human-dependent industry: a fourth-tier city rehab clinic sees 40% profit growth

In a rehabilitation center in Pingdingshan, Henan province, a young therapist is conducting social skills training with an autistic child. On the screen before her, RICE AI — developed by Damihexiaomi, a leading Chinese special-needs children's rehabilitation firm — is recording behavioral data, analyzing intervention progress, and generating next-stage course recommendations. This is not a future scenario but an ongoing transformation, already operational for over six months in cities like Pingdingshan, Cangnan, and Lianyungang.

China's special-needs rehabilitation sector faces a structural talent shortage. According to Ministry of Education data, there were 912,000 special education students nationwide in 2023 but only 77,000 full-time teachers. When including children with autism, speech delays, and other conditions, industry estimates suggest demand in the tens of millions, with a practitioner gap in the millions. The bottleneck is not just quantity but quality — the sector lacks structured accumulation of intervention methods and data. Trainee therapists require years of supervision from experienced mentors, who are in critically short supply.

In October 2024, Damihexiaomi launched RICE AI, a generative AI system for the full special-education rehabilitation workflow. Wang Zhanwei, the company's VP of R&D, described the positioning: "AI cannot replace therapists; AI's value is to make the rehabilitation process from assessment to intervention simpler, enabling more people to do this work, reducing costs, and allowing more children to receive quality services." RICE AI intervenes at three stages: automated assessment reporting (cutting hours to minutes), personalized intervention plan generation based on massive case data, and decision support for junior therapists approaching experienced supervisor quality.

A third-party study of 65 cases (average age 4.3 years) showed that the RICE AI experimental group achieved significant gains across five key domains: social interaction, verbal communication, non-verbal communication, cognitive performance, and imitation ability. The most notable improvement was in non-verbal communication, with post-test scores rising 5.3 points, followed by verbal communication at 5.2 points — both significantly outperforming the control group.

The business impact is equally striking. Pingdingshan Sunflower, an early RICE AI adopter, reduced per-child assessment-to-plan time from two-to-three days to two-to-three hours — a 10x efficiency gain. After launching a "RICE Social Training Course," one parent purchased 1,000 sessions upfront. The clinic's profit grew 40% year-over-year in the first half of 2026. A similar pattern emerged at Fangzhou Rehabilitation Center in Cangnan, Wenzhou, which added over ten new students within a week of deployment.

RICE AI's technical moat rests on Damihexiaomi's 12-year accumulation of over 40,000 real cases, 200 million intervention data points, and 3 million supervisor decisions. Wang explained that general-purpose LLMs lack domain-specific corpus and context understanding to generate usable intervention plans. RICE AI fine-tunes on this proprietary vertical dataset to produce industry-standard personalized plans, covering the full workflow from assessment and planning to lesson preparation, teaching, feedback, and goal adjustment.

In 2025, RICE AI opened to external institutions, transitioning from an internal tool to an industry infrastructure. Over 300 institutions have registered, with more than 100 using it regularly across provinces including Henan, Hubei, Jiangsu, Zhejiang, Guangxi, Hainan, Guangdong, Shandong, and Xinjiang. Wang said Damihexiaomi is "transforming from a chain intervention provider to industry infrastructure."

A key milestone ahead: the RICE AI School-Age Edition launching in August 2026. Focused on classroom adaptation, school integration, and academic support, it has already attracted applications from over 100 schools including special education schools, regular schools, and kindergartens, signaling strong demand for extending AI-assisted support beyond early childhood into the school years.