Neuromnia, an innovative AI-driven platform, is pioneering solutions to tackle some of the most pressing challenges in autism care. With autism affecting one in 36 children, Neuromnia provides clinicians, parents, and teachers with powerful AI-driven tools that enhance productivity, improve treatment quality, and increase access to care for individuals on the autism spectrum.

Harnessing Llama 3.1, Neuromnia recently developed Nia, a human-centric AI co-pilot for Applied Behavior Analysis (ABA) therapy. Nia enhances clinician productivity and creates better access to care, allowing clinicians to focus on delivering quality care at scale.

Neuromnia’s journey with Llama 3.1 began during the research and testing phase of product development. In May 2024, the team started working with the 70B model, and it quickly became clear that Llama 3.1 offered the natural language processing performance necessary to support the company’s mission. They initially used the model to generate behavior intervention plans and skill acquisition recommendations based on client case histories, fine-tuning it with a curated dataset compiled by Co-Founder & Chief Product Officer Josh Farrow, a Board Certified Behavior Analyst (BCBA).

“After testing with Llama, we were impressed by its state-of-the-art performance in natural language processing tasks,” says Co-Founder & CEO Jay Gupta. “Combined with our CPO’s 15 years of clinical expertise, Llama has enabled us to build a comprehensive dataset that will help reduce the administrative burden for clinicians and provide quality care.”

For startups like Neuromnia, open source has been instrumental in building, testing, and scaling solutions. The availability of open source LLMs like Llama 3.1 provides a cost-effective avenue to develop production-quality models for complex tasks that would otherwise be prohibitively expensive. With open source, the team can leverage collective contributions and advancements made by the broader AI community to help refine and fine-tune their capabilities to address the specific needs of the ABA industry.

“Open source has empowered us to build and scale our solutions without succumbing to vendor lock-in,” adds Gupta. “The community support has been a welcome source of information and advice for LLM creation and deployment.”

Neuromnia specifically designed Nia to address ABA workforce shortages by automating key elements of treatment planning, documentation, and modification. By suggesting major treatment components—such as goals, intervention strategies, and procedures—Nia reduces the time clinicians spend on repetitive tasks and allows them to manage larger caseload efficiencies. The team at Neuromnia is also seeking to combat high burnout and turnover rates in the industry by boosting the productivity of ABA professionals through its AI-powered capabilities.

“Many clinics lack robust analytics capabilities,” Gupta notes. “Our platform provides automated, actionable insights to optimize and streamline the entire treatment process, from intake to discharge.”

With the release of Llama 3.1, Neuromnia has taken Nia to the next level. After a few early misconfigurations and bugs, the team was able to resolve issues through trial and error and quickly integrate Llama into their platform. To enhance the model’s output, they incorporated techniques like prompt engineering and retrieval-augmented generation (RAG). Neuromnia has also optimized Llama’s outputs to ensure accuracy and context-appropriate recommendations, training the LLM on clinician-created synthetic data to successfully generate, modify, and validate responses within Nia’s software.

Looking ahead, Neuromnia is committed to continuing its use of Llama, evaluating each new release to ensure its platform remains cutting-edge and capable of meeting the growing demand for autism care solutions.

“While out-of-the-box LLMs can sometimes struggle with technical or complex tasks, we’ve significantly reduced error rates through a combination of prompt engineering, RAG, fine-tuning, and techniques like semantic search,” explains Gupta. “These advancements help ensure that our solution will continue to deliver accurate, hyper-personalized results that make autism support accessible to every family.”

For more information, visit Neuromnia’s website.