Trust Stamp Inc (NASDAQ:IDAI, ISE:AIID) has filed a US provisional patent application for an artificial intelligence-based medical diagnostic support system aimed at helping physicians evaluate diagnoses while addressing common limitations of large language models (LLMs).
The patent application, titled "Medical Diagnostic Assurance System" (MDAS), was filed on June 15, 2026. The company said the technology is designed to reduce issues such as confirmation bias and AI hallucinations that can produce inaccurate or fabricated information.
According to Trust Stamp, MDAS is intended to function as a self-contained sovereign platform that tokenizes patient identities to help safeguard medical records from third-party access. The system is designed for use by qualified medical practitioners.
Rather than generating an initial diagnosis, MDAS uses a three-model consensus process to independently assess and either challenge or support a physician's diagnosis without first being exposed to that original assessment, a design the company said is intended to limit bias.
Trust Stamp CEO Gareth Genner said that reducing sycophantic behavior in LLMs is particularly important in healthcare, where unreliable AI outputs could have serious consequences.
He added that the company is negotiating a pilot program for MDAS and expects the technology to go live in the first quarter of 2027.
The filing expands the company's intellectual property portfolio, which includes 27 issued or allowed patents and seven pending patent applications related to artificial intelligence, biometrics and cryptography.
