Are Artificial Intelligence enabled diagnostic tools for mental health widening health inequalities for people with multiple long term conditions: A mixed methods study from the Methodologies Workstream

NHS clinics are already using artificial intelligence to help diagnose mental health conditions, but no one has checked whether these tools work equally well for people with multiple long-term physical illnesses. The problem is that AI diagnostic tools are typically tested on relatively healthy populations. People with multiple long-term conditions—such as diabetes, heart disease, or chronic pain alongside depression or anxiety—may be misdiagnosed or overlooked if the AI was trained on data that does not reflect their complex health profiles. This project will examine whether current NHS mental health AI tools widen or narrow inequalities related to age, sex, ethnicity, socioeconomic deprivation, and physical or cognitive disabilities. If this research finds that the tools are widening inequalities, the NHS could adjust or redesign them before they become standard practice. If the tools are working fairly, the findings would give clinicians confidence to use them more broadly. Either way, the project will produce practical guidance on how to build and evaluate AI diagnostic tools that do not silently disadvantage the patients who need them most.