Gideon Vos can’t forget the moment his AI model finally explained itself—not just predicting schizophrenia, but revealing why in terms a doctor could trust. At James Cook University in Townsville, Australia, Vos and a multidisciplinary team of engineers, neuroscientists, and psychologists have developed 'explainable' artificial intelligence tools that help distinguish schizophrenia from acute stress using EEG brainwave data—offering new hope for early, accurate diagnosis. Schizophrenia affects about 1% of the global population and carries high mortality rates, often because symptoms emerge before full psychosis, making timely detection critical. Yet diagnosing it remains difficult, especially since stress can mimic or mask symptoms. Traditional AI models trained on EEG data struggle with this complexity, often failing in real-world settings where stress alters brain activity. But Vos’s team didn’t just train their model on standard EEG recordings—they adjusted for stress, using open-access datasets and machine learning algorithms designed to reflect how the brain responds differently under pressure in people with schizophrenia versus healthy individuals. The result? Models that not only classify more accurately but also provide transparent, physiologically consistent explanations that align with established medical science. This is the core of 'explainable AI': not replacing doctors, but empowering them with interpretable insights. As Vos puts it, 'We need the AI model to explain why it can, or why it can't, separate these two groups.' That transparency is what makes the technology trustworthy and clinically useful. For patients in remote or regional areas—where seeing a GP might take hours and a specialist consultation months—this could be transformative. Imagine a future where a smartphone app, guided by AI, flags early warning signs and connects users to care, giving clinicians both data and rationale to act quickly. The research, published in Biomedical Signal Processing and Control, marks a significant step toward integrating AI into mental health care responsibly. By grounding machine learning in medical understanding and prioritizing clarity over black-box predictions, the team is paving the way for AI that doesn’t just predict, but informs, supports, and ultimately humanizes diagnosis.