#10 — At the Intersections — July 18, 2026

From the electrical rhythms of a child's brain to the output of artificial intelligence, complex systems are constantly sending signals. On ComplexityPod, we look at what these signals reveal about health outcomes and approaches to learning. We begin with research on brain activity in pediatric intensive care. Today, we're considering the signals our bodies and technologies send, and what they can tell us. Right, we're looking at how biological markers can correspond with physical states, and even treatment outcomes. We start with the electrical rhythms in the brains of critically ill children. This is a situation where sleep is consistently disrupted, and the problem has been how to monitor it. A study looked at using a tool called amplitude-integrated electroencephalography, or aEEG, to do just that. They used it to track sleep-wake cycles in 118 children on mechanical ventilators. And they did find evidence of these cycles in most of the children. But the amplitude of those signals, the strength, was lower than in healthy children. It did increase after two months of age, though, suggesting a developmental aspect. And there's another pattern: the occurrence of these cycles increased the longer a child was in intensive care. The key finding, though, connects to outcomes. That's the critical part. Reduced sleep-wake cycling was associated with higher mortality. So this tool, aEEG, could be more than just a way to monitor sleep. It might be a way to continuously track a specific risk factor in this population. So, from the biological signals in the brain, we turn to the output of a different system: generative artificial intelligence. A qualitative study, involving Sayra M Cristancho, looked at how 15 multilingual educators are teaching writing now that these AI tools exist. And what they found wasn't just about the technical skills students might gain or lose. The educators were focused on the social and relational questions of literacy itself. They see writing as a practice that's deeply embedded in culture and language. And they're using that perspective to inform how they teach with these new AI tools. That's especially pointed in a world where multilingual writers can already face linguistic discrimination. The study notes these educators have both hopes and fears about AI's role in centering their students' voices. So it shifts the focus from just acquiring a skill—using AI—to what the study terms 'critical AI literacies.' It's about understanding these tools as part of a social and relational practice. For our last piece, we return to the brain, but this time to its chemical signals, not its electrical ones. This is research looking for neurobiological markers that could explain why some people with psychosis don't respond to antipsychotic medication. A project with Lena Palaniyappan sought to identify these markers. The work was a mega-analysis, combining individual data from over 1,100 participants, as well as a meta-analysis of 23 other studies. And they found a clear chemical signature. Compared to people who did respond to the medication, nonresponders had elevated levels of several neurometabolites in the medial frontal cortex. Chemicals like glutamate, choline, and myo-inositol. These levels were also higher when nonresponders were compared to healthy control individuals. And one in particular—myo-inositol—was most pronounced in individuals with treatment-resistant psychosis. These findings provide direct evidence linking nonresponse to specific elevations in brain chemicals. This supports continued work on treatments that act on different pathways, like the glutamate and inflammatory systems. It’s a potential path toward predicting outcomes and developing new treatments for this specific group. That's all for this issue. We will return next week with more research summaries from across the disciplines. That's it for today on ComplexityPod.