When a student is struggling emotionally, the signs often appear in their writing, their typing speed, and their digital behavior long before they verbalize it. In this episode, Lucas and Luna explore Koko Analytics, a startup that trained an AI model on de-identified student counseling chat logs and classroom writing assignments to flag early indicators of depression, anxiety, and suicidal ideation. The company works with 14 school districts across three states, covering roughly 120,000 students, and claims a 78 percent accuracy rate at predicting a future mental health crisis within a 30-day window. Lucas walks through how the model works—analyzing latency between keystrokes, word choice shifts, and usage of absolutist language like 'always' and 'never'—without reading the content of any private conversation. Luna pushes back on ethical concerns: false positives, surveillance, and the risk of labeling kids. The hosts also discuss how Koko trains teachers to interpret the data, the role of school counselors as the final decision-makers, and why the startup refuses to sell to law enforcement. A nuanced look at the intersection of machine learning and adolescent mental health.
#KokoAnalytics #StudentMentalHealth #AIinSchools #MentalHealthDetection #Edtech #BusinessAndTechnology #FexingoBusiness #BusinessPodcast #LucasAndLuna #EducationStartup #SchoolCounseling #YouthDepression #MachineLearning #EthicalAI #PrivacyInEducation #EarlyIntervention #BehavioralAnalytics #StudentWellbeing