The Mission

We are building a startup that researches and develops prototype machine learning models to support clinicians across different areas of mental health. For this challenge, we focus on predicting specific adverse behaviors, such as suicide, self-harm, and aggression, among patients with particular diagnoses in specific clinical settings. By giving clinicians earlier and more reliable risk signals, we aim to help them intervene sooner and improve patient safety and outcomes.

The Challenge

Clinicians often struggle to predict when a patient is at risk of harmful behavior. Current risk assessments rely heavily on clinical judgment and standard questionnaires, which can be time-consuming and inconsistent, and often fail to identify risk in time. This leaves both patients and staff vulnerable and makes it harder to target care where it is needed most.

The solution

We propose AI models trained on clinical data that predict the risk of specific adverse behaviors for defined patient groups and settings. Because each model is tailored to a particular diagnosis and clinical context, its predictions are more relevant and accurate than one-size-fits-all approaches. The models are designed to support clinicians, not replace them, giving them an extra tool to spot risk earlier and decide on timely interventions.