One key component of wound healing is re-epithelialisation, during which cells at the edge of the wound come together to close the wound gap. This process involves cell division, cell migration and changes to cell shape. However, the extent to which these different cell behaviours contribute to wound healing is challenging to interpret. Here, Jake Turley and colleagues apply deep learning models to analyse cell behaviours in the wounded Drosophila pupal wing. The pupal cells express markers of their cell membranes and nuclei, allowing the authors to apply their previously published deep learning models to automatically analyse the cell shape changes and cell divisions that occur during wound healing across a large imaging dataset. They also use single-particle tracking to follow the nuclei and thus analyse cell migration. After charactering these three cell behaviours in wild-type pupae, the authors go on to perturb damage-induced signals in the wing. They find that blocking the calcium wave (the first signal triggered by tissue damage) perturbs all three cell behaviours. By contrast, blocking JNK signalling affects cell shape changes and proliferation, and blocking the wound inflammatory response affects cell migration and proliferation. This study demonstrates how deep learning tools can be applied to explore the interactions between the complex cell and molecular mechanisms that underpin wound healing.