In early 2026, seven of us in our Project Management course at emlyon built a full project dossier for EHPAD Korian Les Terrasses de Blandan, a care home in Lyon with 110 residents. Elderly residents are at higher risk of dehydration, especially during heatwaves, because of reduced thirst sensation, limited mobility and cognitive impairment, and Lyon’s summers aren’t getting any cooler. Most facilities still monitor hydration reactively, with the same manual checklist for every resident regardless of their individual risk. Our proposal was an AI-driven risk scoring interface that combines resident health data with weather forecasts to give nurses a daily prioritised list of who needs attention first, with nurses always able to override the AI’s call.
My main piece was the data model, the UML class diagram that maps what information the system needs and how it all fits together. I worked out which data the tool would draw on, from heart rate, sodium and creatinine levels in the patient health record to temperature, humidity and heatwave alerts in the forecast, and how each piece relates to the others. Patient data and weather data both feed into a central set of dehydration risk scores, which roll up into a daily priority list and connect to each nurse’s profile and assigned residents, so the interface can show every nurse who to see first.

I also helped write the slides framing the problem, contributed to a couple of iterations of our work breakdown structure, and presented the system design to the class, walking through both our use case diagram and the data model. Beyond that, I was in charge of visually streamlining and editing the whole slide deck, so that seven people’s work came together as one coherent presentation.
Looking back with a few more months of study behind me, there are two things I’d change. A single day’s forecast applies to every resident, so weather data should connect to risk scores as one-to-many rather than one-to-one, and the current risk score is stored in two places, which risks the two drifting out of sync.
Built with Caitlin Chesnais-Dixon, Xintian Huang, Mélissa Laine, Thomas Junior Nembot, Christabelle Nguessie Essofack Matsinkou and Océane Thouvard-Kwek.
