Demo accepted at IEEE WiMob 2026: unfolding a DSM-5 assessment down to its formula

Our demonstration Tracing a DSM-5 Depression Assessment from Score to Formula on Edge Hardware has been accepted at IEEE WiMob 2026, the 22nd International Conference on Wireless and Mobile Computing, Networking and Communications, in Avignon, France, from 14 to 16 October. Congratulations, Tibor!

The demo takes the CML Framework from Tibor Haller’s Master’s thesis and puts it on a table where anyone can take it apart. A mini-PC runs the full pipeline and dashboard with no cloud and no uplink, and attendees drive it themselves: run an assessment over a seeded 14-day workload, read whether an episode is likely and which of the nine DSM-5 indicators support that call, then unfold any indicator down to the closed-form operator, its inputs, and the weight of every biomarker behind it.

The third step is the interesting one. On identical input, an attendee swaps the active context profile, say from solitary and digital to social gaming, and the same evidence is re-weighted live. Nothing is learned end to end, so the explanation is not a story told about the computation after the fact, it is the computation, and every run is deterministic and re-runnable. Everything shown runs on synthetic data with no detection accuracy claimed: what is on display is the decision procedure.

The pipeline and dashboard are open source, and the video walks through the dashboard:

Player not loading? Watch on YouTube ↗