Clinical AI reasoning for multimorbid patients: feasibility of calibrated differential diagnosis and treatment-plan evaluation

• Unité de recherche: INNOSUISSE
• Numéro de projet: 139.361 INNO-ICT
• Titre du projet anglais: Clinical AI reasoning for multimorbid patients: feasibility of calibrated differential diagnosis and treatment-plan evaluation

Mots-clé (Anglais)

ICT, eHealth

Résumé des résultats (Abstract) (Anglais)

Patients with several simultaneous conditions drive a disproportionate share of diagnostic error and adverse drug events. ahtida and ZHAW establish the scientific basis for AI reasoning that ranks several conditions at once with calibrated uncertainty, and a valid way to evaluate treatment plans.