Skip to main content

Symptoma, Better Diagnosis for Patients with Rare and Complex Diseases

Deliverables

D2.2 Release question sequences derived from deep learning algorithms.

D2.2 Release question sequences derived from deep learning algorithms. T2.2 Train chatbot: utilizing the same case reports as in WP1. However, instead of searching with all symptoms extracted from the respective case, we start with one symptom only. Deep learning algorithms will then arrive at the best question sequence uncovering the other symptoms thus leading to the right diagnosis. Question sequences should then work for all 44,000 conditions in our database while accounting for disease incidences (a more common disease should have a higher priority than a rare one). T2.3 Test chatbot: in production. As in WP1, we will release question sequences to production, monitor search signals indicating successful questions, and continuously optimize for it.

Searching for OpenAIRE data...

Publications

Vom Symptom zur Diagnose – Tauglichkeit von Symptom-Checkern

Author(s): J. Nateqi, S. Lin, H. Krobath, S. Gruarin, T. Lutz, T. Dvorak, A. Gruschina, R. Ortner
Published in: HNO, Issue 67/5, 2019, Page(s) 334-342, ISSN 0017-6192
DOI: 10.1007/s00106-019-0666-y

Querdenker-Preis

Author(s): DGIM
Published in: DMW - Deutsche Medizinische Wochenschrift, Issue 144/15, 2019, Page(s) 1085-1085, ISSN 0012-0472
DOI: 10.1055/a-0954-8989