Thema: AI for Health

DozentIn(en): Prof. Roland Eils, Julius Upmeier zu Belzen, Benjamin Wild

Maximale Teilnehmerzahl: 4

Zeitraum/Vorbesprechungstermin: nach Absprache

If you have a specific interest or a project idea, we’d love to discuss that in our first meeting. Alternatively, we have prepared two directions to explore: 1) predicting fractures in chest X-ray images, and 2) linking gene expression to tissue types and other phenotypes. Both of these, align partially with existing research projects in our lab, that we would love to try out in these new contexts. In any case, we will develop a pipeline from dataset access, preprocessing, baseline models, ML models and evaluation, as well as planning of future applications and extensions.
We are flexible with regards to the project timeline. If you have any further questions, please feel free to email us:
benjamin.wild@bih-charite.de, julius.upmeier@bih-charite.de

Ort: Digital (A), BIH, Kapelle-Ufer 2 (B)

Kurze inhaltliche Beschreibung:

Project: Machine Learning in Medicine: from idea to tool

• Learn the fundamentals of developing, training and testing deep learning models in the medical domain
• Learn about relevant metrics for evaluation and benchmarking and potential biases to watch out for
• Work on integrating the developed and evaluated models into usable (web) tool

Quantitative Aufteilung: (in %)

Praktische Programmierarbeit: 75%
Soft Skills: 25%

Verwendete Programmiersprache(n): Python (>90%), maybe some javascript for web app

Schwierigkeitsgrad (Acht Sterne verteilt auf drei Bereiche):

A Programmieren ****
B Biologie/Chemie *
C Projektmanagement ***

Erforderliche Vorkenntnisse:

• Experience with the Python programming language
• Fundamental understanding of “What is machine learning"
• Preferably prior experience with PyTorch or other DL-Libraries
• Understanding of neural networks and preferably experience with deep learning

Kontaktadresse, Webseite/Link:

Benjamin Wild
Julius Upmeier zu Belzen
https://www.hidih.org/research/ailslab