Data Science

Data Traffic Control - make your data files human-friendly

Data files are the building materials we work with every day, all day. Working with them should be effortless.

Federated Learning

Today, in every aspect of our lives, everything we do leaves a digital footprint. Health-care institutions are also increasingly collecting more data on their patients, thanks to improved IT infrastructure and new sensors that allow to record vital signs at high frequencies. Most hospitals still store and manage this information locally, under strict data privacy and protection regulations. Although data privacy and protection are of the utmost importance, they become an obstacle for the large-scale analysis of such data and the extraction of new insights that could help doctors in the diagnosis and therapeutic process.

Human Reproduction Reloaded

We are proud to be part of the UZH University Research Priority Programs (URPP) where we with the Schwank lab investigate the use of AI 🐱‍💻 and CRISPR-based technologies ✂️ 🧬 for germline editing. Moreover, these efforts aim at providing policy guidance on the societal impact and legal challenges of rapidly changing medical technology on human reproduction. The goals and objectives of the subproject SP4 CRISPR Technology in Human Reproduction we are working on are detailed below (taken from the URPP Human Reproduction Reloaded official website ).

Patient Similarity Analysis

Since the early days of computing, healthcare professionals have dreamt of using the vast storage and processing powers of computers to sift through vast medical archives and automatically discover new facts and medical knowledge locked inside electronic health records (EHRs) and similar digital patient data archives. This is not unlike the activity of physicians that use their clinical experience to identify patterns (such as symptoms) that are common among a group of patients, leading to new disease classifications and eventually treatment strategies.

A Quest for Reproducible Data Science

Have you ever tried to reproduce someone else’s analysis and got different results?

AutoDiscern: Rating the Quality of Online Health Information with Hierarchical Encoder Attention-based Neural Networks

To help patients find high quality health information online, we developed a Deep Learning system that evaluates the quality of online health articles. The system implements the DISCERN criteria, which checks for references, balanced writing, and more.

MRIdle

Helping reduce idle time in the USZ Radiology department

30 days All-Cause Readmission

Comparing neural-networks versus logistic regression for predicting readmission.

autoDISCERN

Assessing the quality of online health information with AI.

Drug-Drug Interactions

Novel computational method for drug-drug interaction predictions which are an important consideration for patient treatment.