Doctoral Student

Doctoral Student (Ph.D.) in Digital Health: Machine Learning in Medical Care

We are looking for a highly motivated computer science Master’s graduate to fill the project-funded open position on the development and application of machine learning methods in medical care.

Please submit your application documents via email to Bert.Arnrich(at)hpi.de

CASSANDRA – Clinical ASSist AND aleRt Algorithms in visceral surgery

Among the 7.1 million surgical procedures performed annually in Germany, approximately 110,000 operations are related to the four major abdominal organ systems: liver, pancreas, upper gastrointestinal tract (esophagus and stomach) and intestine (small bowl and colon). These rather complex operations are associated with high rates of postoperative complications such as serious bleeding, acute kidney failure, intraabdominal infections and sepsis. In fact, at least one out of four surgical patients experiences at least one severe complication after major abdominal surgery with mortality rates being subsequently as high as 12% according to recent studies. Of particular note, these complications often occur multiple days after surgery - hence, when the patient has already been transferred from an intensive care unit (ICU) to a regular ward with reduced monitoring capacities only. However, early detection of patient deterioration is of key importance to prevent beginning complications from aggravating and finally becoming life-threatening. In septic patients, for instance, every hour of delayed antibiotic therapy induction increases patient mortality by 2%. 

Given the crucial need to improve patient monitoring after surgery, the C linical A ssist AND A lert A lgorithm ( CASSANDRA )-Project aims at developing and evaluating the use of machine learning (ML) algorithms in detecting and predicting postoperative intra-abdominal infections and other complication entities. The computational challenge in this field emerges from combining static preoperative risk parameters (eg. patient age, medical history, etc.) and static intraoperative data (e.g. duration, blood loss, etc.) with dynamic real time parameters on the ICU as well as on regular wards – the latter using a telemetric continuous vital parameter monitoring device. Taken together, the results may pave the way for an autonomous real-time monitoring system on surgical wards in the long term. The CASSANDRA -project will be conducted in close collaboration with the Surgical Department of the Charité - Universitätsmedizin Berlin and is funded by the Innovation Fonds of the Federal Joint Committee (Gemeinsamer Bundesausschuss G-BA) from May 2021 until April 2024.

Responsibilities

Support with the setup of a wireless network on a normal ward in order to collect patient vital signs from wearables in real-time. Develop methods for raw signal data processing. Create a pipeline for automatic signal processing.Build automatic signal quality checks.Clean and filter the signals.Extract condensed information from continuous signals. Develop and evaluate different machine learning models that combine static risk parameters and dynamic continuous vital parameters. Investigate different clinical endpoints like ‘occurrence of a post-surgical complication’ or ‘patient death’.Investigate model performance for different subsets of data representing different times during the inpatient stay.

Qualifications

Master’s degree in computer/data science Experience in data mining, computational statistics and machine learning Excellent skills in high-level programming languages Self-reliance, goal-oriented creativity, dedication, team spirit Excellent English language skills

About the Hasso Plattner Institute 

The Hasso Plattner Institute (HPI) for Digital Engineering gGmbH is an institute for teaching and research founded by Hasso Plattner in 1998 as a unique public-private partnership, partially funded by the Hasso Plattner Foundation. The HPI focusses on the design, construction, maintenance and analysis of large-scale IT systems. In recent years, the HPI has been expanding towards new areas in information technology, among others digital heath.

Since 2018, the new research group Digital Health – Connected Healthcare addresses the collection and analysis of health-related data from daily life and in a clinical context. The interdisciplinary team led by Prof. Dr.-Ing. Bert Arnrich includes computer scientists and data scientists working on projects that address the whole pipeline from data collection, data preparation and analysis through statistical methods and machine learning models, up until the presentation of relevant information to clinical practitioners.

About the project partner Charité 

The Surgical Clinic, Campus Charité Mitte|Campus Virchow Klinikum of the Charité - Universitätsmedizin Berlin is one of the largest surgical departments in Europe with over 5000 cases annually. Conducting more than 170 pancreatic, 350 liver, 500 intestinal and 170 esophageal and gastric surgical procedures per year, it is one of only a few surgical single centres providing a number of operations high enough for the development and evaluation of various machine learning algorithms. The challenging process of systematic data collection and digital storage throughout inpatient stays has already been introduced and continuously been optimized at the Campus Virchow Klinikum as part of the Enhanced Recovery After Surgery (ERAS ©)-program that works an holistic perioperative documentation concept.

---