Analytics in Healthcare

The research group Analytics in Healthcare investigates how data-driven methods can support better decision-making in healthcare. Our work focuses on three areas: predictive analytics, prescriptive analytics, and evaluation research. Across all areas, our goal is to improve patient outcomes while contributing to a more efficient use of scarce healthcare resources.

Predictive Analytics

We develop predictive machine learning models to identify patients at risk and support earlier, more targeted interventions. Our research includes applications such as predicting health deterioration in chronic care patients, identifying underreported diseases in administrative data, and forecasting complications or hospitalizations from remotely collected vital sign data.

By enabling healthcare professionals to focus attention on high-risk patients, predictive analytics can help prevent complications, avoid unplanned hospitalizations, and improve quality of care.

Selected projects

• OptiMaL: Optimizing Critical Patient Pathways with Predictive Machine Learning and Continuous Wearable Data Collection
• Identification of unlabeled sepsis cases to increase the quality of hospital administrative data and downstream prediction tasks and research
• Entity embedding of high-dimensional claims data for hospitalized exacerbation prediction

Prescriptive Analytics

Beyond prediction, we investigate how data can inform optimal treatment decisions for individual patients. Using retrospective clinical and administrative data, we employ causal machine learning methods to learn which decisions lead to the best outcomes for specific patient profiles.

Clinical decisions strongly affect hospital operations and patient flow, particularly in resource-constrained environments such as intensive care units. Our work therefore lies at the intersection of medicine, operations management, econometrics, and data science.

Selected projects

• POM-D-H: Where Precision Medicine and Operations Management Meet – Learning Optimal Decisions in Hospitals (SNSF-funded)
• Developing policies to manage hospital process bottlenecks – a machine learning approach
• Optimal policies for selecting knee replacement patients for rapid recovery post-surgery

Evaluation Research

We also evaluate the effectiveness of new technology such as digital health interventions or robotic-assisted surgery, and health policies. Our research examines both patient and economic outcomes to assess whether innovations and reforms achieve their intended impact.

Selected projects

• Association of hospital digitalization, profitability, and quality
• Evaluation of rapid recovery post-surgery for knee replacement patients
• Effectiveness of the certification of German cancer centers
• Effectiveness of Swiss Hospital Capacity Planning

The research group is led by Dr. Justus Vogel. Our work relies on strong collaborations with hospitals, healthcare providers, and health insurers. We welcome opportunities for collaboration and exchange.

Contact: justus.vogel@unisg.ch