Innovative Statistics for Digital Medicine: New Analysis Tools from Salzburg
The digitization of medicine creates new opportunities in the areas of prevention, diagnosis, and therapy. However, it also poses significant challenges for researchers, as data from wearables, health apps, and mobile sensors is often highly complex and incomplete and varies from person to person.
Anna Eleonora Carrozzo, a researcher at Salzburg Research, is developing new statistical methods specifically tailored to this digital health data.
Statistics for complex health data
Digital technologies such as wearable devices and health apps provide continuous data on heart rate, movement, blood pressure, and sleep patterns. This information has great potential to enable more personalized, patient-centered care, especially for cardiovascular diseases.
However, it is precisely this diversity and irregularity of the data that makes analysis difficult. “High data complexity, outliers, missing values, or low case numbers—all of these factors make statistical evaluations more challenging,” explains Carrozzo. Her goal is to develop statistical methods that can draw valid and clinically relevant conclusions even when case numbers are small, variability is high, or values are erroneous.
These new methods are particularly useful when working with small samples. For instance, they can be used in the study of rare diseases, where traditional statistics often reach their limits due to the small number of cases.
New statistics for new health data
Carrozzo’s research focuses on nonparametric methods, which do not require strict assumptions about the distribution of data. These methods are particularly well-suited to digital health data, which is often incomplete, irregular, or highly individualized.
These innovative approaches aim to close methodological gaps and support medical professionals in evaluating digital health measures in an informed manner, as well as using them in a clinically meaningful way.
From research to practice
The new methods are being implemented as software packages in R with a particular focus on practical applicability. These packages can be used in research and, in the future, in clinical practice. This facilitates the mapping of complex study designs, such as those involving multiple groups, numerous measurement points, or high-dimensional data. This allows well-founded decisions to be made even when traditional statistical methods are limited by sparse data.
The research project, ‘Evaluating Digital Health Interventions with Complex Designs,’ is funded by the Elise Richter Program of the Austrian Science Fund (FWF). Carrozzo is also involved in the EXDIGIT program, which is a collaboration between Salzburg Research and Paris Lodron University of Salzburg. It is funded by the State of Salzburg as part of the WISS 2030 initiative.
More information:
