We assist healthcare professionals in complex decision making by combining heterogeneous patient data in practically usable tools.
Supporting healthcare professionals in their decision making requires the gathering and distilling of essential information from many sources such as vital signs, images, laboratory tests, cognitive tests, etc. Combining such heterogeneous data is essential in attempts to create a holistic view of a person's health or 'disease state''. However, this is often difficult in practice, due to the highly variant nature of different data properties, as well as the fact that measurement data are often incomplete in real-life settings. Additionally, data analysis methods thus far typically concentrate on single-disease views only, and do not take into account similarities between different diseases.
In close collaboration with healthcare professionals, we have developed, data mining and analysis methods, which we refer to as Disease State Index and Disease State Fingerprints, in order to provide insight into highly heterogeneous and possibly incomplete data collections and apply them in decision support for diverse diseases, such as dementias, psychotic disorders and traumatic brain injury. The methods provide information to a researcher or clinician that can be related to understandable physiological and medical processes and thus not function as a black box. The methods have been widely disseminated, and have shown their strength e.g. in the EU projects PredictAD, PredictND www.predictnd.eu, TBIcare and METSY www.metsy.eu.