DBIS

Distributed analytics infrastructures, such as the Personal Health Train (PHT), enable decentralized data processing in compliance with data protection laws. By bringing modeling algorithms to data, this concept enables the processing of sensitive data, such as health care data. The PHT has demonstrated the feasibility and efficiency of this concept. Still, it is relatively inflexible regarding the findability of data to be processed: It simply schedules a train to stop at every selected station.
This thesis aims to design a more efficient and privacy-preserving scheduling strategy that signals the presence of new relevant data, updated data, or other changes to data to train operators.
Thereby, relevant stations can be prioritized or re-evaluated. To this end, the thesis will evaluate to which end such notification mechanisms and improved findability benefit distributed analytics conducted via the PHT infrastructure.