eprintid: 26839
rev_number: 14
eprint_status: archive
userid: 4359
dir: disk0/00/02/68/39
datestamp: 2019-07-24 15:01:19
lastmod: 2019-07-26 15:09:41
status_changed: 2019-07-24 15:01:19
type: conferenceObject
metadata_visibility: show
creators_name: Ballhausen, Hendrik
creators_name: von Maltitz, Marcel
creators_name: Niyazi, Maximilian
creators_name: Kaul, David
creators_name: Belka, Claus
creators_name: Carle, Georg
title: Secure Multiparty Computation in Clinical Research and Digital Health
subjects: ddc-004
subjects: ddc-020
divisions: i-704000
pres_type: poster
abstract: The free flow of information is the lifeblood of the digital economy. In research, the exchange of data is a prime requisite for the generation of new knowledge. In practice, however, there are many barriers to data sharing. Collaborators are reluctant to reveal their proprietary knowledge, consumers are wary of large scale data collection and profiling, regulation restricts what personal information can and cannot be shared across institutional borders.

In clinical research and digital health, there are particulary strict data protection rules in force. Here, we are motivated to seek new methods for knowledge generation, without the problematic exchange of actual patient data. In fact, there is a technology, secure multiparty computation, which allows a number of collaborators to jointly compute about any function, without revealing their private inputs. The method relies entirely on calculations over an encrypted network, without the need for a trusted third party, a central data repository, or even trust between the collaborators.

In a pilot experiment, we demonstrate joint survival analysis based on two separate data bases at LMU Munich and Charité Berlin. Using secure multiparty computation, we are able to identify confounding factors for the survival of patients with glioblastoma. We obtain the same sensitivity as one would achieve by completely pooling the two data bases, and yet we do not actually need to exchange any patient data to perform the calculation.

Going forward, we hope to assemble a collection of libraries for secure multiparty computation in clinical research and digital health. By providing turn-key solutions to the most often used calculations, we hope to reduce barriers to entry for interested researchers and developers. We also hope to create a scientific network of interested institutions and individuals.
date: 2019
id_scheme: DOI
id_number: 10.11588/heidok.00026839
collection: c-51
ppn_swb: 1666709697
own_urn: urn:nbn:de:bsz:16-heidok-268397
language: eng
bibsort: BALLHAUSENSECUREMULT2019
full_text_status: public
place_of_pub: Heidelberg
pages: 1
event_title: E-Science-Tage 2019: Data to Knowledge
event_location: Heidelberg
event_dates: 27.03. - 29.03.2019
citation:   Ballhausen, Hendrik ; von Maltitz, Marcel ; Niyazi, Maximilian ; Kaul, David ; Belka, Claus ; Carle, Georg  (2019) Secure Multiparty Computation in Clinical Research and Digital Health.  [Conference Item]     
document_url: https://archiv.ub.uni-heidelberg.de/volltextserver/26839/3/26839_escience2019_MPC.pdf