Knowledge is power -- and in healthcare, that holds absolutely true. Yet, for an industry that is under financial stress, increasing complexity of disease and co-morbidity, and
burdened by capacity constraints -- why has data not been healthcare's savvier? Three major challenges have inhibited this:
- data is not accessible and remains in silos;
- data is not analysed to derive meaningful clinical insights;
- insight isn't accessible for actioning by providers or patients to self/joint manage their condition.
Our consortium of medical professionals, data scientists, IT-infrastructure experts, machine learning researchers and legal experts have designed Enabling Patient Interventions
to liberate, analyse, and action that data in a trustworthy way. EPI aims to empower patients and providers through self-management, shared management, and personalization across
the full health spectrum. To do so, we will build a fuller picture of the person by linking traditional eHealth data sets with new sources of data. Further, we will develop a
platform based upon a secure and trustworthy distributed data infrastructure, combining data analytics, including machine learning, and health decision support algorithms to
create new, actionable, and personalized insights for prevention, management, and intervention to providers and patients. We will develop new machine learning methods for
determining and analysing optimal interventions within small patient groups.
Our insights will be applied in healthcare use cases representing a spectrum of health management challenges ranging from common chronic to highly lethal orphan diseases, and
will empower better self/joint management of these conditions to improve cost, quality, and outcomes of care.
The overall aim of this project is to explore the use and effectiveness of data driven development of scientific algorithms, supporting personalized self- and joint management
during medical interventions / treatments. The key objective is to use data science promoting health practically with data from various sources to formulate lifestyle advice,
prevention, diagnostics, and treatment tailored to the individual, and to provide personalized, effective, real-time feedback via a concept referred in this proposal as a digital
health twin. The project addresses six research questions:
- Dynamically Analyzing Interventions based on Small Groups: how can we determine, based on as little data as possible, whether an intervention does or does not work
for a small group or even an individual patient?
Dynamically Personalizing the Group: how can we identify effective intervention strategies and optimize personalization strategies applicable for different patient
and lifestyle profiles via dynamic (on-line) clustering of patients? Can those clusters be adapted as new data about patients and results of interventions come in and
as other data may be removed or modified?
Data and Algorithm Distribution: what are the consequences of a distributed, multi-platform, multi-domain, multi-data-source big data infrastructure on the machine
learning algorithms and what are potential consequences on performance?
- Lead CWI: Rosanne Turner, Peter Grunwald
Adaptive health diagnosis leading to optimized intervention: how can we enhance self- / joint management by dynamically integrating updated models generated from
machine learning from various data sources in state of the art health support systems that based on personal health records, knowledge of health modes and effective
- Lead: VU, Corinne Allaart, Henri Bal
Regulatory constraints and data governance: how can we create scalable solutions that meet legal requirements and consent or medical necessity-based access to data
for allowed data processing and preventing breaches of these rules by embedded compliance, providing evidence trails and transparency, thus building trust in a
sensitive big data sharing infrastructure?
- Lead: UvA, Saba Amiri, Adam Belloum
Infrastructure: how can the various requirements from the use-cases be implemented using a single functional ICT-infrastructure architecture?
- Lead: UvA, Milen Girma Kebede, Giovanni Sileno, Tom van Engers
- Lead: UvA, Jamila Kassem, Paola Grosso
- Principle Investigators: Prof.dr.ir. C.T.A.M. de Laat, <firstname.lastname@example.org>, prof. dr. Sander Klous <Klous.Sander@kpmg.nl>
- Project contact: Josine Janus <Janus.Josine@kpmg.nl> or +31 20 656 4357
- This work is part of the project Enabling Personalized Interventions (EPI) and is supported by NWO in the Commit2Data – Data2Person
program under contract 628.011.028.
For more information see: https://enablingpersonalizedinterventions.nl
- Leon Gommans, John Vollbrecht, Betty Gommans - de Bruijn, Cees de Laat, "The Service Provider Group Framework; A framework for arranging trust and power to facilitate
authorization of network services.", Future Generation Computer Systems, (Accepted paper), June 2014
- Leon Gommans, "Multi-Domain Authorization for e-Infrastructures", UvA, Dec 2014.
- Internet2 2012 session: "Trust Framework for Multi-Domain Authorization".
- speakers: Leon Gommans , John Vollbrecht, chair: Cees de Laat.
- Managing Our Hub Economy, Marco Iansiti, Karim R. Lakhani, Harvard Business review, September-October 2017
issue, [local copy]
- NWO press release: Enabling Personalized Interventions - EPI.