Back to search

BEHANDLING-God og treffsikker diagnostikk, behandling og rehabilitering

Improved personalized medicine through machine learning in mental disorders (IMPLEMENT)

Awarded: NOK 2.9 mill.

Psychotic disorders are severe mental illnesses with early onset, frequently chronic course and often lifelong impairment. As a consequence, they cause an enormous healthcare burden, costing close to 100 billion euro annually in Europe alone. The biology of these illnesses is insufficiently understood and no objective tools exist to aid in diagnosis or treatment selection. This sometimes leads to periods of inadequate and ineffective treatment, limiting the opportunity for achieving more optimal clinical outcomes. To address this, IMPLEMENT (I) has successfully developed and applied new machine learning approaches for identification of schizophrenia-associated biological profiles that are being used for subgroup identification, (II) optimized a multimodal treatment response prediction model, achieving an accuracy of 94%, (III) performed extensive work on the preclinical, mechanistic characterization of cognitive processes relevant to schizophrenia, as well as associated methodological development, and (IV) developed an ICT solution for the integrated processing and management of multi-omics data with specific focus on genome-wide association and expression data. With this, IMPLEMENT aims to provide the basis for personalized medicine approaches in schizophrenia, for which currently no robust clinical stratification tools exist.

Work performed as part of the IMPLEMENT project had a strong impact on several positive developments. First, it contributed to mechanisms informed artificial intelligence (AI) to become an important scientific focus in the upcoming, structurally-funded German Center for Mental Health (GCMH), in which IMPLEMENT partner sites CIMH and LMU are coordinating partners. Furthermore, it contributed to the implementation of a national infrastructure with focus on biobanking, omics, and bioinformatics within the GCMH, which is coordinated by E.Schwarz (CIMH). This infrastructure will serve as a knowledge hub to disseminate AI approaches, such as those developed in IMPLEMENT, across the GCMH and beyond. Similarly, machine learning approaches, such as the multimodal stacked generalization and sequential prognostic approach developed in IMPLEMENT (LMU) will likely form a core part of the personalized medicine initiatives pursued in the network. Furthermore, experience in collaborative research and data management made within IMPLEMENT, led to the development of a novel ICT solution by E.Schwarz (CIMH), the foldercase research management platform (www.foldercase.com). It is a freely available scientific project, collaboration and communication management tool, which is already used by approximately 860 scientists, as well as national and international research networks. We expect the platform to form a major technological pillar in the new German Center for Mental Health, fostering collaborative research, data FAIRization, and the development of data science solution. The platform has already been central for the ongoing applications of several large-scale research networks (BMBF-funded SFB, “Sonderforschungsbereich”). IMPLEMENT further played an important role for the award of the annual Chica and Heinz Schaller Research Award (100,000€), and the call to a full professorial position (Hector Institute for Artificial Intelligence in Psychiatry), both to E.Schwarz. IMPLEMENT further led to the initiation of several new research collaborations, including with the Douglas Mental Health University Institute, Dr. Herrmann, University of Heidelberg; Prof. Cecil, Erasmus MC, Prof. Esther Walton, University of Bath, and The H2020 project Early Cause, as well as with members of the FP7 project HELIX. IMPLEMENT results further contributed to a number of successful grant applications focused on the development and application of machine learning technology in psychiatry.

Psychotic disorders are severe mental illnesses with early onset, frequently chronic course and often lifelong impairment. As a consequence, they cause an enormous healthcare burden, costing close to €100 billion annually in Europe alone. The biology of these illnesses is insufficiently understood and no objective tools exist to aid in diagnosis or treatment selection. This leads to long periods of inadequate and ineffective treatment, significantly limiting the opportunity for achieving more optimal clinical outcomes. To address this, IMPLEMENT will develop a translational research framework that identifies biomarkers for treatment-relevant stratification of the most severe psychotic disorder, schizophrenia. Building on known candidates, IMPLEMENT will use advanced machine learning on high-dimensional multi-OMICS and brain scans to identify illness-associated profiles indexing patient subgroups. Using big data approaches (n > 60,000), IMPLEMENT will explore the impact of genetic risk and neurodevelopmental processes on the formation of biological subgroups and use clinical studies of conventional antipsychotic treatment and innovative treatment approaches to tune subgroup profiles towards clinical utility. The IMPLEMENT framework will incorporate preclinical validation to leverage neurobiological understanding and optimize biological subgroup profiles. The clinical utility of these profiles will be validated in independent clinical samples and prospectively recruited subjects. IMPLEMENT will integrate these efforts with ICT development, to optimize the use of high-dimensional datasets across diverse repositories, to optimally harmonize data for personalized medicine investigations and safeguard patient privacy. Overall, IMPLEMENT will provide the basis for biologically-informed personalized medicine approaches in schizophrenia, addressing an enormous unmet medical need in an area of medicine in which currently no robust clinical stratification tools exist.

Publications from Cristin

No publications found

No publications found

No publications found

No publications found

Funding scheme:

BEHANDLING-God og treffsikker diagnostikk, behandling og rehabilitering