Drive 1

DRIVE.AI

Development and validation platform for data-driven AI innovations in medical technology

Learn more
Drive 2

Project description

The DRIVE project aims to address the challenges of AI development with an innovative software platform that bundles the work steps from the idea to approval in a consolidated development platform. Currently, these work steps are distributed between the clinic, the medical device manufacturer and clinical service providers and are characterized by media disruptions and interface problems. This leads to high costs in development and significant delays in approval. The aim of the DRIVE project is to develop this process together with a leading Bavarian university hospital (Augsburg University Hospital) in medical imaging and image processing and to test the optimization of workflows in practice. The platform is provided via a web application based on the principle of “service-oriented architecture”. The performance of the DRIVE platform will be demonstrated using the case study “Glioblastoma follow-up using MRI”.

PROJECT GOALS

Integrated software platform

Research and development of an integrated software platform to accelerate the work steps from the idea to the approval of AI medical devices.

Reference architecture

Establishment of a reference architecture for AI projects with the aim of transferring innovative AI product ideas and translational research to clinical application more quickly.

DRIVE SOLUTION

Raw data extraction

  • Automatic search of existing data sets according to certain specifications (MRI, histology, ICD / OPS, neurological outcome)
  • Selection and signing of cases by the doctor (Quality Gate 1)
  • Automated compilation and anonymization 
  • Secure upload

 

 

 

Generation of training data (annotation)

  • Data collection and verification of anonymization by data protection officers
  • Technical quality check of the overall data records (Quality Gate 2)
  • Annotation of the datasets by medical experts using a customizable integrated annotation tool
  • Quality assurance of the annotations (cross-check, inter-rater reliability, etc.)

AI development and validation

  • Provision of real-world clinical data for medical  AI validation
  • Seamless integration of clinical experts
  • Provision of data logs and overall evaluation for the product file
  • Automatic download of documentation required for regulatory approval

 

 

 

OUTLOOK

HYPOTHESIS

1

Clinical data

The customer underestimates the effort required to obtain clinical data.
2

Acceptance criteria

Acceptance criteria for datasets and annotations needed to mitigate risk and manage expectations.
3

Experience

Limited experience of the medtech industry with stand-alone testing
Interview quotes

“I am looking forward to a platform that resolves the legal issues, solves the issue of anonymization, takes care of workflow management and integrates tools for the annotation process.”

Director R&DDepartment of Surgical Treatment planning applications

“I would like a service where you only have to register and the data is already annotated. All ethical and legal issues must already be clarified.”

Head of AI developmentMedtech manufacturer

“We need a platform where you can redesign the gap between the contract and the data collection. Clinics should have a step-by-step plan (suitable for each country, each location, ...)”

Data ScientistMedtech company

This project is funded by:
Bayern Innovativ

Drive 3

CONSORTIUM PARTNERS

Universitätsklinikum Augsburg and M3i GmbH industry-in-clinic platform

Drive 4

Prof. Dr. med Ehab Shiban

Director (chief) of the Department of Neurosurgery, UK Augsburg

Drive 5

Univ. Prof. Dr. med. Ludwig Christian Hinske

Director of the Institute for Digital Medicine

Drive 6

Basil Berinyuy

Clinical Research Associate, Clinic for Neurosurgery, University Hospital Augsburg

Drive 7

Dr. Stefan Taing

Managing Director M3i GmbH

Drive 8

Dr. med Simon Weidert

Co-Managing Director M3i & Senior Physician, LMU Klinikum

ikenna

Ikenna Ikeliani

Team Leader MxDB Digital Biobank, M3i GmbH

Drive 9
Contact information

Prof. Dr. med Ehab Shiban

ehab.shiban@uk-augsburg.de

Drive 10
Contact information

Dr. Stefan Taing

st@m3i-muenchen.de