Brisbane's Max Kelsen secures data platform trial with Peter MacCallum Cancer Centre

Brisbane-based AI and ML specialist consultancy Max Kelsen has secured a trial for its new data platform with Peter MacCallum Cancer Centre (Peter Mac).

The trial is part of a three-year research project led by Australian research foundation Digital Health Cooperative Research Centre (DHCRC), which aims to help Peter Mac to capture and store its pool of de-identified patient data to support its research and development efforts. Also taking part in the project as research partner is Swinburne University of Technology.

Max Kelsen’s platform would help aggregate previously disparate and siloed data sets, such as patient diagnoses, treatment and post-treatment information, and enable Peter Mac to build digital capability for its health and research services, integrate digital innovations and build a digital ecosystem.

Peter Mac director of digital and healthcare innovations Kate Burbury said the collaboration across academia, healthcare and commercial organisations represents a potentially transformative digital healthcare project.

“The goal of this exciting project is to integrate Peter Mac’s digital data and better understand the benefits that can come from unlocking large health datasets to support collaborative digital health research,” Burbury said.

“We also hope it will assist with establishing a pipeline for the development of regulated, scalable and transferable Software as a Medical Device (SaMD).

“We see significant opportunity for this project to improve clinician decision-making, patient outcomes and drive long term digital health product development and commercialisation.”

Max Kelsen chief executive Nicholas Therkelsen-Terry said the concept of “data driven medicine” has the potential to improve and accelerate patient focused decision making, but before the application of AI, the task has been difficult because of the sheer size and complexity of siloed, clinical data sets and a reluctance of organisations to share them.

“Despite the best intentions of many stakeholders in healthcare and regulation, developing patient-based innovation around data has been costly, time consuming and challenging from a regulatory perspective.

“With this project we are aiming to make it much simpler for researchers to access data which will allow them to build and deploy new innovations in clinical practice under the appropriate regulations – and do this faster and more cost-effectively through the application of game changing AI models.”

Therkelsen-Terry added the project would make “significant improvements” to how healthcare is delivered in Australia and beyond through ethically applied AI and technological innovation in medicine.

“By creating a unique and applied data platform for the consolidation of important data, we can help medical practitioners deliver better and more personalised care,” he said.

Max Kelsen said the platform ensures data security through de-identifying all health records and using a universal consent approach, giving patients control over how their data is shared.

DHCRC chief innovation officer Stefan Harrer said uniting the partners represented a long-term opportunity that has the potential to transform how digital health data is stored and shared.

“This project will initially integrate a novel, scalable data linkage and management platform into Peter Mac's digital infrastructure and then use it to develop a production-grade AI-powered image segmentation module for potential inclusion in clinical workflows,” Harrer said.

“This will set a precedent for empowering clinical institutions to tap into the abundance of health data they hold in highly efficient ways and to deliver real value to patients in trusted and personalised ways.

"Given the unstructured and fragmented nature of health data and its exponentially growing size, AI is poised to play a key role for analysing it. This comes with challenges. In order to gain regulatory approval and the trust of clinician users and patients, AI algorithms need to be fair, transparent and robust.

“Secure data management systems linking different data sources and modalities together efficiently, privately and in an interoperable manner are a key ingredient for developing such responsible AI systems.”