Gaining momentum: Showcasing the DHDP
Supporting predictive biomarker research for better patient outcomes
Building and refining the Digital Health and Discovery Platform (DHDP) is a multi-step collaborative process. Bringing together health researchers, AI and data scientists, and industry partners is complex but once finalized, the DHDP will deliver a secure and streamlined data ecosystem to accelerate precision medicine research.
As we build and refine the Platform, the initial focus is on oncology and neurology. The DHDP’s data-sharing approach will help to accelerate research to detect, diagnose, and treat this disease for better patient outcomes at health and research centres and sites across the country.
But how will the platform function, and what questions will it answer?
Platform demo at TFRI scientific meeting
The Terry Fox Research Institute’s (TFRI) 10th scientific meeting in May 2024 saw a peek ‘under the hood’, showcasing the Platform’s potential and where it fits in with precision oncology.
In a DHDP Plenary Session, the Platform’s functionality was presented, highlighting ways in which DHDP members would be able to use it. The demonstration showcased how the Platform will enable discovery, querying and exploration of clinical, genomic, and other health-related data, supporting access within security and privacy parameters.
Caption: Feedback following the demonstration to 273 conference attendees reported 89% of respondents could or could possibly see themselves and their teams using DHDP in the future.
Data platform for biomarker research
Attendees also saw how the Platform’s ability to support federated learning could enable the exploration of biomarker discovery to support immunotherapy. Immunotherapy can be a life-saving treatment in oncology, but it is expensive and carries a risk of side effects or lack of therapeutic benefits for a subset of patients. It is therefore crucial that treatment is personalized, with careful selection of immunotherapy appropriate for each patient.
What is immunotherapy for cancer? Immuno-oncology helps the patient’s own immune system fight cancer. By identifying potential weaknesses and tumour cell surface markers, it’s possible to get the immune system to recognize and attack cancer cells, to control and destroy them. These predictive biomarkers help oncologists pick the right treatment.
Precision oncology for tailored success
And this is where the DHDP will be able to excel, by facilitating data sharing and analysis across large cohorts of data that can help pinpoint tumour weaknesses through predictive biomarker discovery. This kind of research takes a lot of data. By enabling access to diverse data sets, the DHDP helps researchers explore large cohorts of clinical and research data, to establish links between specific tumour microenvironments, patient characteristics and successful response to immunotherapy.
Defining these predictive biomarkers helps clinicians personalize immunotherapy, by matching patients with drugs that are likely to work well and cause fewer side effects.
Data security, baked in from the start
During a panel discussion on Artificial Intelligence, patient partner moderators James Pereira and Denis Petitclerc, representatives of Canada’s three national AI institutes (Vector Institute, MILA and Amii) and DHDP technology partner, integrate.ai explored the security and safety being built into the Platform from the start to ensure operational confidence.
DHDP is being developed to use a federated learning approach to training AI and machine learning tools for data exploration, discovery and analysis.
“Federated learning is a way to train AI models without anyone seeing or touching your data, offering a way to unlock information to feed new AI applications. … New AI models are being trained collaboratively … on data that never leave[s] [the] private server. … federated learning [is] becoming the standard for meeting a raft of new regulations for handling and storing private data.” IBM https://research.ibm.com/blog/what-is-federated-learning
This federated learning approach in the DHDP means that data scientists can train their AI analytical models and researchers can explore the data, all without moving data files from their source institution. Much safer, since individual owners can set security parameters, and much easier – no more data siloes! And for pan-Canadian collaboration, this is a game changer.
“Many in the demo audience left feeling positive about the DHDP and its approach to patient data access. In an era where we often hear troubling stories about AI and data breaches, it’s reassuring to see a thoughtful approach. Everyone in the MOHCCN’s Gold Cohort will benefit from the Digital Health and Discovery Platform.”
James Pereira – Patient Partner, Marathon of Hope Cancer Centres Network (MOHCCN)