In this episode, I talk with Matt Alderdice, Head of Data Science at Sonrai Analytics, about precision medicine. Sonrai Analytics automates laborious data processes and speeds up new drug and healthcare developments.
- Machine learning for automating time-consuming and tedious analysis of microscopy images.
- Training for machine learning practitioners new to pathology by integrating domain experts with your team.
- Involving stakeholders throughout a project.
- Literature reviews to search for associated publications and potential solutions to avoid overly complicated solutions.
- Validating models with ethnically diverse datasets.
- Analytical validation for differing stains, scanners, and operators.
- Clinical validation on a held out dataset in the same environment as would be in the clinic.
- Identifying relevant metrics from conversations with pathologists, oncologists, nurses, and patients.
- Focus on the problem you’re trying to solve – AI is just a tool.
“It’s improved healthcare. There’s a number of different facets to it. So diagnostic tests can have quite long turnaround time. Machine learning can help improve those turnaround times. There’s some tests that might take 14 days. We can get those down to minutes and hours, they can reduce costs for the health services, and we can get the diagnosis to the patient quicker.”
“That really starts first off by establishing multidisciplinary teams, having pathologists either as part of your company, part of your organization, part of your team, or ensuring that you have access to those experts and utilizing them effectively.”
“How can I get people up to speed with pathology? I get them to do a lab tour… It’s very valuable to machine learning engineers to go in and have a look at how the data is actually generated.”
“From an algorithm point of view, we do a literature review to avoid reinventing the wheel really. I’ve never regretted spending more time researching. I have regretted committing to an implementation too soon. And then, finally, we ask ourselves questions like does the project actually need to have this bleeding edge technique applied to it? Or can it be solved with something relatively simple. And the literature review quite often sheds light on that.”
“We have patient outreach initiatives as part of a lot of our programs where we actually present the concepts that we’re working on, the objectives of our project, and we involve patients or survivors of the disease that we’re working to create solutions for into the process. And each one of those different facets, whether it’s the biostatistician, the nurse, patient, survivor, they all contribute to how we think about what metrics are most important.”
“First and foremost, you have to remember the data that you’re working on isn’t just data. This is data that’s coming from patients, and we’re working to create a solution which is for a patient. And I think that that, when you’re sitting behind a computer or you’re looking down a microscope, it can be easy to forget those things.”