van der Schaar Lab

Revolutionizing Healthcare: AutoPrognosis: Using the next generation of ML tools

24 October 2022, saw the twentieth virtual Revolutionizing Healthcare engagement sessions of the van der Schaar Lab and its audience of practising clinicians. A jubilee that, indeed, had something big on offer: AutoPrognosis 2.0!

This was the first of a double-header on democratising diagnostic and prognostic modelling in healthcare using automated and interpretable machine learning. The second session will take place on 21 November.

At the beginning of the session, Mihaela presented AutoPrognosis 2.0 to an international panel of clinical experts who discussed opportunities and impacts of the tool.

AutoPrognosis 2.0 takes either raw or curated medical datasets, manages missing data, selects the optimal ML pipeline, and provides a report detailing the predictive model, along with explanations of how the model comes to its conclusions. Finally AutoPrognosis 2.0 translates the final predictive model into a web-based interface for clinicians to use.

The session started with said presentation by Mihaela, followed by Dr Eoin McKinney, MD, who gave the clinical perspective on the impact of AutoPrognosis. Then, we had a look at an intriguing case study together with Dr Thomas Callender, MSc MRCP, Wellcome Trust Clinical Training Fellow and Public Health Physician at UCL who has used a version of our tool already.

Finally, we had a round table discussion with experts from the field to explore the opportunities and advances our tool offers for clinicians. Our guests were:

Prof Donald E. Ingber, MD, PhD (Founding Director, Wyss Institute for Biologically Inspired Engineering at Harvard University; Judah Folkman Professor of Vascular Biology, Harvard Medical School & Vascular Biology Program, Boston Children’s Hospital; Hansjörg Wyss Professor of Bioinspired Engineering, Harvard School of Engineering and Applied Sciences)

Prof Suetonia Palmer, MB ChB, PhD, FRACP (Professor of Medicine, University of Otago, New Zealand; Deputy Dean, University of Otago, Christchurch; Renal Physician, Christchurch Hospital, New Zealand)

Dr Paul Goldsmith, MA, BM, BCh, MRCP, PhD (Consultant Neurologist NHS Clinical Senate; Non-executive director of Medicines and Healthcare Products Regulatory Agency (MHRA); Non-executive director of MDU Ltd and MDU Investments Ltd; President, Chief Medical and Innovation Officer, Closed Loop Medicine Ltd; Trustee Radix Big Tent Foundation)

Our panelists answered a number of questions posed by Mihaela and the audience.

We have now published the tool on GitHub and also launched a dedicated website that is full of information surround AutoPrognosis 2.0. This website will grow over the coming weeks with more information, support, demonstrators, feedback, and will be the central point of our journey with this incredible tool.

If you didn’t manage to join us last week, we’d strongly recommend watching the archived video, which is now available on YouTube:

NOTE: This information was up-to-date at the time of the presentation but does not take into account material published since then.

Sign up for our upcoming sessions here.

Andreas Bedorf