van der Schaar Lab

Machine learning for healthcare: Towards a unifying framework

If it were not for the great variability among individuals, medicine might as well be a science and not an art.

Sir William Osler

My purpose is to bring the full power of Machine Learning to bear to change Osler’s mind: to make medicine a science.  This post proposes a framework for this purpose, based on the work I have done and the conversations I have had with clinicians over many years. 

One of the biggest challenges is that clinicians (like you) and machine learners (like me) speak a different language.  I hope that our ongoing conversations (represented by our Revolutionizing Healthcare engagement sessions) will help us to understand each other so that we can work together.

I am particularly grateful to the six clinicians with whom I have started the conversation and who will feature in our upcoming Revolutionizing Healthcare engagement session on November 11. They are Dr. Maxime Cannesson, Dr. David Cox, Dr. Camelia Davtyan, Dr. Brent Ershoff, Dr. David McCartney, and Dr. Henk van Weert

The unifying framework I have in mind has two dimensions: the focus (patient or profession) and the scale (narrow/individual or broad/population-system), leading to four areas, as in the chart below.

Let me expand on each of these areas separately.

Bespoke medicine

Individuals are enormously complex, and differ in many ways – genetic backgrounds, environmental exposures, lifestyles, and treatment histories, to mention just a few. These differences express themselves in different susceptibilities to and manifestations of disease and in variations in response to therapies. Current approaches to personalized/individualized/precision medicine view individuals on the basis of some fixed pattern, defined on the basis of a very limited numbers of factors. 

Bespoke Medicine creates patterns on the basis of all the information available and updates and adapts these patterns as new information becomes available, incorporating the effects of aging, lifestyle changes, onset and evolution of various conditions, as well as progress through a course of treatment. 

Bespoke Medicine will make full use of the power of Machine Learning to create a dynamic and holistic view of the individual in a way that is presently impossible.  In particular, Bespoke Medicine will address many of the central problems of treating patients:

Understanding physiology and symptoms
Using data to discover interactions among factors

“Quantifying” disease
Disease states and durations; early diagnosis

Dynamic phenotyping and forecasting

Discovering Causal pathways

Understanding treatment effects and counterfactual scenarios
Individual variation in efficacy, side effects, interactions etc.

Empowering clinicians

This project aims to make machine learning serve clinicians, not replace them. To this end, the recommendations of machine learning must be made to be interpretable, explainable and trustworthy. 

Systems, pathways, and processes

Machine Learning will improve the way healthcare is delivered on many scales, from individual hospitals to cities to counties to states to countries.

Ideally, every patient should receive the best care – no matter where that care might be delivered – and that care should be delivered in a way that is both effective and cost-efficient. 

Accomplishing this ideal will rely on Machine Learning to integrate data from a vast array of interconnected sources to produce actionable intelligence that will inform all the components of the healthcare system, from the delivery of information and recommendations to providers and patients to the planning and allocation of resources – and everything in-between. 

Population health and public health policy

In 1854, the discovery that cholera was spread through contaminated drinking water required the brilliance and perseverance of John Snow.  Today, given the data, Machine Learning could make the same discovery without human intervention. 

More generally, Machine Learning can discover and disentangle population risks and personalize those risks to various individuals.  It can help to create data-driven guidelines, protocols and standards for screening (who to screen? when to screen? how to screen?  how often to screen?), vaccination (who should be vaccinated and when? how should access to vaccines be prioritized?) and access (who should get the next organ?), etc. 

It can also facilitate learning across locations—even across countries (e.g. lessons from Covid-19). 

We will explain our framework in more depth in our upcoming Revolutionizing Healthcare session, and encourage you to join the discussion on November 11. If you are a practicing clinician, please sign up below.

Mihaela van der Schaar

Mihaela van der Schaar

Mihaela van der Schaar is the John Humphrey Plummer Professor of Machine Learning, Artificial Intelligence and Medicine at the University of Cambridge, a Fellow at The Alan Turing Institute in London, and a Chancellor’s Professor at UCLA.

Mihaela has received numerous awards, including the Oon Prize on Preventative Medicine from the University of Cambridge (2018), a National Science Foundation CAREER Award (2004), 3 IBM Faculty Awards, the IBM Exploratory Stream Analytics Innovation Award, the Philips Make a Difference Award and several best paper awards, including the IEEE Darlington Award.

In 2019, she was identified by National Endowment for Science, Technology and the Arts as the most-cited female AI researcher in the UK. She was also elected as a 2019 “Star in Computer Networking and Communications” by N²Women. Her research expertise span signal and image processing, communication networks, network science, multimedia, game theory, distributed systems, machine learning and AI.

Mihaela’s research focus is on machine learning, AI and operations research for healthcare and medicine.