IMPACTTHE MAGAZINE OF THE AUSTRALIAN ACADEMY OF TECHNOLOGY AND ENGINEERING ATSE.ORG.AU NUMBER 209 | SEPTEMBER 2019 A picture of health How health technology is transforming our lives Contents Annual Academy Oration 16 Cover photo of Dr Erica Smyth & New Fellows Welcome AO FTSE by Frances Andrijich IMPACT is the biannual magazine of the Australian Friday 29 November 2019 Academy of Technology and Engineering. PUBLISHER Australian Academy of Technology and Engineering ADDRESS Level 6, 436 St Kilda Road 24 Melbourne VIC 3004 20 POSTAL ADDRESS JOIN US GPO Box 4055 FEATURES Melbourne VIC 3001 as we welcome more than two dozen 16 Making high quality health-care leaders from across industry, government TELEPHONE accessible to all and academia as Fellows of the Academy. +61 3 9864 0900 By Kiran Mazumdar-Shaw EMAIL The event includes the annual Oration, [email protected] 19 3D vision of cancer research delivered by a Fellow of global standing. CEO Dr Margaret Hartley FTSE 20 Diabetes progress – it’s finger-pricking good EDITOR By Erica Smyth Benjamin Hickey 28 DESIGN 23 Body clock linked to a healthy ticker Elizabeth Geddes By Benjamin Hickey REGULARS ISSN 1326-8708 (print) 24 Skin in the game 4 From the President’s Desk 2207-8223 (electronic) Interview with Tony Weiss 6 Policy PRINT POST Publication number 28 Improving the lives of 12 Academy News 100007367 Australians with dementia COPYRIGHT By Greg Tegart and Anne Livingstone 36 STEM Education © Australian Academy of Technology and Engineering 31 Why I mentor 38 Women in STEM By Academy Fellows DOWNLOAD AS A PDF 46 Technology atse.org.au 34 Precision medicine 51 Engineering By Alan Trounson Pullman on the Park 56 Fellows 40 The Accidental Engineer 192 Wellington Parade Interview with Maria Skyllas-Kazacos 63 Obituaries Melbourne VIC 3002 70 What we’re reading The Academy acknowledges 46 Apollo 11 carried my scientific [email protected] the Traditional Owners of the land experiments to the moon, on which we meet and work. 73 Spotlight We pay our respects to Elders, but so what? past, present and emerging. By Brian O’Brien 52 Safety by design By Mark Hoffman IMPACT SEP—19 3 FROM THE PRESIDENT’S DESK The cost of monitoring devices (eg, single Moving on, assume that we have EHRs that are lead ECGs) has come down dramatically populated with a complete and accurate set By Hugh Bradlow while at the same time their accuracy has of endogenous, genomic and exogenous data. been enhanced. Our ability to collect and Goodness knows how long it will take us to analyse data using new machine learning get there but what could we achieve? The short answer is a transformation of healthcare. algorithms has improved out of sight. Consider the use of machine learning to predict However, “the fault, dear Brutus, lies not in our adverse health events before they occur. In Technology & data stars” but in our data. Machine learning has the hype around machine learning it is usually two preconditions, namely that you collect all forgotten that it is just pattern recognition. If relevant data and that the data is accurate. In you train a neural network with enough patterns, terms of collecting relevant data, work by IBM eventually it will recognise variants of those has indicated that endogenous data (the data patterns with a high probability. So if you are in healthcare arising inside the medical system – pathology able to feed data about an individual (their reports, doctors’ notes, imaging, etc) will influence ECG, their food consumption, exercise, travel, only 10 per cent of the outcomes. Even if we add etc) continuously into a big data environment In theory the use of technology, genomics data to the endogenous data, the it is conceivable that patterns will emerge impact on outcomes rises only by 30 per cent. that enable the pre-conditions for events to specifically the Internet of be recognised before the actual event (eg, a The remaining 60 per cent of outcomes are heart attack) occurs. Things (IoT), should transform influenced by exogenous data – the information healthcare: it should enable about our lifestyles which is not collected by the My favourite scenario is that you are happily medical system. Examples of such exogenous going about your business when an ambulance continuous (instead of data are many but include diet, exercise, appears and carts you off to hospital even though episodic) monitoring of chronic location and travel, human interactions and you are protesting that you feel fine. Seems pretty other considerations. invasive doesn’t it? – but it beats actually having conditions, prediction of adverse a heart attack because if not treated within events, and improved diagnosis. Even if we could determine which data we must 12 hours the heart damage is irreversible. collect, our ability to assemble that data into a In practice however, these format accessible to physicians and researchers There is another consideration in having access to benefits are yet to materialise is constrained by many factors. While the country complete and accurate data, namely healthcare is moving glacially towards a consolidated provider performance. It is well known that the (with some notable exceptions Electronic Health Record (EHR), as individuals medical system regards healthcare as a free in the imaging space). we collect very little of the exogenous data. For market and doctors are not constrained from example, how many people keep track of their charging anything the market will bear, with the daily food intake? result that charges vary by an order of magnitude Why is that? depending on the provider. In a normal free Furthermore, even for the few that do collect such market the consumer can also make a judgement data, the plethora of systems and formats used on the value of the product being provided but to do so do not encourage sharing. Finally, privacy in medicine no such data is available to the considerations quite rightly create significant patient. Even if data is available (eg, success rate (but not insuperable) barriers to data collection of procedures performed) the providers have and sharing. As an example, consider the plausible ways of rationalising their differences discussion about the change in approach to in outcomes – the patient was more difficult, the Australia’s My Health Record from “opt in” to “opt hospital nursing staff was not up to scratch, key out”. Before “opt out” the EHR was barely used. personnel were not available that day, etc. Returning to the 10 per cent of outcomes that Now consider what we could do if we had could be influenced by comprehensive EHRs, accurate data on all the relevant variables: assuming the collection of endogenous data patient data (medical history, activity history, is complete, accuracy is still an issue. Clinical genome, etc); practitioner data (procedure Decision Support (CDS) systems, often based on outcomes, qualifications, professional history, Artificial Intelligence, should be able to assist activity history, food history, etc); data about doctors in avoiding medical errors, determining other staff involved in the treatment (nurses, the optimum treatment for a specific patient’s anaesthetists, etc); hospital data (history, staffing condition, and identifying the critical pathology levels, etc); environmental data (weather, traffic, tests that should be conducted. However, they etc) and so on. are dependent on the data supplied to them Conceivably it would be possible to pick a pattern via the electronic health record system. in an individual healthcare provider’s performance Professor Hugh Bradlow – effectively define a “figure of merit” for the Research has shown that this data has a FTSE ability of that doctor – which would then enable “half-life” of a mere four months due to rapidly Hugh Bradlow is the President patients to determine whether it was worth changing treatment regimes and is often wildly of the Australian Academy of paying extra for that person. I have no idea what Technology and Engineering. inaccurate due to poor record-keeping by a complete data set would like (I have suggested You can hear more of his thoughts physicians. As a result, doctors either ignore or some of the parameters above) nor whether there on Big Data and healthcare override the CDS systems, making them of little in a podcast available on the is an actual pattern that could be picked out for value. This is not a technology issue – it is a human Academy’s website. a given practitioner but I do know that this is an behaviour issue making it much harder, but not active area of research and we should find out impossible, to fix. However, until the health system in the next decade. atse.org.au recognises that the EHR is just the beginning and that they need to tackle the quality, as well as In summary, technology and data have a huge completeness of the data that is used to populate role to play in healthcare. However, we are just at the EHR, progress will be slow. The USA is starting the very beginning of a long and complex journey to tackle these types of problem with their shift which will be more governed by human behaviour to outcomes-based health payments. issues than technology. 4 IMPACT SEP—19 IMPACT SEP—19 5 TRANSPORT REPORT TRANSPORT POLICY POLICY By Kathryn Fagg and Drew Clarke Car Shifting 82% gears Preparing for a transport revolution The first petrol engine car New technologies are emerging in the areas For example, inadequate planning, population the uptake of low and zero emissions vehicles Shifting Gears — Preparing for a of digital and data, communications, sensing growth and the spread of urban centres could and their increased use in government and Transport Revolution was published in Australia hit the streets and spatial, and energy.