Artificial Intelligence
Vanderbilt University Medical Center will leverage GE technology to improve its use of immunotherapy data for cancer treatment.
The company will be highlighting its new Edison platform, AI algorithms built into medical devices as well as diagnostics and therapeutics for precision medicine.
“I think the biggest trend (in healthcare) is towards greater integration. Traditionally, healthcare has been very fragmented, where many different groups serve specific clinical needs without necessarily coordinating with each other. But going forward, the trend is towards integration – not just of things like databases and systems, but integration of the way we process the data and how this influences the clinical workflow,” said Prof Ngiam Kee Yuan, Group Chief Technology Officer of National University Health System (NUHS) in Singapore, in response to what he thought would be the key trends that will impact healthcare systems in future.
It was with the same motivation and mission to best use the healthcare data for research at NUHS that led to the building and development of the DISCOVERY AI platform, which started about four years ago and the platform was officially announced in July 2018. The platform is what Prof Ngiam describes as a ‘sandbox’ that allows the staff at NUHS to develop AI tools in a safe and equitable way – the platform is scalable and can be applied to more than one system within the organisation.
“We saw the opportunity because we had datasets which were large enough to support the development of these AI tools. And one of the advantages we have at NUHS is that we have clinicians and allied health professionals who understand, and are willing to develop these tools. I cannot emphasise enough the importance of having the clinicians onboard throughout the development process.”
Currently, a randomised control trial of a system as part of the platform is a free-text diagnosis machine at the Accident & Emergency (A & E) department. When doctors input a certain set of findings as part of clinical documentation, the machine would automatically provide a suggestion for a diagnosis. The team is exploring diagnosing appendicitis for a start. The trial is slightly under halfway through and Prof Ngiam hopes by the later part of 2019, they would be able to have results, which would be the basis for them to operationalise the AI tool.
One of the early milestones for the NUHS DISCOVERY AI platform is its ability to sustain multiple proof-of-concept projects. With the platform, individual projects are no longer fragmented and there is the ability to aggregate, link and share large data sets.
“It took us four years to get to this point and the next milestone for us is to finish our trials and to actually launch them as “software as medical devices”. Again there are some hurdles to get through before we get there but seeing where we are right now, it is very likely we should be able to get through them.”
Prof Ngiam also pointed out that as the platform is unlike anything they had before and behaves like an advanced form of clinical decision support system (CDSS), which are not based on a set of rules but based on a set of complex trained weights and multiple factors that affect a certain outcome. Despite its complexity, the AI tools need to be thoroughly trialed before it can be used in routine clinical practice.
“This is why we are running it as a trial now in the hospital. In essence, the platform is run under the ambit of a trial to mature the workflow and test the system under real world conditions. Operationally, what the doctors are doing during the trial is no different from what they would have done, except that we collect the data on the basis of the trial,” he added.
Looking to the future, Prof Ngiam and his team is working towards completing the trials in 2019 and hopefully heading towards registration in terms of the platform being used as a medical advisory device.
Data Warehousing
Artificial intelligence can help support and guide care providers' decision-making processes in an era of increasing volume and complexity of data in clinical situations, says Dimitri Fane, director of product management of TrakCare at InterSystems.
Among his pre-show predictions? It's finally telehealth's time to shine. "It's the overnight success story that was 30 years in the making."
Sophie Pinkard discusses her journey in healthcare IT and where the industry is heading in the future.
Wherever you go in Europe, AI is already there. In November 2018, the German government announced its national AI strategy, a draft of which had been published in the summer of 2018 already. Now the strategy has a price tag: €3bn is about to be invested by the German government over the course of six years, the first €500m of which will flow in 2019.
Germany was comparably late. In March 2018, French president Emmanuel Macron announced that his government would invest €1.5bn into AI by 2022. March 2018 also saw the publication of the White Paper ‘Artificial Intelligence at the service of the citizen’ by the Italian government’s Digital Agency. In April 2018, the UK came out with its ‘AI sector deal, worth £1bn, including £300m of private sector investment. And in May last year, Sweden released what they called their National Approach for Artificial Intelligence.
On a European level, the European Commission has published its ‘Communication on Artificial Intelligence for Europe’ in April 2018, to be debated in the European Parliament in due course. The European Commission will also increase its investments into AI under the research and development framework programme Horizon 2020 to around €1.5bn by the end of 2020.
AI: Another word for digitisation?
So what is going on in the old world? Interestingly, many European governments don’t really define what they consider to be ‘AI’. “Many topics that are called AI now were called digitisation before”, said a group leader in the German Federal Ministry for Economic Affairs and Energy recently, adding that he did not want to be quoted with this sentence directly.
Learn more about why AI is more than just a political buzzword in Europe in the recently repositioned HIMSS Insights eBook, a bi-monthly series featuring global examples of projects aiming to foster the development of a digital health ecosystem.
You can read the article in the second eBook which is focused on AI in full here.
Twitter: @DillanYogendra1
Contact the Editor: dyogendra@himss.org
Two major IT players have teamed up to deploy deep-learning AI to cut the time between medical imaging, diagnosis and beginning treatment.
The project, a joint collaboration between Intel and GE Health, is promising to offer physicians automated diagnostic alerts for some conditions within seconds of medical imaging being completed.
It leverages the Intel Distribution of OpenVINO toolkit, running on Intel processor-based X-ray systems to help prioritise and streamline patient care.
Using this system, X-ray technologists, critical care teams and radiologists will be immediately notified to review critical findings that may accelerate patient diagnosis.
Intel Internet of Things Group Health and Life Sciences Sector General Manager David Ryan told HITNA that the AI imaging models are optimised for inference and deployment using the model optimiser component of OpenVINO.
The optimised models are then integrated into the GE application with the OpenVINO inference engine APIs. As X-ray images are acquired by the machine, the inference engine runs them for clinical diagnosis.
GE Healthcare Senior Vice-President of Edison Portfolio Strategy Keith Bigelow said medical imaging is the largest and fastest-growing data source in the healthcare industry.
But, even though it accounts for 90 per cent of all healthcare data, more than 97 per cent of it goes unanalysed or unused.
“Before now, processing this massive volume of medical imaging data could lead to longer turnaround times from image acquisition to diagnosis to care. Meanwhile, patients’ health could decline while they wait for diagnosis,” he said.
“Especially when it comes to critical conditions, rapid analysis and escalation is essential to accelerate treatment.”
According to Bigelow, a key implementation of this technology is providing earlier detection of a potentially life-threatening event – a collapsed lung, also known as pneumothorax.
He said radiologists can now deploy optimised predictive algorithms that scan for and detect pneumothorax “within seconds at the point of care”, allowing rapid response and reprioritisation of an X-ray for clinical diagnosis.
“Deploying deep learning solutions on existing infrastructure delivers the potential to power more efficient and effective care, enhance decision-making, and drive greater value for patients and providers,” he said.
[Read more: AI and machine learning – how soon will it be key to a learning health system? | AI algorithms show promise for colonoscopy screenings]
Ryan said deep learning was a promising approach for radiology because its models can be trained to recognise desired features in an image, such as tumors or anatomies.
“Furthermore, training is done by giving numerous labeled example images to the models, without having to specify the exact features to look for. Deep learning can identify details that can be missed by the human eye,” he said.
According to Ryan, in future applications, deep learning models can be used to identify incidental findings, as well as help radiologists manage their workload, enhance quality of scans, and reduce ‘retakes’, which can cause unnecessary exposure to radiation.
“Deep learning is also showing promising results in image reconstruction from the imaging modalities. Future applications of deep learning can extend beyond imaging data to include electronic health records, pathology, cellular microscopy data, etc. to help develop targeted drugs and achieve precision in medicine,” Ryan said.
Ryan said deep learning was a promising approach for radiology because its models can be trained to recognise desired features in an image, such as tumors or anatomies.
“Furthermore, training is done by giving numerous labeled example images to the models, without having to specify the exact features to look for. Deep learning can identify details that can be missed by the human eye,” he said.
"For the more than 12,000 Australians diagnosed with lung cancer each year, this means a higher chance of survival.”
According to Ryan, in future applications, deep learning models can be used to identify incidental findings, as well as help radiologists manage their workload, enhance quality of scans, and reduce ‘retakes’, which can cause unnecessary exposure to radiation.
“Deep learning is also showing promising results in image reconstruction from the imaging modalities. Future applications of deep learning can extend beyond imaging data to include electronic health records, pathology, cellular microscopy data, etc. to help develop targeted drugs and achieve precision in medicine,” Ryan said.
The project, a joint collaboration between Intel and GE Health, is promising to offer physicians automated diagnostic alerts for some conditions within seconds of medical imaging being completed.
It leverages the Intel Distribution of OpenVINO toolkit, running on Intel processor-based X-ray systems to help prioritise and streamline patient care.
Using this system, X-ray technologists, critical care teams and radiologists will be immediately notified to review critical findings that may accelerate patient diagnosis.
Intel Internet of Things Group Health and Life Sciences Sector General Manager David Ryan explained that the AI imaging models are optimised for inference and deployment using the model optimiser component of OpenVINO.
The optimised models are then integrated into the GE application with the OpenVINO inference engine APIs. As X-ray images are acquired by the machine, the inference engine runs them for clinical diagnosis.
GE Healthcare Senior Vice-President of Edison Portfolio Strategy Keith Bigelow said medical imaging is the largest and fastest-growing data source in the healthcare industry.
But, even though it accounts for 90 per cent of all healthcare data, more than 97 per cent of it goes unanalysed or unused.
“Before now, processing this massive volume of medical imaging data could lead to longer turnaround times from image acquisition to diagnosis to care. Meanwhile, patients’ health could decline while they wait for diagnosis,” he said.
“Especially when it comes to critical conditions, rapid analysis and escalation is essential to accelerate treatment.”
According to Bigelow, a key implementation of this technology is providing earlier detection of a potentially life-threatening event – a collapsed lung, also known as pneumothorax.
He said radiologists can now deploy optimised predictive algorithms that scan for and detect pneumothorax “within seconds at the point of care”, allowing rapid response and reprioritisation of an X-ray for clinical diagnosis.
“Deploying deep learning solutions on existing infrastructure delivers the potential to power more efficient and effective care, enhance decision-making, and drive greater value for patients and providers,” he said.
"For the more than 12,000 Australians diagnosed with lung cancer each year, this means a higher chance of survival.”
Ryan said deep learning was a promising approach for radiology because its models can be trained to recognise desired features in an image, such as tumors or anatomies.
“Furthermore, training is done by giving numerous labeled example images to the models, without having to specify the exact features to look for. Deep learning can identify details that can be missed by the human eye,” he said.
According to Ryan, in future applications, deep learning models can be used to identify incidental findings, as well as help radiologists manage their workload, enhance quality of scans, and reduce ‘retakes’, which can cause unnecessary exposure to radiation.
“Deep learning is also showing promising results in image reconstruction from the imaging modalities. Future applications of deep learning can extend beyond imaging data to include electronic health records, pathology, cellular microscopy data, etc. to help develop targeted drugs and achieve precision in medicine,” Ryan added.
Ryan also said deep learning was a promising approach for radiology because its models can be trained to recognise desired features in an image, such as tumours or anatomies.
“Furthermore, training is done by giving numerous labeled example images to the models, without having to specify the exact features to look for. Deep learning can identify details that can be missed by the human eye,” he said.
According to Ryan, in future applications, deep learning models can be used to identify incidental findings, as well as help radiologists manage their workload, enhance quality of scans, and reduce ‘retakes’, which can cause unnecessary exposure to radiation.
“Deep learning is also showing promising results in image reconstruction from the imaging modalities. Future applications of deep learning can extend beyond imaging data to include electronic health records, pathology, cellular microscopy data, etc. to help develop targeted drugs and achieve precision in medicine,” Ryan said.
This article first appeared on Healthcare IT News Australia.
The European Commission, EU member states, Norway and Switzerland have developed a plan to boost collaboration and further the development and use of AI in Europe, prioritising areas of public interest, including healthcare and transport.
Initiatives will tackle the "low and fragmented" levels of investment in AI in the EU, compared to the US and China, according to the Commission, and the new plan focuses on four areas: fostering talent, making data more available, ensuring trust and increasing investment.
"We agreed to work together to pool data – the raw material for AI – in sectors such as healthcare to improve cancer diagnosis and treatment. We will coordinate investments: our aim is to reach at least €20 billion of private and public investments by the end of 2020. This is essential for growth and jobs," said Vice President for the Digital Single Market Andrus Ansip.
The plan provides a “strategic framework for national AI strategies”, according to information released this month. At this stage, five member states have adopted a national AI strategy, backed by a dedicated budget: the UK, France, Finland, Sweden and Germany.
Through Horizon 2020, working with member states, the Commission will support the development of a common (anonymised) database of health images based on patients voluntarily donating their data, initially targeting the most common forms of cancer to help improve diagnosis by using AI.
Meanwhile, common European "data spaces" will be created to enable seamless data sharing across borders, and advanced degrees in AI will be supported through initiatives such as offering dedicated scholarships.
“These common European data spaces will aggregate data, both for the public sector and for business-to-business needs, across Europe and make it available to train AI on a scale that will enable the development of new products and services.
“The rapid development and adoption of European rules such as interoperability requirements and standards is essential. The EU must also provide support to ensure that these data sets can be seamless accessed, exchanged and reused," the Commission said.
Meanwhile, a group bringing together experts from academia, industry and civil society is working on developing ethics guidelines for the development and use of AI, with an initial version that will be open for consultation expected to be published shortly. The final document will be released in March next year.
Twitter: @1Leontina
Contact the author: lpostelnicu@himss.org