PERTH, Australia – There is pervasive use of artificial intelligence and machine learning (AI/ML) across the health care industry in Australia, and excitement is building on the opportunities it offers to technologies and ultimately to patients, Ausbiotech CEO Lorraine Chiroiu told BioWorld.
"AI/ML is transforming clinical practice in terms of clinical trials, diagnosis, treatment, decision-making, early detection and preventative health," she said.
AI is being used for everything from smart medical records to the systems that help set appointments, to hospital records and diagnostic and pathology tests. It's being used in diagnostics for cancer patients to redirect the best treatment regimens based on a number of patient variables, and patient records can be aggregated so that algorithms can narrow down diagnoses.
AI is changing the precision around surgeries like knee replacements by using robotic surgery to diagnose the exact angles, Brandon Capital Managing Director Chris Nave told BioWorld.
AI/ML doesn't make the diagnosis, but it informs the clinician, providing a hand in the procedure, but that hand is a robot.
"For me, the interesting part is how do you make sure you use AI in a way that improves patient outcomes, but doesn't disenfranchise the clinician or lead to inappropriate diagnoses or inappropriate treatment? I don't think you can ever remove the human hand from patient management," said Nave. "For example, AI is used in analysis in radiography to help scan for fractures or tumors, and then if there's a hit, it goes on to the radiologist who will then do the clinical assessment, and that's where it has a huge benefit.
"Across the globe, clinicians spend more time doing data entry in patients' records than they do treating patients, and I think AI has a real opportunity to change that," Nave said.
Some predictions suggest that AI could replace human jobs by 50%, he said, but that won't be the case in health care.
Increasing capacity in health care
"I think it's going to increase capacity to the health care system, which is creaking at the seams, to actually provide better health care. If you can remove the amount of time that doctors have to do data entry for each of their patients, or if you can make sure that an individual patient with a specific cancer gets the best resume of treatment for their cancer type, age and background, they could get a better outcome.
"I don't think AI is going to be as big a threat as it may be with manufacturing, and I think it has the opportunity to improve the system," Nave said. "But we also think it's not without its challenges. For us to invest, it has to address an immediate clinical problem, it needs to change patient outcomes or lower the cost of healthcare, and it needs to be protectable. That's a challenging concept when you're talking about using macro patient data to try and inform a treatment in an individual patient."
Standards Australia released a discussion paper on developing standards for AI applications, and it asks stakeholders to comment on what they see as the greatest examples, needs and opportunities for adopting AI and how Australians would use it. It asks what Australia's competitive advantages are and where Australia could be a global leader in this space.
The consultation asks stakeholders to what extent standards should play a role in providing a practical solution for implementing AI, as well as the anticipated costs and benefits. If standards are relevant, should they focus on Australian views or be international, the paper asks.
Funding new, innovative startups and rapidly growing companies is critical for Australia's economy to grow and diversify, particularly as the mining boom matures and traditional manufacturing diminishes.
A recent Data61 report suggests that the health care sector in Australia will see the highest five-year growth in employment, growing at roughly 18.7%. A data science research and engineering organization, Data61 is partnering with the Australian government to predict the impact of artificial intelligence and machine learning on Australia across a number of metrics.
"There are emerging industries that show promising growth in VC investment. The health care, life sciences, energy and environment sectors are likely to be key areas of focus among VC investors," according to the report.
ANDHealth (Australia's National Digital Health Initiative), a federally and commercially funded consortium is focused on building a digital health sector in Australia, and it's working with medium-stage digital health companies to bring them to proof-of-concept, clinical validation, investment readiness and market entry.
The digital health ecosystem traditionally has been fragmented, with most of the investments in digital health going to developing a nationwide electronic health record program called "my health record" for the country's universal health care system patients, said ANDHealth Managing Director Bronwyn Le Grice.
Complicating the landscape is the fact that "many innovators come straight out of the tech world where regulation is a new thing," she said.
Bridging the gap between digital tech and health care
When digital technology companies talk to local VCs, they fall into a "no-man's land between traditional health care investors and ICT [information and communication technology] investors," Le Grice said, explaining that ICT investors don't generally play in the regulated medical space, and the health care VCs don't often play in the digital space. One of ANDHealth's goals is to help bridge that gap.
ANDHealth has actively screened 207 companies in the last 18 months, and 9% of those are companies focused on AI/ML, but 33% have data analytics and encompass some aspects of AI/ML, Le Grice said. In fact, most mobile health and telemedicine companies have some level of algorithm technology and machine learning inside them. Altogether about 40% of the companies ANDHealth has screened would fall under this definition.
Pharma companies are by far the biggest users of this type of technology, she said.
Anton Van Den Hengel, director of the Australian Institute of Machine Learning, said that digital technology has already revolutionized commerce, banking, leisure, communications, and it is starting to revolutionize health.
"Health data is important, but broader data can also be powerful and relates factors that are beyond health data. We're currently not using it because everyone is so paranoid about privacy. The question is, who is going to do it first?" he said.
Van Den Hengel said the next great divide will be the data divide. "We need to exploit the data we already have so we can do something good with the data."
Cybersecurity is a real issue, he said, but it's independent of machine learning. And, although machine learning "is watching you, it can't understand what you're doing.
"Humans aren't good at learning from long-term results or measuring long-term trends. Machine learning is good at both," Van Den Hengel said. "The role we have to play is putting them all together and connecting the right sentences. Getting the right data into a centralized device and getting machine learning to solve the problem."
Innovators should band together for different disease areas whereby one app would be available for diabetes, another for respiratory disease, and another for substance abuse so that the constellation of diseases is covered, suggested Anand Iyer, co-founder and chief strategy officer at WellDoc.
He co-created Bluestar – the first app to be cleared by the FDA – to help patients manage their diabetes. He spent 20 years in the wireless industry and saw converging factors in the growing global diabetes epidemic and believed that AI/ML could drive patient engagement to better manage care and reduce costs at the same time.
With these apps comes a host of data that can loop back to the patient to provide feedback, and it can also be used to gain insight on patient preferences and pattern recognition. In this way, data can be extrapolated to predict future events such as if a patient will have a hypoglycemic event or if medication needs to be adjusted.
In Australia's vast remote Outback, apps like Bluestar could help solve geographic challenges.
Researchers harness AI for sequencing, diagnostics
PERTH, Australia – Brisbane's Translational Research Institute (TRI) is working with Siemens Heathineers at Draper Laboratories using magnetic resonance spectroscopy to learn more about the chemical content of tissues and organs, providing a deeper understanding and earlier detection of conditions like post traumatic stress disorder (PTSD), said TRI CEO Carolyn Mountford.
Using AI, clinicians can identify the changes in a patient's brain chemistry associated with PTSD.
For the TRI study, factors include PTSD and blast exposure to develop the algorithm, and then classifications are generated based on data collected from patients. These classifications can be compared to inform clinicians.
The U.S. Department of Defense and Australian military are working with Mountford's team to develop the new magnetic resonance spectroscopy (MRS) in vivo to diagnose changes to brain chemistry associated with brain injury and post-traumatic stress disorder.
By using a scanner at deployment sites, soldiers could be scanned before they are deployed, and then scanned again when they return, she said.
The team is identifying biomarkers that distinguish post-traumatic stress disorder and mild traumatic brain injury using advanced MRS.
MRS also offers the ability to assess women at high risk of breast cancer in whom very early tissue changes can be identified. The technology can identify metabolic deregulations in breast tissue that precede tumor growth, detecting the cellular changes leading to the disease years before the cancer emerges. – Tamra Sami, Staff Writer
AI algorithms give evidence of breast density in cancer screenings
PERTH, Australia – Wellington, New Zealand-based Volpara Health Technologies uses imaging and artificial intelligence to detect breast cancer early. The company offers a suite of breast imaging tools that enable personalized breast cancer screening based on objective measurements of volumetric breast density, compression and radiation dose.
Its Volparadensity program provides an objective volumetric measure of breast density from both digital mammography and tomosynthesis data.
Although mammography is the most effective tool for breast cancer screening, it doesn't work equally well for all women, particularly those with dense breasts. Breast density has not only been linked to an increased risk of breast cancer, it also dramatically impacts early detection. Several large studies have confirmed that as density increases, the accuracy of mammography decreases.
Cloud technology allows individual imaging centers to calculate the amount of dose given to a woman as well as the correct amount of compression. Algorithms also analyze positioning and can spot technicians that might need to adjust their practices.
Radiologists have access to previous mammograms, and by using machine learning they can more easily mine that huge amount of information to spot subtle changes for what is clinically significant. – Tamra Sami, Staff Writer
Using AI to sequence the human genome
PERTH, Australia – An AI application spun out of Australia's QIMR Berghofer Medical Research Institute is a human sequencing project aimed at developing clinical assays for cancer patients and diagnostics labs.
Nic Waddell, head of QIMR Berghofer's Medical Genomics group, co-founded startup Genomiqa, which will offer hospitals, clinicians and pharma companies analysis of data from whole genome sequencing. She said the group can sequence the whole human genome in a few days.
Co-founder John Pearson, who leads the genome informatics group at QIMR, created the software for medical researchers and developed analysis pipelines for whole genome, exome and panel sequencing.
The group is working with big data company Max Kelson to better predict cancer treatment outcomes harnessing AI and genomics. The project is focused on identifying reliable markers to develop tests that predict which patients will benefit most from immunotherapy prior to treatment.
The key to predicting patient treatment outcomes lies in finding and interpreting the patterns and genes of significance in the genomes of patients who have responded best to previous treatments. – Tamra Sami, Staff Writer
AI technology to automate lab processes
PERTH, Australia – Adelaide-based LBT Innovations is launching its AI platform that automates screening and interpreting culture plates for microbiology applications in the lab. The Automated Plate Assessment System (APAS) uses intelligent learning and machine learning technology to read and interpret the presence of bacteria in culture plates.
The company has two modules for analyzing urine and methicillin-resistant Staphylococcus aureus (MRSA), which represent 60% to 75% of the specimens that go through the manual reading system, LBT CEO Brent Barnes told BioWorld.
"The gold standard is still a manual plate reading, so you still need a microbiologist to do some antibiotic sensitivity reading for those positive readings to get a drug recommendation," Barnes said, "but the automated step helps remove all the negatives and allows these highly skilled microbiologists to focus on the positives."
The negatives generally represent more than 90% of the samples, he said.
"A scientist can read maybe 40 to 60 plates per hour, and our technology can read 200 plates per hour."
The urine module is the first and only FDA-cleared automated reading and interpretation system. For that clearance, LBT Innovations conducted a 10,000-patient clinical trial whereby a manual loading system was built to compare head-to-head analyses of three independent microbiologists comparing the same plates manually.
Those trial results were submitted to the FDA for a de novo designation, and the company received clearance in October 2016 to market the device as a Class II device in the U.S.
A 510(k) supplement will be needed for the MRSA indication to show the FDA that the automated instrument is equivalent to the manual loading instrument.
The next application is an intelligent imaging software solution to track the progress of long-term chronic wounds. The hand-held device will take a picture of the wound and then trained algorithms will analyze it to provide an objective analysis. – Tamra Sami, Staff Writer