AI-Enabled Organ Software
Human brilliance in science and bio-medical engineering has enabled many great discoveries including the early Australian-led discovery of the Cochlear implant (bionic ear). More recently human genius has also underpinned the Bivacor TAH heart device, a Queensland invention with the capacity to replace the total function of the human heart. During his childhood, Daniel Timms took in the mechanics of plumbing practiced by his father and he drew on that knowledge to equip the Bivacor artificial heart with its rotary blood-pumping technology. That kind of genius will be retained if we carefully marry human and artificial intelligence.
From pre-natal predictions of an ‘at-risk’ infant’s chances of having a developmental delay (made possible by wearable sensors that monitor a baby’s movements) to the implantation of artificial organ software, AI will revolutionise healthcare. Early discoveries in bionics made little use of AI, but most future innovations will use it in some way. AI will bring new innovations to enhance our senses (hearing, vision, touch and taste) and it has the potential to take mind-controlled bionic limbs and prosthetics to the next level.
Australia’s most recent moon-shot discovery in human bionics is AI-enabled artificial pancreas software. Brisbane-based Dr Nigel Greenwood together with Professor Jenny Gunton and their Diabetes Neuromathix (DNx) team have successfully created this AI-enabled organ software to markedly improve the quality of life and longevity of ‘brittle’ Type 1 diabetics.
Dr Nigel Greenwood
As Dr Greenwood explains, “Self-administration of insulin is a difficult thing to do for around 75% of people living with diabetes, most especially those who experience a rollercoaster of changes in their blood glucose levels. If they get their insulin dosing wrong, they risk blindness, kidney failure and other organ damage. As a result, there has been a vigorous global race to invent artificial pancreas technology that delivers an accurate insulin dosage.
Dr Greenwood’s Diabetes Neuromathix (DNx) pancreas (http://neuromathix.com/) is the first in the world to deliver a personalised, sophisticated computational model of diabetes in the body. The model is interpreted by two competing AI systems. In simple terms, the second AI system uses ‘known unknowns’ about the person’s diabetic history to try to wreck the first AI system’s treatment strategy. The first system fights back and data generated from this ‘AI battle or game’ provide the insights needed to internally develop and deliver a customised insulin strategy for that individual.
Neuromathix AI enabled artificial pancreas software
This ‘adversarial AI’ embedded in the Neuromathix pancreas is a totally new form of AI that interprets a new form of machine learning. It is based on ecosystem dynamics instead of the human brain. Operating partly in the Cloud and partly in an App, the pancreas software delivers customised treatment strategies for Type 1 diabetics more quickly and efficiently than AI based on conventional machine learning or fuzzy logic.
The software relies on high-performance computers (built on Nvidia Tesla GPU architecture) and algorithms sourced from advanced robotics applications and modified for medical contexts. These algorithms are protected by patents in Australia and other key markets worldwide.
Neuromathix Team MachineGenes is in the top 10 teams among those competing worldwide for the prestigious IBM Watson AI X-Prize (http://machinegenes.com/xprize.html) to be announced late 2019.
