02-08-2021 | | By Robin Mitchell
Recently, researchers have been able to combine GPS with AI to detect early-onset Alzheimer’s in drivers which a high degree of accuracy. Why is detecting Alzheimer’s early important, what did the researchers achieve, and how does it demonstrate the importance of AI in medical diagnosis?
There are very few individuals who enjoy arranging doctor appointments, having blood taken, and waiting for results. Despite the unpleasantness of such experiences, getting diagnosed as early as possible for diseases provides the best chance for treatment and survival. For example, many millions around the world still die as a result of perfectly treatable conditions such as prostate and breast cancer, and this is due to a lack of awareness and desire to get tested.
Alzheimer’s is a particularly nasty disease for a multitude of reasons. Firstly, Alzheimer’s is a degenerative disease that attacks the brain meaning that suffers begin to lose the ability to take care of themselves, perform common daily tasks, and even recognize loved ones. Secondly, the effect of Alzheimer’s takes its toll on those who care for the suffering; anyone who has spoken to suffers from degenerative brain diseases knows the pain of not being recognized. Thirdly, there is no known cure that can completely eliminate the disease meaning that a diagnosis is essentially a death sentence.
However, researchers around the world continue to investigate the disease looking for methods for either stopping it entirely or slowing it down. While those in the later stages of Alzheimer’s have extremely limited options, those who have been diagnosed with the disease in its early stages can take advantage of clinical trials, experimental drugs, and lifestyle changes that may help to increase life expectancy. But getting an early diagnosis can be invasive and difficult with some tests including blood tests, DNA sequencing, and PET scans of the brain. Unless someone starts showing symptoms, diagnosing Alzheimer’s in those who show no family history of the condition is difficult and unlikely.
Recently, researchers combined multiple forms of technology to create a system that is able to detect early-onset Alzheimer’s with a high degree of accuracy. Simply put, the researchers proposed that those suffering from Alzheimer’s would make different decisions when driving compared to those who do not have the disease. For example, Alzheimer drivers may drive for shorter periods of time, stick to commonly used routes, travel less at night, and make abrupt changes to their driving.
To record such data, the researchers selected 139 people in the US to have a GPS unit installed in their vehicles. Once enough data was gathered, the researchers then fed the data into a custom AI system designed using Python to look for patterns in driving. Half of the selected individuals had already been positively tested for Alzheimer’s while the others had been tested to be negative for the disease. Testing of Alzheimer’s in the positive half of the group was done using medical methods including spinal fluid tests and PET scans.
The results of the experiment showed a detection accuracy of 82% which improved to over 90% when combined with genetic marker tests that indicate risk factors (a positive genetic market test doesn’t mean Alzheimer’s will occur, only that it’s a risk factor). However, it should be noted that users of the GPS systems were over the age of 65 meaning that such a system may not be able to detect Alzheimer’s in the young where symptoms are not present.
The use of AI to diagnose Alzheimer’s early demonstrates the importance of AI in medical diagnosis and how the medical industry should be turning to its power. Doctors are human, and humans can only absorb so much information. While a human doctor has the advantage of being able to read and understand human emotion (in reality, doctors rarely understand emotion nor care for it), an AI is able to read huge amounts of cases and learn from millions of diagnoses. AI also carries no bias or prejudice meaning that patients are more likely to be correctly diagnosed, and the ability for AI to continually improve itself creates a diagnoses machine that only ever gets better.
However, the medical industry for whatever reason seems to be resistant to using such tools when diagnosing patients. It is unlikely for doctors to be concerned with losing jobs to AI as doctors are required to authorize treatments and prescriptions. Therefore, the resistance to AI and computer systems in the medical industry may stem from arrogance and the self-importance generally exhibited by doctors. The idea that an AI can diagnose patients better than a doctor may be seen as impossible by medical professionals as AI is generally unable to determine patients that may be lying about their symptoms.
It is clear that AI will eventually exceed human capability in medical diagnosis, and humans can use AI tools to help better provide treatment. However, AI will also be important in early detection, and such data could be extracted from everyday items including smartphones, IoT devices, and vehicles.