Alzheimer’s Disease, a form of dementia, could be considered one of the more terrifying illnesses one can get. It starts slowly, with minor memory problems followed by disorientation and confusion (particularly in decision making). Issues with speech and language usually follow. In addition to the more common symptoms, the patient may suffer from personality changes as well as experiencing hallucinations. Patients lose themselves and usually require expert care, which costs time and money.
Alzheimer’s disease affects one in six people over the age of 80 in the UK. One in every 20 cases affects people aged 40 to 65. This is known as early or young-onset Alzheimer’s, causing a very similar disease but with a younger face.
There is no cure for Alzheimer’s and it needs to be caught in the very early stages to be treated, often with drugs that attempt to minimise the symptoms and slow its progression. Unfortunately, if the disease is at the memory loss stage, it is already too late.
Alzheimer’s is caused by a build-up of proteins in the brain, which causes plaques or ‘tangles’. These masses lead to a loss in the connections between nerve cells and the eventual death of the neurones. They form recognisable traces on Magnetic Resonance Imaging (MRI) scans, which are the main method of diagnoses. Currently, MRIs are only conducted once symptoms start to appear and have to be analysed by highly-trained doctors.
There is a desperate need for better diagnostic methods. Dr Mallar Chakravarty of the Douglas Mental Health University Institute and colleagues at the University of Toronto have designed a new beacon of hope. Their new Artificial Intelligence (AI) algorithm is able to predict the cognitive decline that ultimately leads to Alzheimer’s disease. The novel machine-learning model, called Longitudinal Siamese Network (LSN), uses a multitude of data from over 800 people to learn the combined MRI signatures, genetic information and clinical data that predict the development of Alzheimer’s. The AI can predict if an individual is likely to develop symptomatic Alzheimer’s in the next five years, which is far earlier than current methods and could make all the difference.
The technology uses cortical thickness (thickness of certain parts of brain tissue) for diagnosis, which is currently the best-known marker. The studies attempting to find biomarkers are being stumped by the same issues that face treatments – we aren’t catching it early enough. A limited database of patients with an early diagnosis means that there are no long-term studies that properly define each stage of the disease. We don’t properly know how it works because we haven’t been able to watch it develop. This AI could provide a wealth of new research information, as well as slowing people’s disease if they can identify it and get treatment quickly enough.
However, despite AIs being used in almost every area of our lives (including our phones, our social media and our judiciary systems), we become suspicious of utilising them in health care. Maybe it’s because there is more at stake. The implications of a machine getting something wrong in social media is not on the same level as a glitch in healthcare. In one, you get an advert that’s not tailored to your interests popping up on your timeline. In the other, someone is misdiagnosed and can’t get the treatment they need. The difference is stark.
At the same time, as junior doctors are continually punished by their contracts and forced to work long hours making them more susceptible to mistakes. The machine can’t become tired or be less focused because they have a cold or had an argument with a family member at home. The machine can go all day and all night for 365 days a year unless it has a malfunction. It could diagnose 100 patients in a fraction of the time that it would take an exhausted human doctor.
But AI is not being designed to replace doctors. Rather, as Dr Chakravarty said, it should start acting as a “doctor’s assistant”. If it can also increase the success of research into diagnostic biomarkers, there is no reason that the tests couldn’t run in synergy with each other. At-risk groups could become more routinely screened, in the same way as various types of cancer.
AIs are already being used in our healthcare system. Cancer screenings are progressively being introduced to diagnostic techniques. Doctors will often look over the results, but it is still proving to be very useful. Unlike physical diseases, such as cancer, the diseases of the brain are lagging behind in both diagnosis and treatment. In spite of the reduced resources for brain research, Dr Chakravarty’s work, with her colleagues, has provided a breakthrough. Such work will not only benefit long-term patients and their families, but will also speed up the search for a cure for Alzheimer’s disease.
Image Credit: erik_stein via Pixabay